Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 6
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

PSPU-Net for Automatic Short Axis Cine MRI Segmentation of Left and Right Ventricles

  • Pérez-Pelegrí, Manuel
  • Moratal, David|||0000-0002-2825-3646
  • Monmeneu, Jose V.
  • López-Lereu, María P.
  • Ruiz-España, Silvia
  • Del-Canto, Irene
  • Bodi, Vicente
[EN] Characterization of the heart anatomy and function is mostly done with magnetic resonance image cine series. To achieve a correct characterization, the volume of the right and left ventricle need to be segmented, which is a timeconsuming task. We propose a new convolutional neural network architecture that combines U-net with PSP modules (PSPU-net) for the segmentation of left and right ventricle cavities and left ventricle myocardium in the diastolic frame of short-axis cine MRI images and compare its results against a classic 3D U-net architecture. We used a dataset containing 399 cases in total. The results showed higher quality results in both segmentation and final volume estimation for a test set of 99 cases in the case of the PSPU-net, with global dice metrics of 0.910 and median absolute relative errors in volume estimations of 0.026 and 0.039 for the left ventricle cavity and myocardium and 0.051 for the right ventricles cavity., DM acknowledges financial support from the Conselleria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana (grants AEST/2019/037 and AEST/2020/029), from the Agencia Valenciana de la Innovacion, Generalitat Valenciana (ref. INNCAD00/19/085), and from the Centro para el Desarrollo Tecnologico Industrial (Programa Eurostars-2, actuacion Interempresas Internacional), Spanish Ministerio de Ciencia, Innovacion y Universidades (ref. CIIP20192020). We are grateful to Andres Larroza for his valuable technical assistance in the project.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Automatic left ventricle volume calculation with explainability through a deep learning weak-supervision methodology

  • Pérez-Pelegrí, Manuel
  • Moratal, David|||0000-0002-2825-3646
  • Monmeneu, Jose V.
  • López-Lereu, María P.
  • Pérez-Pelegrí, Lucía
  • Maceira, Alicia M.
  • Bodi, Vicente
[EN] Background and objective: Magnetic resonance imaging is the most reliable imaging technique to assess the heart. More specifically there is great importance in the analysis of the left ventricle, as the main pathologies directly affect this region. In order to characterize the left ventricle, it is necessary to extract its volume. In this work we present a neural network architecture that is capable of directly estimating the left ventricle volume in short axis cine Magnetic Resonance Imaging in the end-diastolic frame and provide a segmentation of the region which is the basis of the volume calculation, thus offering explain-ability to the estimated value. Methods: The network was designed to directly target the volumes to estimate, not requiring any labeled segmentation on the images. The network was based on a 3D U-net with extra layers defined in a scan-ning module that learned features like the circularity of the objects and the volumes to estimate in a weakly-supervised manner. The only targets defined were the left ventricle volumes and the circularity of the object detected through the estimation of the pi value derived from its shape. We had access to 397 cases corresponding to 397 different subjects. We randomly selected 98 cases to use as test set. Results: The results show a good match between the real and estimated volumes in the test set, with a mean relative error of 8% and a mean absolute error of 9.12 ml with a Pearson correlation coefficient of 0.95. The derived segmentations obtained by the network achieved Dice coefficients with a mean value of 0.79. Conclusions: The proposed method is capable of obtaining the left ventricle volume biomarker in the end-diastole and offer an explanation of how it obtains the result in the form of a segmentation mask without the need of segmentation labels to train the algorithm, making it a potentially more trustworthy method for clinicians and a way to train neural networks more easily when segmentation labels are not readily available., The authors acknowledge financial support from the Consel-leria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana (grants AEST/2019/037 and AEST/2020/029) , from the Agencia Valenciana de la Innovacion, Generalitat Valenciana (ref. INNCAD00/19/085) , and from the Centro para el Desarrollo Tecnologico Industrial (Programa Eurostars2, actuacion Interempresas Internacional) , Spanish Ministerio de Ciencia, Innovacion y Universidades (ref. CIIP-20192020) .
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Glioblastomas and brain metastases differentiation following an MRI texture analysis-based radiomics approach

