Pasar al contenido principal

IA EXPLICABLE HACIA EL DIAGNOSTICO POR COMPUTADOR USABLE Y FIABLE DE ENFERMEDADES NEURODEGENERATIVAS: TENDIENDO CONOCIMIENTO ENTRE EL CEREBRO Y EL OJO A TRAVES DEL NERVIO OPTI

PID2022-138703OB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2022
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDAD DE ZARAGOZA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 10
Encontrada(s) 1 página(s)

Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Explainable artificial intelligence toward usable and trustworthy computer-aided diagnosis of multiple sclerosis from Optical Coherence Tomography

  • Hernandez, Monica
  • Ramon-Julvez, Ubaldo
  • Vilades, Elisa
  • Cordon, Beatriz
  • Mayordomo, Elvira
  • Garcia-Martin, Elena
Background: Several studies indicate that the anterior visual pathway provides information about the dynamics of axonal degeneration in Multiple Sclerosis (MS). Current research in the field is focused on the quest for the most discriminative features among patients and controls and the development of machine learning models that yield computer-aided solutions widely usable in clinical practice. However, most studies are conducted with small samples and the models are used as black boxes. Clinicians should not trust machine learning decisions unless they come with comprehensive and easily understandable explanations. Materials and methods: A total of 216 eyes from 111 healthy controls and 100 eyes from 59 patients with relapsing-remitting MS were enrolled. The feature set was obtained from the thickness of the ganglion cell layer (GCL) and the retinal nerve fiber layer (RNFL). Measurements were acquired by the novel Posterior Pole protocol from Spectralis Optical Coherence Tomography (OCT) device. We compared two black-box methods (gradient boosting and random forests) with a glass-box method (explainable boosting machine). Explainability was studied using SHAP for the black-box methods and the scores of the glass-box method. Results: The best-performing models were obtained for the GCL layer. Explainability pointed out to the temporal location of the GCL layer that is usually broken or thinning in MS and the relationship between low thickness values and high probability of MS, which is coherent with clinical knowledge.Conclusions: The insights on how to use explainability shown in this work represent a first important step toward a trustworthy computer-aided solution for the diagnosis of MS with OCT.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Extending the reach of the point-to-set principle

  • Lutz, Jack H.
  • Lutz, Neil
  • Mayordomo, Elvira
The point-to-set principle of J. Lutz and N. Lutz (2018) has recently enabled the theory of computing to be used to answer open questions about fractal geometry in Euclidean spaces [Rn]. These are classical questions, meaning that their statements do not involve computation or related aspects of logic. In this paper we extend the reach of the point-to-set principle from Euclidean spaces to arbitrary separable metric spaces X. We first extend two algorithmic dimensions—computability-theoretic versions of classical Hausdorff and packing dimensions that assign dimensions [dim(x)] and [Dim(x)] to individual points [x E X] —to arbitrary separable metric spaces and to arbitrary gauge families. Our first two main results then extend the point-to-set principle to arbitrary separable metric spaces and to a large class of gauge families. We demonstrate the power of our extended point-to-set principle by using it to prove new theorems about classical fractal dimensions in hyperspaces.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Insights into traditional Large Deformation Diffeomorphic Metric Mapping and unsupervised deep-learning for diffeomorphic registration and their evaluation

  • Hernandez, Monica
  • Ramon Julvez, Ubaldo
This paper explores the connections between traditional Large Deformation Diffeomorphic Metric Mapping methods and unsupervised deep-learning approaches for non-rigid registration, particularly emphasizing diffeomorphic registration. The study provides useful insights and establishes connections between the methods, thereby facilitating a profound understanding of the methodological landscape. The methods considered in our study are extensively evaluated in T1w MRI images using traditional NIREP and Learn2Reg OASIS evaluation protocols with a focus on fairness, to establish equitable benchmarks and facilitate informed comparisons. Through a comprehensive analysis of the results, we address key questions, including the intricate relationship between accuracy and transformation quality in performance, the disentanglement of the influence of registration ingredients on performance, and the determination of benchmark methods and baselines. We offer valuable insights into the strengths and limitations of both traditional and deep-learning methods, shedding light on their comparative performance and guiding future advancements in the field.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

