Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 35
Encontrada(s) 1 página(s)

Biblos-e Archivo. Repositorio Institucional de la UAM
Publication Artículo científico (JournalArticle)

KRAS p.G12C mutation occurs in 1% of EGFR-mutated advanced non-small-cell lung cancer patients progressing on a first-line treatment with a tyrosine kinase inhibitor

  • Serna-Blasco, R.
  • Sánchez-Herrero, E.
  • Sanz-Moreno, S.
  • Rodriguez-Festa, A.
  • García-Veros, E.
  • Casarrubios, M.
  • Sierra-Rodero, B.
  • Laza-Briviesca, R.
  • Cruz-Bermúdez, A.
  • Mielgo-Rubio, X.
  • Sánchez-Hernández, A.
  • Uribelarrea, E. A.
  • Calvo de Juan, Virginia
  • Romero, A.
  • Provencio Pulla, Mariano
Background: KRAS is mutated in ∼30% of non-small-cell lung cancer (NSCLC) but it has also been identified as one of the mechanisms underlying resistance to tyrosine kinase inhibitors (TKIs) in EGFR-positive NSCLC patients. Novel KRAS inhibitors targeting KRAS p.G12C mutation have been developed recently with promising results. The proportion of EGFR-positive NSCLC tumours harbouring the KRAS p.G12C mutation upon disease progression is completely unexplored. Materials and methods: Plasma samples from 512 EGFR-positive advanced NSCLC patients progressing on a first first-line treatment with a TKI were collected. The presence of KRAS p.G12C mutation was assessed by digital PCR. Results: Overall, KRAS p.G12C mutation was detected in 1.17% of the samples (n = 6). In two of these cases, we could confirm that the KRAS p.G12C mutation was not present in the pre-treatment plasma samples, supporting its role as an acquired resistance mutation. According to our data, KRASG12C patients showed similar clinicopathological characteristics to those of the rest of the study cohort and no statistically significant associations between any clinical features and the presence of the mutation were found. However, two out of six KRASG12C tumours harboured less common EGFR driver mutations (p.G719X/p.L861Q). All KRASG12C patients tested negative for the presence of p.T790M resistance mutation. Conclusions: The KRAS p.G12C mutation is detected in 1% of EGFR-positive NSCLC patients who progress on a first line with a TKI. All KRASG12C patients were negative for the presence of the p.T790M mutation and they did not show any distinctive clinical feature, This work was supported by CLARIFY project (https://www.clarify2020.eu/). ES-H was supported by the Consejería de Ciencia, Universidades e Innovación of the Comunidad de Madrid (Doctorados Industriales of the Comunidad de Madrid) [grant number IND2019/BMD-17258]. EG-V was supported by AECC (Asociación española Contra el Cáncer) grant ‘Programa de Prácticas de Laboratorio de curso académico AECC 2020’ (no grant number).
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Estudio, informe, memoria (Report)

Cancer Survivorship and AI for Well-being White Paper: A European research projects collaborative perspective

  • Callejas Carrión, Zoraida|||0000-0001-8891-5237
  • Flynn, Tom
The Cancer Survivorship - AI for Well-being cluster (#CS_AIW) has brought together 11 EU-funded projects working in artificial intelligence (AI) for healthcare and well-being with the aim of transcending the individual project experiences and stop working in silos. The White Paper presents the #CS_AIW initiative and our collective shared vision, providing an overview of the participating projects and it reports on the achievements and lessons learned from our collaboration, including both the benefits obtained and the difficulties addressed. The experience with #CS_AIW to date has proven that collaboration within the cluster has been extremely beneficial and useful to foster an ethos of cross-fertilization engagement, share best practices, boost cross-sector collaboration, increase the participation of end user communities, and reach a wider audience. The activities organized, which encompass meetings, roundtable discussions, podcasts, a better practices guide, a common data space and the white paper itself, have proven to be an inherited valued initiative to generate, activate and communicate knowledge and innovation in the areas of mental health, well-being, cancer recovery, patient support and participatory research., The projects in #CS_AIW have received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no.: ASCAPE - Ref 875351 BD4QoL - Ref 875192 CAPABLE - Ref 875052 CLARIFY - Ref 860627 FAITH - Ref 875358 LIFECHAMPS - Ref 875329 MENHIR - Ref 823907 ONCORELIEF - Ref 875392 PERSIST - Ref 875406 QUALITOP - Ref 875171 REBECCA - Ref 65231
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Contribución de congreso (ConferenceOutput)

Probabilistic Modeling of Inter- and Intra-observer Variability in Medical Image Segmentation

  • Schmidt, Arne
  • Morales Álvarez, Pablo
  • Molina Soriano, Rafael
Medical image segmentation is a challenging task, particularly due to inter- and intra-observer variability, even between medical experts. In this paper, we propose a novel model, called Probabilistic Inter-Observer and iNtra- Observer variation NetwOrk (Pionono). It captures the labeling behavior of each rater with a multidimensional probability distribution and integrates this information with the feature maps of the image to produce probabilistic segmentation predictions. The model is optimized by variational inference and can be trained end-to-end. It outperforms state-of-the-art models such as STAPLE, Probabilistic UNet, and models based on confusion matrices. Additionally, Pionono predicts multiple coherent segmentation maps that mimic the rater’s expert opinion, which provides additional valuable information for the diagnostic process. Experiments on real-world cancer segmentation datasets demonstrate the high accuracy and efficiency of Pionono, making it a powerful tool for medical image analysis., European Union’s H2020 research and innovation programme (Marie Skłodowska Curie grant agreement No 860627, CLARIFY Project), Spanish Ministry of Science and Innovation (project PID2019-105142RB-C22), FEDER/Junta de Andalucía-Consejería de Transformación Económica, Industria, Conocimiento y Universidades (project P20 00286)
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Artículo científico (JournalArticle)

Introducing instance label correlation in multiple instance learning. Application to cancer detection on histopathological images

  • Morales Álvarez, Pablo
  • Schmidt, Arne
  • Hernández-Lobato, José Miguel
  • Molina Soriano, Rafael
In the last years, the weakly supervised paradigm of multiple instance learning (MIL) has become very popular in many different areas. A paradigmatic example is computational pathology, where the lack of patch-level labels for whole-slide images prevents the application of supervised models. Probabilistic MIL methods based on Gaussian Processes (GPs) have obtained promising results due to their excellent uncertainty estimation capabilities. However, these are general-purpose MIL methods that do not take into account one important fact: in (histopathological) images, the labels of neighboring patches are expected to be correlated. In this work, we extend a state-of-the-art GP-based MIL method, which is called VGPMIL-PR, to exploit such correlation. To do so, we develop a novel coupling term inspired by the statistical physics Ising model. We use variational inference to estimate all the model parameters. Interestingly, the VGPMIL-PR formulation is recovered when the weight that regulates the strength of the Ising term vanishes. The performance of the proposed method is assessed in two real-world problems of prostate cancer detection. We show that our model achieves better results than other state-of-the-art probabilistic MIL methods. We also provide different visualizations and analysis to gain insights into the influence of the novel Ising term. These insights are expected to facilitate the application of the proposed model to other research areas., European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska Curie grant agreement No 860627 (CLARIFY Project), Spanish Ministry of Science and Innovation under project PID2019-105142RBC22, University of Granada and FEDER/Junta de Andalucía under project B-TIC-324-UGR20 (Proyectos de I+D+i en el marco del Programa Operativo FEDER Andalucía), Margarita Salas postdoctoral fellowship (Spanish Ministry of Universities with Next Generation EU funds)
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Contribución de congreso (ConferenceOutput)

Combining Attention-based Multiple Instance Learning and Gaussian Processes for CT Hemorrhage Detection

  • Wu, Yunan
  • Schmidt, Arne
  • Hernández Sánchez, Enrique
  • Molina Soriano, Rafael
This work has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska Curie grant agreement No 860627 (CLARIFY Project) and also from the Spanish Ministry of Science and Innovation under project PID2019-105142RB-C22., Intracranial hemorrhage (ICH) is a life-threatening emer- gency with high rates of mortality and morbidity. Rapid and accurate de- tection of ICH is crucial for patients to get a timely treatment. In order to achieve the automatic diagnosis of ICH, most deep learning models rely on huge amounts of slice labels for training. Unfortunately, the manual annotation of CT slices by radiologists is time-consuming and costly. To diagnose ICH, in this work, we propose to use an attention-based multiple instance learning (Att-MIL) approach implemented through the combi- nation of an attention-based convolutional neural network (Att-CNN) and a variational Gaussian process for multiple instance learning (VGP- MIL). Only labels at scan-level are necessary for training. Our method (a) trains the model using scan labels and assigns each slice with an at- tention weight, which can be used to provide slice-level predictions, and (b) uses the VGPMIL model based on low-dimensional features extracted by the Att-CNN to obtain improved predictions both at slice and scan levels. To analyze the performance of the proposed approach, our model has been trained on 1150 scans from an RSNA dataset and evaluated on 490 scans from an external CQ500 dataset. Our method outperforms other methods using the same scan-level training and is able to achieve comparable or even better results than other methods relying on slice- level annotations., European Commission 860627, Spanish Government PID2019-105142RB-C22
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Artículo científico (JournalArticle)

The Devil is in the Details: Whole Slide Image Acquisition and Processing for Artifacts Detection, Color Variation, and Data Augmentation: A Review