  • Ortiz-Ramón, Rafael
  • Ruiz-España, Silvia
  • Molla-Olmos, Enrique
  • Moratal, David|||0000-0002-2825-3646
[EN] Purpose: To evaluate the potential of 2D texture features extracted from magnetic resonance (MR) images for differentiating brain metastasis (BM) and glioblastomas (GBM) following a radiomics approach. Methods: This retrospective study included 50 patients with BM and 50 with GBM who underwent T1-weighted MRI between December 2010 and January 2017. Eighty-eight rotation-invariant texture features were computed for each segmented lesion using six texture analysis methods. These features were also extracted from the four images obtained after applying the discrete wavelet transform (88 features x 4 images). Three feature selection methods and five predictive models were evaluated. A 5-fold cross-validation scheme was used to randomly split the study group into training (80 patients) and testing (20 patients), repeating the process ten times. Classification was evaluated computing the average area under the receiver operating characteristic curve. Sensibility, specificity and accuracy were also computed. The whole process was tested quantizing the images with different gray-level values to evaluate their influence in the final results. Results: Highest classification accuracy was obtained using the original images quantized with 128 gray-levels and a feature selection method based on the p-value. The best overall performance was achieved using a support vector machine model with a subset of 32 features (AUC = 0.896 +/- 0.067, sensitivity of 82% and specificity of 80%). Naive Bayes and k-nearest neighbors models showed also valuable results (AUC approximate to 0.8) with a lower number of features (< 13), thus suggesting that these models may be more generalizable when using external validations. Conclusion: The proposed radiomics MRI approach is able to discriminate between GBM and BM with high accuracy employing a set of 2D texture features, thus helping in the diagnosis of brain lesions in a fast and noninvasive way., This work has been partially funded by the Spanish Ministerio de Economia y Competitividad (MINECO, Spain) and FEDER funds [grant number BFU2015-64380-C2-2-R]. David Moratal acknowledges financial support from the Conselleria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana (grants AEST/2017/013, AEST/2018/021, and AEST/2019/037), from the Agencia Valenciana de la Innovacion, Generalitat Valenciana (ref. INNCAD00/19/085), and from the Centro para el Desarrollo Tecnologico Industrial (Programa Eurostars-2, actuacion Interempresas Internacional), Spanish Ministerio de Ciencia, Innovacion y Universidades (ref. CIIP-20192020). Rafael Ortiz-Ramon was supported by grant ACIF/2015/078 from the Conselleria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana (Spain). Document
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Magnetic Resonance Assessment of Left Ventricular Ejection Fraction at Any Time Post-Infarction for Prediction of Subsequent Events in a Large Multicenter STEMI Registry

  • Gavara-Doñate, Josep|||0000-0002-3483-7066
  • Moratal, David|||0000-0002-2825-3646
  • Marcos-Garces, Victor
  • Lopez-Lereu, Maria P.
  • Monmeneu, José V.
  • Rios-Navarro, César
  • De Dios, Elena
  • Pérez, Nerea
  • Merenciano, Héctor
  • Gabaldon, Ana
  • Cànoves, Joaquim
  • Racugno, Paolo
  • Bonanad, Clara
  • Miñana, Gema
  • Nunez, Julio
  • Nunez, Eduardo
[EN] Background Magnetic resonance imaging (MRI) is the most accurate imaging technique for left ventricular ejection fraction (LVEF) quantification, but as yet the prognostic value of LVEF assessment at any time after ST-segment elevation myocardial infarction (STEMI) for subsequent major adverse cardiac event (MACE) prediction is uncertain. Purpose To explore the prognostic impact of MRI-derived LVEF at any time post-STEMI to predict subsequent MACE (cardiovascular death or re-admission for acute heart failure). Study Type Prospective. Population One thousand thirteen STEMI patients were included in a multicenter registry. Field Strength/Sequence 1.5-T. Balanced steady-state free precession (cine imaging) and segmented inversion recovery steady-state free precession (late gadolinium enhancement) sequences. Assessment Post-infarction MRI-derived LVEF (reduced [r]: <40%; mid-range [mr]: 40%-49%; preserved [p]: >= 50%) was sequentially quantified at 1 week and after >3 months of follow-up. Statistical Tests Multi-state Markov model to determine the prognostic value of each LVEF state (r-, mr- or p-) at any time point assessed to predict subsequent MACE. A P-value During a 6.2-year median follow-up, 105 MACE (10%) were registered. Transitions toward improved LVEF predominated and only r-LVEF (at any time assessed) was significantly related to a higher incidence of subsequent MACE. The observed transitions from r-LVEF, mr-LVEF, and p-LVEF states to MACE were: 15.3%, 6%, and 6.7%, respectively. Regarding the adjusted transition intensity ratios, patients in r-LVEF state were 4.52-fold more likely than those in mr-LVEF state and 5.01-fold more likely than those in p-LVEF state to move to MACE state. Nevertheless, no significant differences were found in transitions from mr-LVEF and p-LVEF states to MACE state (P-value = 0.6). Data Conclusion LVEF is an important MRI index for simple and dynamic post-STEMI risk stratification. Detection of r-LVEF by MRI at any time during follow-up identifies a subset of patients at high risk of subsequent events. Level of Evidence 2 Technical Efficacy Stage 2, This work was supported by "Instituto de Salud Carlos III," "Fondos Europeos de Desarrollo Regional FEDER" (grants PI15/00531, PI17/01836, PI20/00637, and CIBERCV16/11/00486), and "Marato TV3" (grant 20153030-31-32), a grant from the Catalonian Society of Cardiology 2015 and a grant from La Caixa Foundation (HR17-00527). David Moratal and Jose Gavara acknowledge financial support from the "Conselleria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana" (grants AEST/2019/037 and AEST/2020/029), "Agencia Valenciana de la Innovacion, Generalitat Valenciana" (ref. INNCAD00/19/085 and INNCAD/2020/84), and "Centro para el Desarrollo Tecnologico Industrial" (Programa Eurostars-2, actuacion Interempresas Internacional), Spanish "Ministerio de Ciencia, Innovacion y Universidades" (ref. CIIP-20192020).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