EPDiff-JF-Net: Adjoint Jacobi Fields for Diffeomorphic Registration Networks

  • Ramón-Júlvez, Ubaldo
  • Hernández, Mónica
  • Mayordomo, Elvira
Large Deformation Diffeomorphic Metric Mapping (LDDMM) is a well-established diffeomorphic registration method, a critical step for many medical imaging applications. Of special interest are its geodesic shooting variants, which enable statistical shape analysis from transformations usable in computational anatomy studies. This paper introduces a novel deep learning-based unsupervised approach for diffeomorphic image registration called EPDiff-JF-Net. Our method predicts an initial velocity field, performs geodesic shooting to obtain the corresponding path of diffeomorphisms, and utilizes an adjoint Jacobi fields layer to calculate the relevant gradients for training from parallel transport along the geodesic. The model is trained in a fully unsupervised end-to-end manner, with no requirement of ground-truth in the training loss. Experimental results on 3D brain MRI datasets demonstrate the effectiveness of EPDiff-JF-Net, outperforming EPDiff-based LDDMM and deep learning methods while significantly reducing computation time.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

EPDIFF-JF-NET: Adjoint Jacobi Fields for Diffeomorphic Registration Networks

  • Ramón Júlvez, Ubaldo
  • Hernández Giménez, Mónica
  • Mayordomo Cámara, Elvira
This paper presents a deep learning unsupervisedapproach for diffeomorphic image registrationcalled EPDiff-JF-Net. We propose a novel paralleltransport layer to compute the gradients necessaryfor training with adjoint Jacobi fields. We test ourmethod on two independent brain MRI datasets andobtain state-of-the-art results.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

PDE-LDDMM meets NODEs: Introducing neural ordinary differential equation solvers in PDE-constrained Large Deformation Diffeomorphic Metric Mapping

  • Hernandez, Monica
Non-rigid image registration is a crucial task in various medical applications, allowing the alignment of images with complex spatial or temporal variations. This paper introduces NODEO-LDDMM and NODEO-PDE-LDDMM, two innovative deep-learning-based approaches that bridge the gap between Large Deformation Diffeomorphic Metric Mapping (LDDMM) and neural ordinary differential equations (NODEs). LDDMM and PDE-LDDMM offer mathematically well-established formulations for diffeomorphic registration, while NODEs provide the flexibility of deep-learning in the solution of the ODEs involved in both methods. Both NODEO-LDDMM and NODEO-PDE-LDDMM include the strengths of deep-learning into LDDMM, enabling a robust optimization with a good balance between accuracy and transformation smoothness in their solutions. Our proposed methods reached or outperformed their traditional counterparts and the nearly diffeomorphic deep-learning-based approaches selected as benchmarks. This work contributes to advancing non-rigid image registration techniques, with a methodology suited to overcome some of the limitations of deep-learning in medical image registration.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Addressing cross-population domain shift in chest X-ray classification through supervised adversarial domain adaptation

  • Musa, Aminu
  • Prasad, Rajesh
  • Hernández, Mónica
Medical image analysis, empowered by artificial intelligence (AI), plays a crucial role in modern healthcare diagnostics. However, the effectiveness of machine learning models hinges on their ability to generalize to diverse patient populations, presenting domain shift challenges. This study explores the domain shift problem in chest X-ray classification, focusing on cross-population variations, especially in underrepresented groups. We analyze the impact of domain shifts across three population datasets acting as sources using a Nigerian chest X-ray dataset acting as the target. Model performance is evaluated to assess disparities between source and target populations, revealing large discrepancies when the models trained on a source were applied to the target domain. To address with the evident domain shift among the populations, we propose a supervised adversarial domain adaptation (ADA) technique. The feature extractor is first trained on the source domain using a supervised loss function in ADA. The feature extractor is then frozen, and an adversarial domain discriminator is introduced to distinguish between the source and target domains. Adversarial training fine-tunes the feature extractor, making features from both domains indistinguishable, thereby creating domain-invariant features. The technique was evaluated on the Nigerian dataset, showing significant improvements in chest X-ray classification performance. The proposed model achieved a 90.08% accuracy and a 96% AUC score, outperforming existing approaches such as multi-task learning (MTL) and continual learning (CL). This research highlights the importance of developing domain-aware models in AI-driven healthcare, offering a solution to cross-population domain shift challenges in medical imaging.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Normality, Relativization, and Randomness