  • Kanwal, Neel
  • Pérez Bueno, Fernando
  • Schmidt, Arne
  • Engan, Kjersti
  • Molina Soriano, Rafael
Whole Slide Images (WSI) are widely used in histopathology for research and the diagnosis of different types of cancer. The preparation and digitization of histological tissues leads to the introduction of artifacts and variations that need to be addressed before the tissues are analyzed. WSI preprocessing can significantly improve the performance of computational pathology systems and is often used to facilitate human or machine analysis. Color preprocessing techniques are frequently mentioned in the literature, while other areas are usually ignored. In this paper, we present a detailed study of the state-of-the-art in three different areas of WSI preprocessing: Artifacts detection, color variation, and the emerging field of pathology-specific data augmentation. We include a summary of evaluation techniques along with a discussion of possible limitations and future research directions for new methods., European Commission 860627, Ministerio de Ciencia e Innovacion (MCIN)/Agencia Estatal de Investigacion (AEI) PID2019-105142RB-C22, Fondo Europeo de Desarrollo Regional (FEDER)/Junta de Andalucia-Consejeria de Transformacion Economica, Industria, Conocimiento y Universidades B-TIC-324-UGR20, Instituto de Salud Carlos III Spanish Government European Commission BES-2017-081584
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Artículo científico (JournalArticle)

Efficient Cancer Classification by Coupling Semi Supervised and Multiple Instance Learning

  • Schmidt, Arne
  • Molina Soriano, Rafael
The annotation of large datasets is often the bottleneck in the successful application of artificial intelligence in computational pathology. For this reason recently Multiple Instance Learning (MIL) and Semi Supervised Learning (SSL) approaches are gaining popularity because they require fewer annotations. In this work we couple SSL and MIL to train a deep learning classifier that combines the advantages of both methods and overcomes their limitations. Our method is able to learn from the global WSI diagnosis and a combination of labeled and unlabeled patches. Furthermore, we propose and evaluate an efficient labeling paradigm that guarantees a strong classification performance when combined with our learning framework. We compare our method to SSL and MIL baselines, the state-of-the-art and completely supervised training. With only a small percentage of patch labels our proposed model achieves a competitive performance on SICAPv2 (Cohen's kappa of 0.801 with 450 patch labels), PANDA (Cohen's kappa of 0.794 with 22,023 patch labels) and Camelyon16 (ROC AUC of 0.913 with 433 patch labels). Our code is publicly available at https://github.com/arneschmidt/ssl_and_mil_cancer_classification., European Union's Horizon 2020 Research and Innovation Program through the Marie Skodowska Curie (Cloud Artificial Intelligence For pathologY (CLARIFY) Project) 860627, Spanish Government PID2019-105142RB-C22
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Artículo científico (JournalArticle)

Proportion constrained weakly supervised histopathology image classification

  • Silva Rodríguez, Julio
  • Schmidt, Arne
  • Molina Soriano, Rafael
Multiple instance learning (MIL) deals with data grouped into bags of instances, of which only the global information is known. In recent years, this weakly supervised learning paradigm has become very popular in histological image analysis because it alleviates the burden of labeling all cancerous regions of large Whole Slide Images (WSIs) in detail. However, these methods require large datasets to perform properly, and many approaches only focus on simple binary classification. This often does not match the real-world problems where multi-label settings are frequent and possible constraints must be taken into account. In this work, we propose a novel multi-label MIL formulation based on inequality constraints that is able to incorporate prior knowledge about instance proportions. Our method has a theoretical foundation in optimization with logbarrier extensions, applied to bag-level class proportions. This encourages the model to respect the proportion ordering during training. Extensive experiments on a new public dataset of prostate cancer WSIs analysis, SICAP-MIL, demonstrate that using the prior proportion information we can achieve instance-level results similar to supervised methods on datasets of similar size. In comparison with prior MIL settings, our method allows for ∼ 13% improvements in instance-level accuracy, and ∼ 3% in the multi-label mean area under the ROC curve at the bag-level., Spanish Government PID2019-105142RB-C2, European Commission 860627, Generalitat Valenciana/European Union through the European Regional Development Fund (ERDF) of the Valencian Community IDIFEDER/2020/030, Universitat Politecnica de Valencia
Digibug. Repositorio Institucional de la Universidad de Granada
Publication Artículo científico (JournalArticle)

Bibliometric analysis of the global scientific production on machine learning applied to different cancer types

  • Ruiz Fresneda, Miguel Ángel|||0000-0001-6349-6566
  • Gijón Gijón, Alfonso
  • Morales Álvarez, Pablo
This work has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska Curie grant agreement no. 860627 (CLARIFY Project), from the Spanish Ministry of Science and Innovation under project PID2019-105142RB-C22, and by FEDER/Junta de Andalucia-Consejeria de Transformacion Economica, Industria, Conocimiento y Universidades under the project P20_00286. Funding for open access charge: Universidad de Granada/CBUA., Cancer disease is one of the main causes of death in the world, with million annual cases in the last decades. The need to find a cure has stimulated the search for efficient treatments and diagnostic procedures. One of the most promising tools that has emerged against cancer in recent years is machine learning (ML), which has raised a huge number of scientific papers published in a relatively short period of time. The present study analyzes global scientific production on ML applied to the most relevant cancer types through various bibliometric indicators. We find that over 30,000 studies have been published so far and observe that cancers with the highest number of published studies using ML (breast, lung, and colon cancer) are those with the highest incidence, being the USA and China the main scientific producers on the subject. Interestingly, the role of China and Japan in stomach cancer is correlated with the number of cases of this cancer type in Asia (78% of the worldwide cases). Knowing the countries and institutions that most study each area can be of great help for improving international collaborations between research groups and countries. Our analysis shows that medical and computer science journals lead the number of publications on the subject and could be useful for researchers in the field. Finally, keyword co-occurrence analysis suggests that ML-cancer research trends are focused not only on the use of ML as an effective diagnostic method, but also for the improvement of radiotherapy- and chemotherapy-based treatments., Horizon 2020 European Union under the Marie Sklodowska Curie 860627, Spanish Government PID2019-105142RB-C22, FEDER/Junta de Andalucía-Consejería de Transformación Económica, Industria, Conocimiento y Universidades P20_00286, Universidad de Granada/CBUA
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Histological interpretation of spitzoid tumours: an extensive machine learning-based concordance analysis for improving decision making

  • Mosquera-Zamudio, Andrés
  • Launet, Laetitia Mariana
  • Wiedemeyer, Katharina
  • López-Takegami, Juan C.
  • Palma, Luis F.
  • Undersrud, Erling
  • Janssen, Emilius
  • Brenn, Thomas
  • Monteagudo, Carlos
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] The histopathological classification of melanocytic tumours with spitzoid features remains a challenging task. We confront the complexities involved in the histological classification of these tumours by proposing machine learning (ML) algorithms that objectively categorise the most relevant features in order of importance. The data set comprises 122 tumours (39 benign, 44 atypical and 39 malignant) from four different countries. BRAF and NRAS mutation status was evaluated in 51. Analysis of variance score was performed to rank 22 clinicopathological variables. The Gaussian naive Bayes algorithm achieved in distinguishing Spitz naevus from malignant spitzoid tumours with an accuracy of 0.95 and kappa score of 0.87, utilising the 12 most important variables. For benign versus non-benign Spitz tumours, the test reached a kappa score of 0.88 using the 13 highest-scored features. Furthermore, for the atypical Spitz tumours (AST) versus Spitz melanoma comparison, the logistic regression algorithm achieved a kappa value of 0.66 and an accuracy rate of 0.85. When the three categories were compared most AST were classified as melanoma, because of the similarities on histological features between the two groups. Our results show promise in supporting the histological classification of these tumours in clinical practice, and provide valuable insight into the use of ML to improve the accuracy and objectivity of this process while minimising interobserver variability. These proposed algorithms represent a potential solution to the lack of a clear threshold for the Spitz/spitzoid tumour classification, and its high accuracy supports its usefulness as a helpful tool to improve diagnostic decision-making.; Our machine learning models improve histopathological diagnostic decision-making, ranking the importance of the histological variables for the classification of Spitz and spitzoid tumours. image, This work has received funding from Horizon 2020, the European Commission's Framework Programme for Research and Innovation, under the grant agreement no. 860627 (CLARIFY), PI20/00094, Instituto de Salud Carlos III, FEDER European Funds and INNEST/2021/321 (SAMUEL). The work of A.C. has been supported by the ValgrAI-Valencian Graduate School and Research Network for Artificial Intelligence and Generalitat Valenciana and Universitat Politecnica de Valencia (PAID-PD-22).The material from Norway has been collected as part of the project 'Pathology services in the Western Norway Health Region-a centre for applied digitisation' which is financed through a strategic investment from the Western Norway Health Authority.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

A Deep Embedded Framework for Spitzoid Neoplasm Classification Using DNA Methylation Data

  • del Amor, Rocío
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Monteagudo, Carlos
  • Garzón, María José
  • García-Giménez, José Luis
[EN] Spitzoid melanocytic tumors (SMT) are a group of neoplasms that represent a formidable diagnostic challenge for dermatopathologists. DNA methylation (DNAm) is a welldefined epigenetic factor that has an important role in the development of these lesions. In this work, we propose different deep-learning-based approaches to address the Spitzoid neoplasms detection from DNAm. We use an autoencoder and a variational autoencoder for dimensionality reduction with a subsequently supervised classification. Additionally, we present a deep embedded refined clustering algorithm able to optimize the latent space at the same time that the non-supervised classification task is performed. This novel approach in DNAm supposes a step forward in the SMT detection as suggest the obtained results (acc = 0.9). Additionally, making use of the resulting model, we present a subspace-prototypical-based approach for the prognostic prediction of uncertain malignant potential samples, which is nowadays the hottest open area in SMT detection., This work has received funding from Horizon 2020, the European Union¿s Framework Programme for Research and Innovation, under grant agreement No. 860627 (CLARIFY), the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN) and SICAP (DPI2016-77869-C2-1-R) and GVA through project PROMETEO/2019/109
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Uninformed Teacher-Student for hard-samples distillation in weakly supervised mitosis localization