Comparative Analysis of Tagging and Feature-Tracking Cardiac MRI Techniques for the Evaluation of Cardiac Deformation

  • Tejero-Cervera, Pascual
  • Del-Canto, Irene
  • López-Lereu, María P.
  • Monmeneu, Jose V.
  • Ruiz-España, Silvia
  • Santabárbara, José Manuel
  • Maceira, Alicia
  • Bodi, Vicente
  • Moratal, David|||0000-0002-2825-3646
[EN] Currently, there is a high prevalence of cardiovascular diseases, being the leading cause of mortality worldwide. Thus, a rapid and effective evaluation of the heart is increasingly relevant to diagnose the patient early and achieve appropriate treatment. For this reason, this study aims to compare myocardial deformation parameters such as circumferential strain (segmental, base, mid, apical and global) and global torsion values by the CMR tissue tagging reference technique in SPAMM cine images and the promising feature-tracking technique in b-SSFP cine images, in 31 healthy subjects and 35 patients diagnosed with infarct. The results obtained showed that there is a high agreement in the inter-technique and inter-observer reproducibility for all myocardial parameters evaluated. Therefore, it can be determined the CMR featuretracking technique is highly reproducible for circumferential strain and torsion parameters and could currently be a reliable clinical alternative to assess myocardial function., DM acknowledges financial support from the Conselleria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana (grants AEST/2019/037 and AEST/2020/029), from the Agencia Valenciana de la Innovacion, Generalitat Valenciana (ref. INNCAD00/19/085), and from the Centro para el Desarrollo Tecnologico Industrial (Programa Eurostars-2, Interempresas Internacional), Spanish Ministerio de Ciencia, Innovacion y Universidades (ref. CIIP-20192020).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Determination of Non-Invasive Biomarkers for the Assessment of Fibrosis, Steatosis and Hepatic Iron Overload by MR Image Analysis. A Pilot Study

  • Meneses, Alba
  • Santabárbara, José Manuel
  • Romero, Juan Antonio
  • Aliaga, Roberto
  • Maceira, Alicia María
  • Moratal, David|||0000-0002-2825-3646
[EN] The reference diagnostic test of fibrosis, steatosis, and hepatic iron overload is liver biopsy, a clear invasive procedure. The main objective of this work was to propose HSA, or human serum albumin, as a biomarker for the assessment of fibrosis and to study non-invasive biomarkers for the assessment of steatosis and hepatic iron overload by means of an MR image acquisition protocol. It was performed on a set of eight subjects to determine fibrosis, steatosis, and hepatic iron overload with four different MRI sequences. We calibrated longitudinal relaxation times (T1 [ms]) with seven human serum albumin (HSA [%]) phantoms, and we studied the relationship between them as this protein is synthesized by the liver, and its concentration decreases in advanced fibrosis. Steatosis was calculated by means of the fat fraction (FF [%]) between fat and water liver signals in "fat-only images" (the subtraction of in-phase [IP] images and out-of-phase [OOP] images) and in "water-only images" (the addition of IP and OOP images). Liver iron concentration (LIC [mu mol/g]) was obtained by the transverse relaxation time (T2* [ms]) using Gandon's method with multiple echo times (TE) in T2-weighted IP and OOP images. The preliminary results showed that there is an inverse relationship (r = -0.9662) between the T1 relaxation times (ms) and HSA concentrations (%). Steatosis was determined with FF > 6.4% and when the liver signal was greater than the paravertebral muscles signal, and thus, the liver appeared hyperintense in fat-only images. Hepatic iron overload was detected with LIC > 36 mu mol/g, and in these cases, the liver signal was smaller than the paravertebral muscles signal, and thus, the liver behaved as hypointense in IP images., This research was funded by "Conselleria d'Educacio, Investigacio, Cultura i Esport, Generalitat Valenciana" (grants AEST/2019/037 and AEST/2020/029), from the "Agencia Valenciana de la Innovacion, Generalitat Valenciana" (ref. INNCAD00/19/085 and INNCAD/2020/84), and from the "Centro para el Desarrollo Tecnologico Industrial" (Programa Eurostars-2, actuacion Interempresas Internacional), Spanish "Ministerio de Ciencia, Innovacion y Universidades" (ref. CIIP-20192020).