  • Calvert, Wesley
  • Gruner, Emma
  • Mayordomo, Elvira
  • Turetsky, Daniel
  • Villano, Java Darleen
Normal numbers were introduced by Borel and later proven to be a weak notion of algorithmic randomness. We introduce here a natural relativization of normality based on generalized number representation systems. We explore the concepts of supernormal numbers that correspond to semicomputable relativizations, and that of highly normal numbers in terms of computable ones. We prove several properties of these new randomness concepts. Both supernormality and high normality generalize Borel absolute normality. Supernormality is strictly between 2-randomness and effective dimension 1, while high normality corresponds exactly to sequences of computable dimension 1 providing a more natural characterization of this class.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

A Systematic Review of Cross-Population Shifts in Medical Imaging Analysis with Deep Learning

  • Musa, Aminu
  • Prasad, Rajesh
  • Onwualu, Peter
  • Hernández, Mónica
Deep learning has achieved expert-level performance in medical imaging analysis. However, models often fail to generalize across patient populations due to cross-population domain shifts, distributional differences arising from demographic variability, variations in imaging protocols, scanner hardware, and differences in disease prevalence. This challenge limits the real-world deployment and can increase health inequities. This review systematically examines the nature, causes, and impact of cross-population domain shift in deep learning-based medical imaging analysis. We analyzed 50 peer-reviewed studies from 2020 to 2025, evaluating the proposed methodologies for handling population shifts, the datasets employed, and the metrics used to assess performance. Our findings demonstrate that performance degradation ranged from 10–25% when models were tested on unseen populations, emphasizing the substantial impact of domain shifts on model generalizability. The literature reveals that mitigation strategies broadly fall into two categories: data-centric approaches, such as augmentation and harmonization, and model-centric approaches, including domain adaptation, transfer learning, adversarial learning, multi-task learning, and continual learning. While domain adaptation and transfer learning are the most widely used, their performance gains across populations remain modest, ranging from 5–15%, and are not supported by external validation. Our synthesis reveals a significant reliance on large, publicly available datasets from limited regions, with an underrepresentation of data from low- and middle-income countries. Evaluation practices are inconsistent, with few studies employing standardized external test sets. This review provides a structured taxonomy of mitigation techniques, a refined analysis of domain shift characteristics, and an in-depth critique of methodological challenges. We highlight the urgent need for more geographically and demographically inclusive datasets, adaptable modeling techniques, and standardized evaluation protocols to enable accurate and equitable AI-driven diagnostics across diverse populations. Finally, we outline future research directions to guide the development of robust, generalizable, and fair models for medical imaging analysis. El aprendizaje profundo ha alcanzado un rendimiento de nivel experto en el análisis de imágenes médicas. Sin embargo, los modelos a menudo no logran generalizar entre poblaciones de pacientes debido a cambios de dominio entre poblaciones, diferencias de distribución derivadas de la variabilidad demográfica, variaciones en los protocolos de imagen, el hardware del escáner y diferencias en la prevalencia de enfermedades. Este desafío limita la implementación en el mundo real y puede aumentar las inequidades en salud. Esta revisión examina sistemáticamente la naturaleza, las causas y el impacto de los cambios de dominio entre poblaciones en el análisis de imágenes médicas basado en aprendizaje profundo. Analizamos 50 estudios revisados ​​por pares de 2020 a 2025, evaluando las metodologías propuestas para manejar los cambios de población, los conjuntos de datos empleados y las métricas utilizadas para evaluar el rendimiento. Nuestros hallazgos demuestran que la degradación del rendimiento osciló entre el 10 % y el 25 % cuando los modelos se probaron en poblaciones no vistas, lo que enfatiza el impacto sustancial de los cambios de dominio en la generalización del modelo. La literatura revela que las estrategias de mitigación se dividen en dos categorías principales: enfoques centrados en datos, como la ampliación y la armonización, y enfoques centrados en modelos, que incluyen la adaptación de dominio, el aprendizaje por transferencia, el aprendizaje adversario, el aprendizaje multitarea y el aprendizaje continuo. Si bien la adaptación de dominio y el aprendizaje por transferencia son los más utilizados, sus mejoras de rendimiento en diferentes poblaciones siguen siendo modestas, oscilando entre el 5 % y el 15 %, y carecen de validación externa. Nuestra síntesis revela una dependencia significativa de grandes conjuntos de datos públicos procedentes de regiones limitadas, con una escasa representación de datos de países de ingresos bajos y medios. Las prácticas de evaluación son inconsistentes, y pocos estudios emplean conjuntos de prueba externos estandarizados. Esta revisión proporciona una taxonomía estructurada de las técnicas de mitigación, un análisis refinado de las características del cambio de dominio y una crítica exhaustiva de los desafíos metodológicos. Destacamos la necesidad urgente de contar con conjuntos de datos más inclusivos desde el punto de vista geográfico y demográfico, técnicas de modelado adaptables y protocolos de evaluación estandarizados para permitir diagnósticos precisos y equitativos basados ​​en IA en diversas poblaciones. Por último, describimos las futuras líneas de investigación para guiar el desarrollo de modelos robustos, generalizables y justos para el análisis de imágenes médicas.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Understanding Trade-Offs in Continuous Neural Representations for Diffeomorphic Image Registration: A Comparative Study of Implicit Neural Representations and Neural Ordinary Differential Equations