  • Fernández-Martín, Claudio
  • Silva-Rodríguez, Julio
  • Kiraz, Umay
  • Janssen, Emiel A.M.
  • Morales, Sandra|||0000-0003-0763-1545
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Background and Objective: Mitotic activity is a crucial biomarker for diagnosing and predicting outcomes for different types of cancers, particularly breast cancer. However, manual mitosis counting is challenging and time-consuming for pathologists, with moderate reproducibility due to biopsy slide size, low mitotic cell density, and pattern heterogeneity. In recent years, deep learning methods based on convolutional neural networks (CNNs) have been proposed to address these limitations. Nonetheless, these methods have been hampered by the available data labels, which usually consist only of the centroids of mitosis, and by the incoming noise from annotated hard negatives. As a result, complex algorithms with multiple stages are often required to refine the labels at the pixel level and reduce the number of false positives. Methods: This article presents a novel weakly supervised approach for mitosis detection that utilizes only image-level labels on histological hematoxylin and eosin (H&E) images, avoiding the need for complex labeling scenarios. Also, an Uninformed Teacher-Student (UTS) pipeline is introduced to detect and distill hard samples by comparing weakly supervised localizations and the annotated centroids, using strong augmentations to enhance uncertainty. Additionally, an automatic proliferation score is proposed that mimicks the pathologist-annotated mitotic activity index (MAI). The proposed approach is evaluated on three publicly available datasets for mitosis detection on breast histology samples, and two datasets for mitotic activity counting in whole-slide images. Results: The proposed framework achieves competitive performance with relevant prior literature in all the datasets used for evaluation without explicitly using the mitosis location information during training. This approach challenges previous methods that rely on strong mitosis location information and multiple stages to refine false positives. Furthermore, the proposed pipeline for hard-sample distillation demonstrates promising dataset-specific improvements. Concretely, when the annotation has not been thoroughly refined by multiple pathologists, the UTS model offers improvements of up to in mitosis localization, thanks to the detection and distillation of uncertain cases. Concerning the mitosis counting task, the proposed automatic proliferation score shows a moderate positive correlation with the MAI annotated by pathologists at the biopsy level on two external datasets. Conclusions: The proposed Uninformed Teacher-Student pipeline leverages strong augmentations to distill uncertain samples and measure dissimilarities between predicted and annotated mitosis. Results demonstrate the feasibility of the weakly supervised approach and highlight its potential as an objective evaluation tool for tumor proliferation., This work was funded by the Horizon 2020 European Union research and innovation programme under the Marie Sklodowska Curie grant agreement No 860627 (CLARIFY Project) . The work of Sandra Morales has been co-funded by the Universitat Politecnica de Valencia, Spain through the program PAID-10-20. The work of J. Silva-Rodriguez was carried out during his previous position at Universitat Politecnica de Valencia. This work was partially funded by Generalitat Valenciana through project CIPROM/2022/20 and with Ayuda a Primeros Proyectos de Investigacion (PAID-06-23) , Vicerrectorado de Investigacion of the Universitat Politecnica de Valencia. Funding for open access charge: Universitat Politecnica de Valencia (PAID-12-23).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Annotation protocol and crowdsourcing multiple instance learning classification of skin histological images: The CR-AI4SkIN dataset

  • del Amor, Rocío
  • Morales, Sandra|||0000-0003-0763-1545
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Pérez-Cano, Jose
  • López-Pérez, Miguel
  • Terradez, Liria
  • Aneiros-Fernandez, Jose
  • Mateos, Javier
  • Molina, Rafael
[EN] Digital Pathology (DP) has experienced a significant growth in recent years and has become an essential tool for diagnosing and prognosis of tumors. The availability of Whole Slide Images (WSIs) and the implementation of Deep Learning (DL) algorithms have paved the way for the appearance of Artificial Intelligence (AI) systems that support the diagnosis process. These systems require extensive and varied data for their training to be successful. However, creating labeled datasets in histopathology is laborious and time-consuming. We have developed a crowdsourcing-multiple instance labeling/learning protocol that is applied to the creation and use of the CR-AI4SkIN dataset.2 CR-AI4SkIN contains 271 WSIs of 7 Cutaneous Spindle Cell (CSC) neoplasms with expert and non-expert labels at region and WSI levels. It is the first dataset of these types of neoplasms made available. The regions selected by the experts are used to learn an automatic extractor of Regions of Interest (ROIs) from WSIs. To produce the embedding of each WSI, the representations of patches within the ROIs are obtained using a contrastive learning method, and then combined. Finally, they are fed to a Gaussian process-based crowdsourcing classifier, which utilizes the noisy non-expert WSI labels. We validate our crowdsourcing-multiple instance learning method in the CR-AI4SkIN dataset, addressing a binary classification problem (malign vs. benign). The proposed method obtains an F1 score of 0.7911 on the test set, outperforming three widely used aggregation methods for crowdsourcing tasks. Furthermore, our crowdsourcing method also outperforms the supervised model with expert labels on the test set (F1-score = 0.6035). The promising results support the proposed crowdsourcing multiple instance learning annotation protocol. It also validates the automatic extraction of interest regions and the use of contrastive embedding and Gaussian process classification to perform crowdsourcing classification tasks., This work has received funding from the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN) and Spanish Ministry of Science and Innovation through project PID2022-140189OB-C22, from Horizon 2020, the European Union's Framework Programme for Research and Innovation, under the grant agreement No. 860627 (CLARIFY) , grant B-TIC-324-UGR20 funded by Consejeria de Universidad, Investigacion e Innovacion (Junta de Andalucia) and by "ERDF A way of making Europe", and GVA through the project INNEST/2021/321 (SAMUEL) . The work of Rocio del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263) . The work of Miguel Lopez Perez has been supported by the University of Granada postdoctoral program "Contrato Puente". The work of Sandra Morales has been co-funded by the Universitat Politecnica de Valencia through the program PAID-10-20.r Union's Framework Programme for Research and Innovation, under the grant agreement No. 860627 (CLARIFY) , grant B-TIC-324-UGR20 funded by Consejeria de Universidad, Investigacion e Innovacion (Junta de Andalucia) and by "ERDF A way of making Europe", and GVA through the project INNEST/2021/321 (SAMUEL) . The work of Rocio del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263) . The work of Miguel Lopez Perez has been supported by the University of Granada postdoctoral program "Contrato Puente". The work of Sandra Morales has been co-funded by the Universitat Politecnica de Valencia through the program PAID-10-20.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Beyond nest size: the clinicopathological spectrum of large nested melanocytic tumours and the value of comparative genomic hybridisation and messenger RNA expression analysis

  • Mosquera-Zamudio, Andrés
  • Pérez-Debén, Silvia
  • Porcar-Saura, Saray
  • Casabó-Vallés, German
  • Martínez-Rodríguez, Miguel
  • Garzón, María José
  • García-López, Eva
  • Monteagudo, Carlos
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Large nested melanomas (LNMs) are a rare subtype of naevoid melanoma consisting of large junctional melanocytic nests that are more common in older individuals and/or associated with sun damage. However, the presence of large melanocytic nests alone does not lead to a diagnosis of malignancy, as they can also be found in melanocytic naevi. LNMs are challenging because they lack most classic histological features of malignancy and require thorough clinicopathological evaluation. This ambiguity calls for a critical reassessment of the current diagnostic criteria for the subclassification of benign or malignant within the spectrum of large nested melanocytic tumours (LNMTs). Eighteen LNMTs and six special-site melanocytic naevi (SSMNs) were studied using different approaches: clinical features, dermoscopy, histopathology, immunohistochemistry, array comparative genomic hybridisation (aCGH) and messenger RNA (mRNA) sequencing analysis, with the aim of identifying novel and reproducible criteria for the recognition of LNMs.Careful clinicopathological evaluation of the 18 LNMTs led to the diagnosis of seven LNMs and 11 large nested melanocytic naevi (LNMNs). Lentiginous spread and nest bridging were significantly associated with LNMs after Holm-Bonferroni correction. Asymmetry, largest nest size, poor lateral demarcation, the number of colours and dermoscopic structures, and preferentially expressed antigen in melanoma (PRAME) immunostaining were more common in LNMs but did not reach statistical significance. Four of seven LNMs and nine of 11 LNMNs lacked the BRAF V600E mutation. Regarding aCGH, no LNMN or SSMN cases had >= 3 copy number variations (CNVs), in contrast to 50% of LNM cases. Importantly, LNM and LNMN could be distinguished by differential mRNA expression of nine genes. Our study demonstrates that there is a spectrum of LNMTs and that the clinicopathological diagnosis of LNM, for which we support the term 'late-onset nested naevoid melanomas', can be significantly strengthened by the presence of lentiginous pattern, nest bridging, gene CNV and differential mRNA expression., CM, VN and AM-Z are supported by Horizon 2020 Marie Sklodowska Curie Action 860627 CLARIFY grant from the European Commission; CM, VN, EG-L and SP-D are supported by INNEST/2021/302/58 grant from Agencia Valenciana de la Innovacion-AVI Generalitat Valenciana. CM is also supported by PI20/00094 grant from Instituto de Salud Carlos III, Spain; and FEDER European funds. The authors state that there are no conflicts of interest to disclose.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Proportion constrained weakly supervised histopathology image classification

  • Silva-Rodríguez, Julio
  • Schmidt, Arne
  • Sales, María A.
  • Molina, Rafael
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Multiple instance learning (MIL) deals with data grouped into bags of instances, of which only the global information is known. In recent years, this weakly supervised learning paradigm has become very popular in histological image analysis because it alleviates the burden of labeling all cancerous regions of large Whole Slide Images (WSIs) in detail. However, these methods require large datasets to perform properly, and many approaches only focus on simple binary classification. This often does not match the real-world problems where multi-label settings are frequent and possible constraints must be taken into account. In this work, we propose a novel multi-label MIL formulation based on inequality constraints that is able to incorporate prior knowledge about instance proportions. Our method has a theoretical foundation in optimization with log-barrier extensions, applied to bag-level class proportions. This encourages the model to respect the proportion ordering during training. Extensive experiments on a new public dataset of prostate cancer WSIs analysis, SICAP-MIL, demonstrate that using the prior proportion information we can achieve instance-level results similar to supervised methods on datasets of similar size. In comparison with prior MIL settings, our method allows for similar to 13% improvements in instance-level accuracy, and similar to 3% in the multi-label mean area under the ROC curve at the bag-level., This work was supported by the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C2. The work of A. Schmidt was funded from the European Union's Horizon 2020 research and innovation programme under the Marie SkAodowska Curie grant agreement No 860627 (CLARIFY Project). The hardware NVIDIA DGXA100 used in the experimental part of this work was funded by Generalitat Valenciana/European Union through the European Regional Development Fund (ERDF) of the Valencian Community (IDIFEDER/2020/030). Funding for open access charge: Universitat Politecnica de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Constrained multiple instance learning for ulcerative colitis prediction using histological images

  • del Amor, Rocío
  • Meseguer-Esbrí, Pablo|||0000-0001-7821-6168
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Lorenzo Parigi, Tommaso
  • Villanacci, Vincenzo
  • Launet, Laetitia
  • Bazarova, Alina
  • Tontini, Gian Eugenio
  • Bisschops, Raf
  • De Hertogh, Gert
  • Ferraz, Jose G.
  • Götz, Martin
  • Gui, Xianyong
  • Hayee, Bu Hussain
  • Lazarev, Mark
[EN] Background and Objective: Ulcerative colitis (UC) is an inflammatory bowel disease (IBD) affecting the colon and the rectum characterized by a remitting-relapsing course. To detect mucosal inflammation as-sociated with UC, histology is considered the most stringent criteria. In turn, histologic remission (HR) correlates with improved clinical outcomes and has been recently recognized as a desirable treatment target. The leading biomarker for assessing histologic remission is the presence or absence of neutrophils. Therefore, the finding of this cell in specific colon structures indicates that the patient has UC activity. However, no previous studies based on deep learning have been developed to identify UC based on neu-trophils detection using whole-slide images (WSI). Methods: The methodological core of this work is a novel multiple instance learning (MIL) framework with location constraints able to determine the presence of UC activity using WSI. In particular, we put forward an effective way to introduce constraints about positive instances to effectively explore additional weakly supervised information that is easy to obtain and enjoy a significant boost to the learning process. In addition, we propose a new weighted embedding to enlarge the relevance of the positive instances. Results: Extensive experiments on a multi-center dataset of colon and rectum WSIs, PICASSO-MIL, demon-strate that using the location information we can improve considerably the results at WSI-level. In com-parison with prior MIL settings, our method allows for 10% improvements in bag-level accuracy. Conclusion : Our model, which introduces a new form of constraints, surpass the results achieved from current state-of-the-art methods that focus on the MIL paradigm. Our method can be applied to other histological concerns where the morphological features determining a positive WSI are tiny and similar to others in the image., This work has received funding from Horizon 2020, the European Unions Framework Programme for Research and Innovation, under grant agreement No. 860627 (CLARIFY) , the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN) and GVA through projects PROMETEO/2019/109 and INNEST/2021/321 (SAMUEL) . Roco del Amor and Adrin Colomer work have also been supported by the Spanish Government under FPU Grant (FPU20/05263) and the Universitat Politcnica de Valncia (PAID-10-21-Subprograma 1), respectively. We gratefully acknowledge the support from the Generalitat Valenciana (GVA) with the donation of the DGX A100 used for this work, action co-financed by the European Union through the Operational Program of the European Regional Development Fund of the Comunitat Valenciana 2014¿2020 (IDIFEDER/2020/030).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Federating Medical Deep Learning Models from Private Jupyter Notebooks to Distributed Institutions

  • Launet, Laetitia Mariana
  • Wang, Yuandou
  • Koulouzis, Spiros
  • Bianchi, Riccardo
  • Mosquera-Zamudio, Andrés
  • Monteagudo, Carlos
  • Zhao, Zhiming
  • Colomer, Adrián|||0000-0002-7616-6029
  • Igual García, Jorge|||0000-0003-3408-4014
  • Pulgarín-Ospina, Cristian Camilo|||0000-0002-9759-3791
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Deep learning-based algorithms have led to tremendous progress over the last years, but they face a bottleneck as their optimal development highly relies on access to large datasets. To mitigate this limitation, cross-silo federated learning has emerged as a way to train collaborative models among multiple institutions without having to share the raw data used for model training. However, although artificial intelligence experts have the expertise to develop state-of-the-art models and actively share their code through notebook environments, implementing a federated learning system in real-world applications entails significant engineering and deployment efforts. To reduce the complexity of federation setups and bridge the gap between federated learning and notebook users, this paper introduces a solution that leverages the Jupyter environment as part of the federated learning pipeline and simplifies its automation, the Notebook Federator. The feasibility of this approach is then demonstrated with a collaborative model solving a digital pathology image analysis task in which the federated model reaches an accuracy of 0.8633 on the test set, as compared to the centralized configurations for each institution obtaining 0.7881, 0.6514, and 0.8096, respectively. As a fast and reproducible tool, the proposed solution enables the deployment of a cross-country federated environment in only a few minutes., This work has been partially funded by the European Union s Horizon 2020 research and innovation programme with the project CLARIFY under Marie Sklodowska-Curie (860627), ENVRI-FAIR (824068), BlueCloud (862409), and ARTICONF (825134). This work is also supported by LifeWatch ERIC, GVA through projects PROMETEO/2019/109 and INNEST/2021/321 (SAMUEL), and the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN). The work of Adrián Colomer has been supported by the ValgrAI Valencian Graduate School and Research Network for Artificial Intelligence & Generalitat Valenciana and Universitat Politècnica de València (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A novel self-learning framework for bladder cancer grading using histopathological images

  • García-Pardo, José Gabriel
  • Esteve, Anna
  • Ramos, David
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] In recent times, bladder cancer has increased significantly in terms of incidence and mortality. Currently, two subtypes are known based on tumour growth: non-muscle invasive (NMIBC) and muscle-invasive bladder cancer (MIBC). In this work, we focus on the MIBC subtype because it has the worst prognosis and can spread to adjacent organs. We present a self-learning framework to grade bladder cancer from histological images stained by immunohistochemical techniques. Specifically, we propose a novel Deep Convolutional Embedded Attention Clustering (DCEAC) which allows for the classification of histological patches into different levels of disease severity, according to established patterns in the literature. The proposed DCEAC model follows a fully unsupervised two-step learning methodology to discern between non-tumour, mild and infiltrative patterns from high-resolution 512 x 512 pixel samples. Our system outperforms previous clustering-based methods by including a convolutional attention module, which enables the refinement of the features of the latent space prior to the classification stage. The proposed network surpasses state-of-the-art approaches by 2-3% across different metrics, reaching a final average accuracy of 0.9034 in a multi-class scenario. Furthermore, the reported class activation maps evidence that our model is able to learn by itself the same patterns that clinicians consider relevant, without requiring previous annotation steps. This represents a breakthrough in MIBC grading that bridges the gap with respect to training the model on labelled data., This work has been partially funded by the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska Curie grant agreement No 860627 (CLARIFY Project), the State Research Spanish Agency under the AI4SKIN project (PID2019- 105142RB-C21) and GVA through project PROMETEO/2019/109. The work of Gabriel García has been supported by the State Research Spanish Agency PTA2017-14610-I. The equipment used for this research has been funded by the European Union within the operating Program ERDF of the Valencian Community 2014-2020 with the grant number IDIFEDER/2020/030.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Labeling confidence for uncertainty-aware histology image classification

  • del Amor, Rocío
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Silva-Rodríguez, Julio
[EN] Deep learning-based models applied to digital pathology require large, curated datasets with high-quality (HQ) annotations to perform correctly. In many cases, recruiting expert pathologists to annotate large databases is not feasible, and it is necessary to collect additional labeled data with varying label qualities, e.g., pathologists-in-training (henceforth, non-expert annotators). Learning from datasets with noisy labels is more challenging in medical applications since medical imaging datasets tend to have instance-dependent noise and suffer from high inter/intra-observer variability. In this paper, we design an uncertainty-driven labeling strategy with which we generate soft labels from 10 non-expert annotators for multi-class skin cancer classification. Based on this soft annotation, we propose an uncertainty estimation-based framework to handle these noisy labels. This framework is based on a novel formulation using a dual-branch min-max entropy calibration to penalize inexact labels during the training. Comprehensive experiments demonstrate the promising performance of our labeling strategy. Results show a consistent improvement by using soft labels with standard cross-entropy loss during training (similar to 4.0% F1-score) and increases when calibrating the model with the proposed min-max entropy calibration (similar to 6.6% F1-score). These improvements are produced at negligible cost, both in terms of annotation and calculation., Funding from Horizon 2020, the European Union's Framework Programme for Research and Innovation, under the grant agreement No. 860627 (CLARIFY) , the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN) and GVA through the project INNEST/2021/321 (SAMUEL). The work of Rocio del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263). The work of J. Silva-Rodriguez was carried out during his previous position at Universitat Politecnica de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