  • Rodriguez-Sanz, Salvador
  • Paesa-Lia, Carlos
  • Hernandez, Monica
Non-rigid image registration is a fundamental problem in medical imaging and a representative example of continuous transformation modeling in image processing. Diffeomorphic registration methods, such as Large Deformation Diffeomorphic Metric Mapping (LDDMM) and its PDE-constrained variants (PDE-LDDMM), provide mathematically grounded formulations with strong geometric guarantees for transformation quality. However, existing approaches face persistent trade-offs between numerical stability, accuracy, and computational efficiency. Recent work has explored implicit neural representations (INRs) and neural ordinary differential equations (NODEs) as flexible neural representations for modeling continuous transformations. Despite their increasing adoption, their practical behavior and limitations in diffeomorphic registration remain insufficiently understood. In this paper, we present a unified formulation of INR- and NODE-based registration methods within LDDMM and PDE-LDDMM, enabling a systematic and controlled comparison across architectures, sampling strategies, and numerical solvers. Our analysis reveals fundamental trade-offs between these approaches. In particular, we show that MLP-based INR formulations introduce significant computational overhead and rely on sampling strategies that can degrade smoothness and lead to the increased occurrence of non-diffeomorphic transformations at higher resolutions. Moreover, these approximations do not fully alleviate the computational cost, with some variants exceeding the costs of expensive classical optimization-based methods. In contrast, NODE-based formulations and downsampling strategies consistently provide transformations with more controlled Jacobian extrema while maintaining competitive computational performance. Among the evaluated methods, the original NODE-LDDMM and NODE-PDE-LDDMM formulations achieve the most favorable trade-offs between registration accuracy, geometric consistency, and computational efficiency. These findings provide clear insights into the design of neural representations for continuous transformation modeling, with practical implications for diffeomorphic registration and computational anatomy applications.