ProGleason-GAN: Conditional Progressive Growing GAN for prostatic cancer Gleason Grade patch synthesis

  • Golfe-San Martín, Alejandro|||0000-0001-5095-9904
  • del Amor, Rocío
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Sales, María A.
  • Terradez, Liria
[EN] Background and objective: Prostate cancer is one of the most common diseases affecting men. The main diagnostic and prognostic reference tool is the Gleason scoring system. An expert pathologist assigns a Gleason grade to a sample of prostate tissue. As this process is very time-consuming, some artificial intelligence applications were developed to automatize it. The training process is often confronted with insufficient and unbalanced databases which affect the generalisability of the models. Therefore, the aim of this work is to develop a generative deep learning model capable of synthesising patches of any selected Gleason grade to perform data augmentation on unbalanced data and test the improvement of classification models.Methodology: The methodology proposed in this work consists of a conditional Progressive Growing GAN (ProGleason-GAN) capable of synthesising prostate histopathological tissue patches by selecting the desired Gleason Grade cancer pattern in the synthetic sample. The conditional Gleason Grade information is introduced into the model through the embedding layers, so there is no need to add a term to the Wasserstein loss function. We used minibatch standard deviation and pixel normalisation to improve the performance and stability of the training process.Results: The reality assessment of the synthetic samples was performed with the Frechet Inception Distance (FID). We obtained an FID metric of 88.85 for non-cancerous patterns, 81.86 for GG3, 49.32 for GG4 and 108.69 for GG5 after post-processing stain normalisation. In addition, a group of expert pathologists was selected to perform an external validation of the proposed framework. Finally, the application of our proposed framework improved the classification results in SICAPv2 dataset, proving its effectiveness as a data augmentation method.Conclusions: ProGleason-GAN approach combined with a stain normalisation post-processing provides state-of-the-art results regarding Frechet's Inception Distance. This model can synthesise samples of noncancerous patterns, GG3, GG4 or GG5. The inclusion of conditional information about the Gleason grade during the training process allows the model to select the cancerous pattern in a synthetic sample. The proposed framework can be used as a data augmentation method., This work has received funding & nbsp;from Horizon 2020 , the European Union's Framework Programme for Research and Innovation, under grant agreement no. 860627 (CLARIFY), the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN) and GVA through projects PROME-TEO/2019/109 and INNEST/2021/321 (SAMUEL). The work of Adrian Colomer has been supported by the ValgrAI - Valencian Graduate School and Research Network for Artificial Intelligence & Generalitat Valenciana and Universitat Politecnica de Valencia (PAID-PD-22). Rocio del Amor has been supported by the Spanish Government under FPU Grant (FPU20/05263).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

MICIL: Multiple-Instance Class-Incremental Learning for skin cancer whole slide images

  • Meseguer-Esbrí, Pablo|||0000-0001-7821-6168
  • del Amor, Rocío
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Artificial intelligence (AI) agents encounter the problem of catastrophic forgetting when they are trained in sequentially with new data batches. This issue poses a barrier to the implementation of AI -based models in tasks that involve ongoing evolution, such as cancer prediction. Moreover, whole slide images (WSI) play a crucial role in cancer management, and their automated analysis has become increasingly popular in assisting pathologists during the diagnosis process. Incremental learning (IL) techniques aim to develop algorithms capable of retaining previously acquired information while also acquiring new insights to predict future data. Deep IL techniques need to address the challenges posed by the gigapixel scale of WSIs, which often necessitates the use of multiple instance learning (MIL) frameworks. In this paper, we introduce an IL algorithm tailored for analyzing WSIs within a MIL paradigm. The proposed M ultiple I nstance C lass- I ncremental L earning ( MICIL ) algorithm combines MIL with class-IL for the first time, allowing for the incremental prediction of multiple skin cancer subtypes from WSIs within a class-IL scenario. Our framework incorporates knowledge distillation and data rehearsal, along with a novel embedding-level distillation, aiming to preserve the latent space at the aggregated WSI level. Results demonstrate the algorithm's effectiveness in addressing the challenge of balancing IL-specific metrics, such as intransigence and forgetting, and solving the plasticity-stability dilemma., This work has received funding from Horizon 2020, the European Union's Framework Programme for Research and Innovation, under grant agreement No. 860627 (CLARIFY), the Spanish Ministry of Economy and Competitiveness through projects PID2019-105142RB-C21(AI4SKIN) and PID2022-140189OB-C21 (ASSIST), and GVA through the project INNEST/2021/321 (SAMUEL). The work of Rocio del Amorand Pablo Meseguer has been supported by the Spanish Ministry of Universities under an FPU Grant (FPU20/05263) and valgrAI - Valencian Graduate School and Research Network of Artificial Intelligence, respectively.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Deep Learning for Skin Melanocytic Tumors in Whole-Slide Images: A Systematic Review

  • Mosquera-Zamudio, Andrés
  • Launet, Laetitia
  • Tabatabaei, Zahra
  • Parra-Medina, Rafael
  • Oliver Moll, Javier
  • Monteagudo, Carlos
  • Janssen, Emiel
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Simple Summary Deep learning (DL) is expanding into the surgical pathology field and shows promising outcomes in diminishing subjective interpretations, especially in dermatopathology. We aim to show the efforts of implementing DL models for melanocytic tumors in whole slide images. Four electronic databases were systematically searched, and 28 studies were identified. Our analysis revealed four research trends: DL models vs. pathologists, diagnostic prediction, prognosis, and regions of interest. We also highlight relevant issues that must be considered to implement these models in real scenarios taking into account pathologists' and engineers' perspectives. The rise of Artificial Intelligence (AI) has shown promising performance as a support tool in clinical pathology workflows. In addition to the well-known interobserver variability between dermatopathologists, melanomas present a significant challenge in their histological interpretation. This study aims to analyze all previously published studies on whole-slide images of melanocytic tumors that rely on deep learning techniques for automatic image analysis. Embase, Pubmed, Web of Science, and Virtual Health Library were used to search for relevant studies for the systematic review, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. Articles from 2015 to July 2022 were included, with an emphasis placed on the used artificial intelligence methods. Twenty-eight studies that fulfilled the inclusion criteria were grouped into four groups based on their clinical objectives, including pathologists versus deep learning models (n = 10), diagnostic prediction (n = 7); prognosis (n = 5), and histological features (n = 6). These were then analyzed to draw conclusions on the general parameters and conditions of AI in pathology, as well as the necessary factors for better performance in real scenarios., This work has received funding from the European Union's Horizon 2020 Programme for Research and Innovation, under the Marie Sklodowska Curie grant agreement No. 860627 (CLARIFY). The work is also supported by project INNEST/2021/321 (SAMUEL), PAID-10-21 - Subprograma 1 and PAID-PD-22 for postdoctoral research, and PI20/00094, Instituto de Salud Carlos III, y Fondos Europeos FEDER.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A Transparent Search-Based Framework for Skin and Breast Cancer Diagnosis Using Siamese Networks

  • Tabatabaei, Zahra
  • Oliver Moll, Javier
  • Naranjo, Valery
  • Colomer, Adrián|||0000-0002-7616-6029
[EN] Computer aid diagnosis has developed digital pathology with Deep Learning (DL)-based tools to assist pathologists in decision-making. Content-Based Histopathological Image Retrieval (CBHIR) is a novel tool to seek highly correlated patches in terms of similarity in histopathological features. In this work, we proposed two CBHIR approaches on breast (Breast-twins) and skin cancer (Skin-twins) datasets for robust and accurate patch-level retrieval, integrating a custom-built Siamese network as a feature extractor. The proposed Siamese network generalizes for unseen images by focusing on the similar histopathological features of the input pairs. The proposed CBHIR approaches are evaluated on the Breast (public) and Skin (private) datasets with top K accuracy. Finding the optimum amount of K is challenging; also, as much as K increases, the dissimilarity between the query and the returned images increases, which might mislead the pathologists. This paper is tackling this issue for the first time on histopathological images by evaluating the top first retrieved images, while such approaches are underexplored in existing literature. The Breasttwins model achieves 70% of the F1-score at the top first, which exceeds the other recent methods at a higher amount of K such as 5 and 400. Skin-twins overpasses the recently proposed Convolutional Auto Encoder by 67%, increasing the precision. Besides, the Skin-twins model tackles the challenges of Spitzoid Tumors of Uncertain Malignant Potential (STUMP) to assist pathologists with retrieving top K images and their corresponding labels. So, this approach can offer a more explainable tool to pathologists in terms of transparency, trustworthiness, or reliability, This work was supported by a research grant (VIL57398) from VILLUM FONDEN. Also, the skin data is from the CLARIFY Project, which is funded by the European Union s Horizon 2020 research and innovation program under the Marie Skodowska-Curie grant agreement No. 860627.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Efficient Cancer Classification by Coupling Semi Supervised and Multiple Instance Learning

  • Schmidt, Arne
  • Silva-Rodríguez, Julio
  • Molina, Rafael
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] The annotation of large datasets is often the bottleneck in the successful application of artificial intelligence in computational pathology. For this reason recently Multiple Instance Learning (MIL) and Semi Supervised Learning (SSL) approaches are gaining popularity because they require fewer annotations. In this work we couple SSL and MIL to train a deep learning classifier that combines the advantages of both methods and overcomes their limitations. Our method is able to learn from the global WSI diagnosis and a combination of labeled and unlabeled patches. Furthermore, we propose and evaluate an efficient labeling paradigm that guarantees a strong classification performance when combined with our learning framework. We compare our method to SSL and MIL baselines, the state-of-the-art and completely supervised training. With only a small percentage of patch labels our proposed model achieves a competitive performance on SICAPv2 (Cohen's kappa of 0.801 with 450 patch labels), PANDA (Cohen's kappa of 0.794 with 22,023 patch labels) and Camelyon16 (ROC AUC of 0.913 with 433 patch labels). Our code is publicly available at https://github.com/arneschmidt/ssl_and_mil_cancer_classification., This work was supported in part by the European Union's Horizon 2020 Research and Innovation Program through the Marie Skodowska Curie (Cloud Artificial Intelligence For pathologY (CLARIFY) Project) under Grant 860627, and in part by the Spanish Ministry of Science and Innovation under Project PID2019-105142RB-C22.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Artificial intelligence in computational pathology - challenges and future directions

  • Morales, Sandra|||0000-0003-0763-1545
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Engan, Kjersti
[EN] The field of digital histopathology has seen incredible growth in recent years. Digital pathology is becoming a relevant tool in healthcare, industrial and research sectors to reduce the saturation of pathology departments and improve the productivity of pathologists by increasing diagnostic accuracy and reducing turnaround times. Artificial Intelligence (AI) algorithms may be used for the identification of relevant regions, extraction of features from a histological image and overall classification of images into specific classes. The combination of digital histopathology imaging and AI therefore presents a significant opportunity for the support of the pathologists' tasks and opens up a whole new world of computational analysis. In this paper, we have analysed the present, the challenges and the future of the computational pathology discussing the different existing strategies to overcome its main limitations and ensure the computational pathology acceptance. The lack of labelled data, which is the possibly largest challenge for all medical AI applications, is even more pronounced in computational pathology because of the multi-gigapixel nature of the images and high data heterogeneity. We consider the future of the computational pathology is the combination of weak label strategies with active learning and crowdsourcing scenarios since it would remove some of the workload from clinical experts and manual annotation obtaining clinically satisfactory performance with minimal annotation effort. In addition, we believe areas such as explainable AI, data fusion and secure role-based data sharing will be receiving increasing research attention in computational pathology in the close future., This work has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skodowska Curie grant agreement No 860627 (CLARIFY Project) and GVA through project PROMETEO/2019/109. The work of San-dra Morales has been co-funded by the Universitat Politecnica de Valencia through the program PAID-10-20
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Self-contrastive weakly supervised learning framework for prognostic prediction using whole slide images

  • Fuster, Saul
  • Khoraminia, Farbod
  • Silva-Rodríguez, Julio
  • Kiraz, Umay
  • van Leenders, Geert J. L. H.
  • Eftestol, Trygve
  • Janssen, Emiel A.M.
  • Zuiverloon, Tahlita C. M.
  • Engan, Kjersti
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] We present a pioneering investigation into the application of deep learning techniques to analyze histopathological images for addressing the substantial challenge of automated prognostic prediction. Prognostic prediction poses a unique challenge as the ground truth labels are inherently weak, and the model must anticipate future events that are not directly observable in the image. To address this challenge, we propose a novel three-part framework comprising of a convolutional network based tissue segmentation algorithm for region of interest delineation, a contrastive learning module for feature extraction, and a nested multiple instance learning classification module. Our study explores the significance of various regions of interest within the histopathological slides and exploits diverse learning methods in real-world clinical scenarios. The pipeline is initially validated on artificially generated data and a simpler diagnostic task. Transitioning to prognostic prediction, tasks become more challenging. Employing bladder cancer as use case, our best models yield an AUC of 0.721 and 0.678 for recurrence and treatment outcome prediction respectively for a private data cohort. Altogether, this research serves as an initial investigation on the shortcomings of histopathological image analysis for treatment outcome prediction., This research has received funding from the European Union's Horizon 2020 research and innovation program under grant agreements 860627 (CLARIFY). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A deep embedded refined clustering approach for breast cancer distinction based on DNA methylation

  • del Amor, Rocío
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Monteagudo, Carlos
[EN] Epigenetic alterations have an important role in the development of several types of cancer. Epigenetic studies generate a large amount of data, which makes it essential to develop novel models capable of dealing with large-scale data. In this work, we propose a deep embedded refined clustering method for breast cancer differentiation based on DNA methylation. In concrete, the deep learning system presented here uses the levels of CpG island methylation between 0 and 1. The proposed approach is composed of two main stages. The first stage consists in the dimensionality reduction of the methylation data based on an autoencoder. The second stage is a clustering algorithm based on the soft assignment of the latent space provided by the autoencoder. The whole method is optimized through a weighted loss function composed of two terms: reconstruction and classification terms. To the best of the authors¿ knowledge, no previous studies have focused on the dimensionality reduction algorithms linked to classification trained end-to-end for DNA methylation analysis. The proposed method achieves an unsupervised clustering accuracy of 0.9927 and an error rate (%) of 0.73 on 137 breast tissue samples. After a second test of the deep-learning-based method using a different methylation database, an accuracy of 0.9343 and an error rate (%) of 6.57 on 45 breast tissue samples are obtained. Based on these results, the proposed algorithm outperforms other state-of-the-art methods evaluated under the same conditions for breast cancer classification based on DNA methylation data., This work has received funding from Horizon 2020, the European Union's Framework Programme for Research and Innovation, under grant Agreement No. 860627 (CLARIFY), the Spanish Ministry of Economy and Competitiveness through project PID2019105142RB-C21 (AI4SKIN) and SICAP (DPI2016-77869-C2-1-R) and GVA through Project PROMETEO/2019/109.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

WWFedCBMIR: World-Wide Federated Content-Based Medical Image Retrieval

  • Tabatabaei, Zahra
  • Wang, Yuandou
  • Oliver Moll, Javier
  • Zhao, Zhiming
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] The paper proposes a federated content-based medical image retrieval (FedCBMIR) tool that utilizes federated learning (FL) to address the challenges of acquiring a diverse medical data set for training CBMIR models. CBMIR is a tool to find the most similar cases in the data set to assist pathologists. Training such a tool necessitates a pool of whole-slide images (WSIs) to train the feature extractor (FE) to extract an optimal embedding vector. The strict regulations surrounding data sharing in hospitals makes it difficult to collect a rich data set. FedCBMIR distributes an unsupervised FE to collaborative centers for training without sharing the data set, resulting in shorter training times and higher performance. FedCBMIR was evaluated by mimicking two experiments, including two clients with two different breast cancer data sets, namely BreaKHis and Camelyon17 (CAM17), and four clients with the BreaKHis data set at four different magnifications. FedCBMIR increases the F1 score (F1S) of each client from 96% to 98.1% in CAM17 and from 95% to 98.4% in BreaKHis, with 11.44 fewer hours in training time. FedCBMIR provides 98%, 96%, 94%, and 97% F1S in the BreaKHis experiment with a generalized model and accomplishes this in 25.53 fewer hours of training., This study is funded by European Union s Horizon 2020 research and innovation program under the Marie Skodowska-Curie grant agreement No. 860627 (CLARIFY Project). The work of Adrián Colomer has been supported by the ValgrAI Valencian Graduate School and Research Network for Artificial Intelligence and Generalitat Valenciana and Universitat Politècnica de València (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Choosing only the best voice imitators: Top-K many-to-many voice conversion with StarGAN

  • Fernández-Martín, Claudio
  • Panariello, Claudio
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Voice conversion systems have become increasingly important as the use of voice technology grows. Deep learning techniques, specifically generative adversarial networks (GANs), have enabled significant progress in the creation of synthetic media, including the field of speech synthesis. One of the most recent examples, StarGAN-VC, uses a single pair of generator and discriminator to convert voices between multiple speakers. However, the training stability of GANs can be an issue. The Top-K methodology, which trains the generator using only the best K generated samples that "fool"the discriminator, has been applied to image tasks and simple GAN architectures. In this work, we demonstrate that the Top-K methodology can improve the quality and stability of converted voices in a state-of-the-art voice conversion system like StarGAN-VC. We also explore the optimal time to implement the Top-K methodology and how to reduce the value of K during training. Through both quantitative and qualitative studies, it was found that the Top-K methodology leads to quicker convergence and better conversion quality compared to regular or vanilla training. In addition, human listeners perceived the samples generated using Top-K as more natural and were more likely to believe that they were produced by a human speaker. The results of this study demonstrate that the Top-K methodology can effectively improve the performance of deep learning-based voice conversion systems., The work of C. Fernandez-Martin was funded by the Horizon 2020 of the European Union research and innovation programme under the Marie Sklodowska Curie grant agreement No 860627 (CLARIFY Project).The work of Adrian Colomer has been supported by the ValgrAI -Valencian Graduate School and Research Network for Artificial Intelligence & Generalitat Valenciana and Universitat Politecnica de Valencia, Spain (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A Spitzoid Tumor dataset with clinical metadata and Whole Slide Images for Deep Learning models

  • Mosquera-Zamudio, Andrés
  • Launet, Laetitia Mariana
  • Moscardó, Anaïs
  • Monteagudo, Carlos
  • del Amor, Rocío
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Spitzoid tumors (ST) are a group of melanocytic tumors of high diagnostic complexity. Since 1948, when Sophie Spitz first described them, the diagnostic uncertainty remains until now, especially in the intermediate category known as Spitz tumor of unknown malignant potential (STUMP) or atypical Spitz tumor. Studies developing deep learning (DL) models to diagnose melanocytic tumors using whole slide imaging (WSI) are scarce, and few used ST for analysis, excluding STUMP. To address this gap, we introduce SOPHIE: the first ST dataset with WSIs, including labels as benign, malignant, and atypical tumors, along with the clinical information of each patient. Additionally, we explain two DL models implemented as validation examples using this database., This work has received funding from Horizon 2020, the European Union's Framework Programme for Research and Innovation, under the grant agreement No. 860627 (CLARIFY), PI20/ 00094, Instituto de Salud Carlos III and FEDER European Funds, INNEST/2021/321 (SAMUEL) and the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN). The work of Rocio del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263), and that of Adrian Colomer has been supported by the ValgrAI - Valencian Graduate School and Research Network for Artificial Intelligence & Generalitat Valenciana and Universitat Politecnica de Valencia (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Validation of an Artificial Intelligence Model for Breast Cancer Molecular Subtyping Using Hematoxylin and Eosin-Stained Whole-Slide Images in a Population-Based Cohort

  • Kiraz, Umay
  • Fernández-Martín, Claudio
  • Rewcastle, Emma
  • Gudlaugsson, Einar G.
  • Skaland, Ivar
  • Janssen, Emiel A. M.
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Morales, Sandra|||0000-0003-0763-1545
[EN] Simple Summary Breast cancer is a complex disease that can be classified into different biological subtypes. Correctly identifying these subtypes is essential in determining the most effective treatment for each patient. However, current methods such as gene expression testing and immunohistochemistry are either expensive, time-consuming, or not widely available in all healthcare settings. In this study, we explored whether a computer-based approach using artificial intelligence can accurately predict breast cancer subtypes by analyzing routine pathology slides stained with hematoxylin and eosin. This real-world validation study shows that this method can identify certain subtypes with promising accuracy, offering a faster and more accessible alternative to existing techniques. This research may help improve diagnostic processes, especially in hospitals with limited resources, and support more personalized treatment decisions for patients with breast cancer.Abstract Background/Objectives: Breast cancer (BC) is the most commonly diagnosed cancer in women and the leading cause of cancer-related deaths globally. Molecular subtyping is crucial for prognosis and treatment planning, with immunohistochemistry (IHC) being the most commonly used method. However, IHC has limitations, including observer variability, a lack of standardization, and a lack of reproducibility. Gene expression profiling is considered the ground truth for molecular subtyping; unfortunately, this is expensive and inaccessible to many institutions. This study investigates the potential of an artificial intelligence (AI) model to predict BC molecular subtypes directly from hematoxylin and eosin (H&E)-stained whole-slide images (WSIs). Methods: A pretrained deep learning framework based on multiple-instance learning (MIL) was validated on the Stavanger Breast Cancer (SBC) dataset, consisting of 538 BC cases. Three classification tasks were assessed, including two-class [triple negative BC (TNBC) vs. non-TNBC], three-class (luminal vs. HER2-positive vs. TNBC), and four-class (luminal A vs. luminal B vs. HER2-positive vs. TNBC) groups. Performance metrics were used for the evaluation of the AI model. Results: The AI model demonstrated strong performance in distinguishing TNBC from non-TNBC (AUC = 0.823, accuracy = 0.833, F1-score = 0.824). However, performance declined with an increasing number of classes. Conclusions: The study highlights the potential of AI in BC molecular subtyping from H&E WSIs, offering an easily applicable and standardized method to IHC. Future improvements should focus on optimizing multi-class classification and validating AI-based methods against gene expression analyses for enhanced clinical applicability., This work was funded by the Horizon 2020 European Union research and innovation program under the Marie Sklodowska Curie grant agreement No 860627 (CLARIFY Project). This work was partially funded by Ayuda a Primeros Proyectos de Investigacion (PAID-06-23), Vicerrectorado de Investigacion of the Universitat Politecnica de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Toward More Transparent and Accurate Cancer Diagnosis With an Unsupervised CAE Approach

  • Tabatabaei, Zahra
  • Oliver Moll, Javier
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] According to the Global Cancer Observatory, 2020, breast cancer is the most prevalent cancer type in both genders (11.7%), while prostate cancer is the second most common cancer type in men (14.1%). In digital pathology, Content-Based Medical Image Retrieval (CBMIR) is a powerful tool for improving cancer diagnosis by searching for similar histopathological Whole Slide Images (WSIs). CBMIR empowers pathologists to explore similar patches to their query, enhancing diagnostic reliability and accuracy. In this paper, a customized unsupervised Convolutional Auto Encoder (CAE) was developed in the proposed Unsupervised CBMIR (UCBMIR) to replicate the traditional cancer diagnosis workflow, offering the potential to enhance diagnostic accuracy and efficiency by reducing pathologists¿ workload. Furthermore, it provides a more transparent supporting tool for pathologists in cancer diagnosis. UCBMIR was evaluated using two widely used numerical techniques in CBMIR, visual techniques, and compared with a classifier. Validation encompassed three data sets, including an external evaluation to demonstrate its effectiveness. UCBMIR achieved 99% and 80% top 5 recalls on BreaKHis and SICAPv2 with the first evaluation technique while using the second technique, it reached 91% and 70% precision for BreaKHis and SICAPv2, respectively. Moreover, UCBMIR displayed a strong capability to identify diverse patterns, yielding 81% accuracy in the top 5 predictions on an external image from Arvaniti. The proposed unsupervised CBMIR tool delivered 83% accuracy in retrieving images with the same cancer type., This study is funded by the European Union s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 860627 (CLARIFY Project). The work of Adrián Colomer has been supported by Ayuda a Primeros Proyectos de Investigación (PAID-06-22), Vicerrectorado de Investigacion de la Universitat Politecnica de Valencia (UPV).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Tesis doctoral (DoctoralThesis)

Deep learning strategies for histological image retrieval

  • Tabatabaei, Zahra
[ES] Según World Health Organization (WHO), el cáncer es una de las principales causas de muerte a nivel mundial, con cerca de 10 millones de fallecimientos en 2020. Esto significa que aproximadamente una de cada seis muertes es causada por el cáncer. Para prevenir y disminuir esta enorme cantidad de muertes, es necesario un diagnóstico preciso del cáncer. Las técnicas basadas en Deep Learning (DL) han ofrecido algunas técnicas en el Diagnóstico Asistido por Computadora (CAD) para ayudar a los médicos con su diagnóstico. Estas técnicas no solo disminuyen la carga de trabajo de los patólogos, sino que también aumentan la precisión de sus diagnósticos con menos costos. Las colecciones de imágenes de alta resolución, como las láminas histopatológicas y las exploraciones médicas, han mejorado el rendimiento de estas técnicas. En esta tesis, nos enfocamos principalmente en imágenes histopatológicas escaneadas por escáneres de Whole Slide Images (WSI). Estas imágenes se introducen en métodos basados en DL, que emplean Redes Neuronales Convolucionales (CNN) para detectar las anomalías y los patrones en el tejido escaneado. Estas técnicas son capaces de analizar el tejido para disminuir los impactos de los errores humanos en el diagnóstico del cáncer. Content-Based Medical Image Retrieval (CBMIR) es uno de estos métodos que recientemente ha captado la atención de los investigadores en patología digital. En esta tesis, proponemos tres marcos CBMIR sobre imágenes histopatológicas con dos técnicas basadas en DL que se presentan en diferentes escenarios. En cuanto a los obstáculos potenciales que un CBMIR en patología digital podría enfrentar, incluida la limitación de recursos de GPU, la falta de suficientes conjuntos de datos, y las estrictas regulaciones de privacidad de datos para el intercambio de datos. En relación con estas complejidades, nos enfocamos en el aprendizaje federado en la segunda clase de nuestra investigación. En esta sección, combinamos los conceptos de Federated Learning (FL) con un marco CBMIR para imitar un CBMIR Federado Mundial (FedCBMIR) en imágenes histológicas de cáncer de mama. En esta investigación, seguimos tres escenarios para imitar los tres casos de uso de FedCBMIR en el flujo de trabajo médico. En la última contribución de esta tesis, el enfoque principal es una estrategia basada en aprendizaje contrastivo. Proponemos un marco CBMIR que puede superar las técnicas anteriores con el top K (K>1) y también tener un alto rendimiento en la recuperación de imágenes en el top primero. Además, otra contribución de esta tesis es resolver los desafíos que los patólogos tienen al clasificar los Tumores Spitzoides de Potencial Maligno Incierto (STUMP). Los STUMP presentan un dilema diagnóstico debido a su intrincada histología, creando desafíos para establecer parámetros claros entre nevos benignos y melanomas potencialmente malignos. Para ayudar a los patólogos a enfrentar esta complejidad, el marco puede proporcionar parches similares al top K para ellos con sus etiquetas correspondientes. En resumen, los marcos CBMIR y CBHIR propuestos en esta tesis contribuyen al diagnóstico del cáncer de próstata, mama y piel a partir de imágenes histopatológicas mediante el uso de FEs basados en DL en diferentes escenarios. Estos no solo mejoran la precisión y la eficiencia del diagnóstico del cáncer, sino que también prometen facilitar la detección temprana y las estrategias de tratamiento personalizado. Aprovechar estos marcos en el diagnóstico actual del cáncer podría conducir en última instancia a mejores resultados para los pacientes, menores costos de atención médica y una mayor calidad de vida para las personas afectadas por el cáncer de próstata, mama y piel. Estos avances tienen el potencial de impulsar un cambio social positivo y contribuir a la lucha global contra el cáncer., [CA] Segons l'Organització Mundial de la Salut (OMS), el càncer és una de les principals causes de mort a nivell mundial, amb prop de 10 milions de defuncions en 2020. Això significa que aproximadament una de cada sis morts és causada pel càncer. Per prevenir i disminuir aquesta enorme quantitat de morts, és necessari un diagnòstic precís del càncer. Les tècniques basades en Deep Learning (DL) han ofert algunes tècniques en el Diagnòstic Assistit per Ordinador (CAD) per ajudar els metges amb el seu diagnòstic. Aquestes tècniques no només disminueixen la càrrega de treball dels patòlegs, sinó que també augmenten la precisió dels seus diagnòstics amb menys costos. Les col·leccions d'imatges d'alta resolució, com les làmines histopatològiques i les exploracions mèdiques, han millorat el rendiment d'aquestes tècniques. En aquesta tesi, ens enfoquem principalment en imatges histopatològiques escanejades per escàners de Whole Slide Images (WSI). Aquestes imatges s'introdueixen en mètodes basats en DL, que empren Xarxes Neuronals Convolucionals (CNN) per detectar les anomalies i els patrons en el teixit escanejat. Aquestes tècniques són capaces d'analitzar el teixit per disminuir els impactes dels errors humans en el diagnòstic del càncer. El Content-Based Medical Image Retrieval (CBMIR) és un d'aquests mètodes que recentment ha captat l'atenció dels investigadors en patologia digital. En aquesta tesi, proposem tres marcs CBMIR sobre imatges histopatològiques amb dues tècniques basades en DL que es presenten en diferents escenaris. Pel que fa als obstacles potencials que un CBMIR en patologia digital podria afrontar, inclou la limitació de recursos de GPU, la manca de suficients conjunts de dades, i les estrictes regulacions de privadesa de dades per a l'intercanvi de dades. En relació amb aquestes complexitats, ens enfoquem en l'aprenentatge federat en la segona classe de la nostra investigació. En aquesta secció, combinem els conceptes de Federated Learning (FL) amb un marc CBMIR per imitar un CBMIR Federat Mundial (FedCBMIR) en imatges histològiques de càncer de mama. En aquesta investigació, seguim tres escenaris per imitar els tres casos d'ús de FedCBMIR en el flux de treball mèdic. En l'última contribució d'aquesta tesi, l'enfocament principal és una estratègia basada en aprenentatge contrastiu. Proposem un marc CBMIR que pot superar les tècniques anteriors amb el top K (K>1) i també tenir un alt rendiment en la recuperació d'imatges en el top primer. A més, una altra contribució d'aquesta tesi és resoldre els desafiaments que els patòlegs tenen a l'hora de classificar els Tumors Spitzoides de Potencial Maligne Incert (STUMP). Els STUMP presenten un dilema diagnòstic a causa de la seva intricada histologia, creant desafiaments per establir paràmetres clars entre nevus benignes i melanomes potencialment malignes. Per ajudar els patòlegs a enfrontar aquesta complexitat, el marc pot proporcionar parches similars al top K per a ells amb les seves etiquetes corresponents. En resum, els marcs CBMIR i CBHIR proposats en aquesta tesi contribueixen al diagnòstic del càncer de pròstata, mama i pell a partir d'imatges histopatològiques mitjançant l'ús de FEs basats en DL en diferents escenaris. Aquests no només milloren la precisió i l'eficiència del diagnòstic del càncer, sinó que també prometen facilitar la detecció primerenca i les estratègies de tractament personalitzat. Aprofitar aquests marcs en el diagnòstic actual del càncer podria conduir en última instància a millors resultats per als pacients, menors costos d'atenció mèdica i una major qualitat de vida per a les persones afectades pel càncer de pròstata, mama i pell. Aquests avenços tenen el potencial d'impulsar un canvi social positiu i contribuir a la lluita global contra el càncer., [EN] According to the World Health Organization (WHO), cancer is one of the leading causes of death worldwide, with nearly 10 million deaths in 2020. This means that approximately one in six deaths is caused by cancer. To prevent and decrease this enormous number of deaths, an accurate cancer diagnosis is necessary. Deep Learning (DL)-based techniques have offered some methods in Computer-Aided Diagnosis (CAD) to assist doctors with their diagnoses. These techniques not only reduce the workload of pathologists but also increase the accuracy of their diagnoses at lower costs. Collections of high-resolution images, such as histopathological slides and medical scans, have improved the performance of these techniques. In this thesis, we focus mainly on histopathological images scanned by Whole Slide Image (WSI) scanners. These images are introduced into DL-based methods, which employ Convolutional Neural Networks (CNN) to detect anomalies and patterns in the scanned tissue. These techniques can analyze the tissue to reduce the impacts of human errors in cancer diagnosis. Content-Based Medical Image Retrieval (CBMIR) is one of these methods that has recently attracted the attention of researchers in digital pathology. In this thesis, we propose three CBMIR frameworks on histopathological images with two DL-based techniques presented in different scenarios. Regarding potential obstacles that a CBMIR in digital pathology might face, including the limitation of GPU resources, the lack of sufficient datasets, and strict data privacy regulations for data sharing. Considering these complexities, we focus on federated learning in the second part of our research. In this section, we combine the concepts of Federated Learning (FL) with a CBMIR framework to simulate a World-Wide Federated CBMIR (FedCBMIR) on histological images of breast cancer. In this research, we follow three scenarios to mimic the three use cases of FedCBMIR in the medical workflow. In the final contribution of this thesis, the main focus is a contrastive learning-based strategy. We propose a CBMIR framework that can surpass previous techniques with the top K (K>1) and also have high performance in retrieving images at the top first. Additionally, another contribution of this thesis is to solve the challenges that pathologists face in grading Spitzoid Tumors of Uncertain Malignant Potential (STUMP). STUMPs present a diagnostic dilemma due to their intricate histology, creating challenges for establishing clear parameters between benign nevi and potentially malignant melanomas. To assist pathologists in coping with this complexity, the framework can provide top K similar patches for them with their corresponding labels. In summary, the CBMIR and CBHIR frameworks proposed in this thesis contribute to the diagnosis of prostate, breast, and skin cancer from histopathological images using DL-based FEs in different scenarios. These not only improve the accuracy and efficiency of cancer diagnosis but also promise to facilitate early detection and personalized treatment strategies. Leveraging these frameworks in current cancer diagnosis could ultimately lead to better patient outcomes, lower healthcare costs, and a higher quality of life for individuals affected by prostate, breast, and skin cancer. These advances have the potential to drive positive social change and contribute to the global fight against cancer., This study is funded by European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 860627 (CLAR- IFY Project). The work of Adrián Colomer has been supported by the ValgrAI – Valencian Graduate School and Research Network for Artificial Intelligence & Gen- eralitat Valenciana and Universitat Politècnica de València (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

HistoColAi: An open-source web platform for collaborative digital histology image annotation with AI-driven predictive integration

  • Pulgarín-Ospina, Cristian Camilo|||0000-0002-9759-3791
  • del Amor, Rocío
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Silva-Rodríguez, Julio José
[EN] Digital pathology is now a standard component of the pathology workflow, offering numerous benefits such as high-detail whole slide images and the capability for immediate case sharing between hospitals. Recent advances in deep learning-based methods for image analysis make them a potential aid in digital pathology. However, A significant challenge in developing computer-aided diagnostic systems for pathology is the lack of intuitive, open-source web applications for data annotation. This paper proposes a web service that efficiently provides a tool to visualize and annotate digitized histological images, integrating AI-driven predictive insights. While the tool is capable of handling various image formats, its primary use case is for Whole Slide Imaging (WSI) in the TIFF format, specifically tailored for histopathology applications. This innovative integration not only revolutionizes accessibility but also democratizes the utilization of complex deep-learning models for pathologists unfamiliar with such tools. Moreover, to demonstrate the effectiveness of this approach, we present a use case centered on the diagnosis of spindle cell skin neoplasm involving multiple annotators. Additionally, we conduct a usability study, showing the feasibility of the developed tool., This work has received funding from Horizon 2020, the European Union's Framework Programme for Research and Innovation, under the grant agreement No. 860627 (CLARIFY) , the Spanish Ministry of Economy and Competitiveness through project PID2019-105142RB-C21 (AI4SKIN) and GVA through the project INNEST/2021/321 (SAMUEL) . he work of Cristian Camilo Pulgarin Ospina has been supported by the Spanish State Research Agency (PRE2020-093271) . The work of Rocio del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263) . The work of J. Silva-Rodriguez was carried out during his previous position at Universitat Politecnica de Valencia. We gratefully acknowledge the support from the Generalitat Valenciana (GVA) with the donation of the DGX A100 used for this work, an action co-financed by the European Union through the Operational Program of the European Regional Development Fund of the Comunitat Valenciana 2014 2020 (IDIFEDER/2020/030).
Addi. Archivo Digital para la Docencia y la Investigación
Publication Artículo científico (JournalArticle)

Advancing Content-Based Histopathological Image Retrieval Pre-Processing: A Comparative Analysis of the Effects of Color Normalization Techniques

  • Tabatabaei, Zahra
  • Pérez Bueno, Fernando
  • Colomer, Adrián
  • Oliver Moll, Javier
  • Molina, Rafael
  • Valery Naranjo
Published on 1 March 2024, Content-Based Histopathological Image Retrieval (CBHIR) is a search technique based on the visual content and histopathological features of whole-slide images (WSIs). CBHIR tools assist pathologists to obtain a faster and more accurate cancer diagnosis. Stain variation between hospitals hampers the performance of CBHIR tools. This paper explores the effects of color normalization (CN) in a recently proposed CBHIR approach to tackle this issue. In this paper, three different CN techniques were used on the CAMELYON17 (CAM17) data set, which is a breast cancer data set. CAM17 consists of images taken using different staining protocols and scanners in five hospitals. Our experiments reveal that a proper CN technique, which can transfer the color version into the most similar median values, has a positive impact on the retrieval performance of the proposed CBHIR framework. According to the obtained results, using CN as a pre-processing step can improve the accuracy of the proposed CBHIR framework to 97% (a 14% increase), compared to working with the original images., This study was by the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 860627 (CLARIFY Project). The work of Adrián Colomer was supported by Ayuda a Primeros Proyectos de Investigación (PAID-06-22), Vicerrectorado de Investigacion de la Universitat Politecnica de Valencia (UPV).