Showing papers by "Gemma Piella published in 2019"

Machine learning-based phenogrouping in heart failure to identify responders to cardiac resynchronization therapy.

Abstract: The work of S. Sanchez‐Martinez was supported by a fellowship from ‘la Caixa’ Banking Foundation. C. Butakoff was supported by a grant from the Fundacio La Marato de TV3 (n. 20154031), Spain. N. Duchateau was supported by ‘Programme Avenir Lyon Saint‐Etienne’ (PALSE‐IMPULSION‐2016, Lyon, France). MADIT‐CRT was sponsored by Boston Scientific, while no additional funding was provided for this analysis. This study was also partially supported by the Spanish Ministry of Economy and Competitiveness (grant TIN2014‐52923‐R; Maria de Maeztu Units of Excellence Programme ‐ MDM‐2015‐0502) and FEDER.

Medical-based deep curriculum learning for improved fracture classification

Abstract: Current deep-learning based methods do not easily integrate to clinical protocols, neither take full advantage of medical knowledge. In this work, we propose and compare several strategies relying on curriculum learning, to support the classification of proximal femur fracture from X-ray images, a challenging problem as reflected by existing intra- and inter-expert disagreement. Our strategies are derived from knowledge such as medical decision trees and inconsistencies in the annotations of multiple experts, which allows us to assign a degree of difficulty to each training sample. We demonstrate that if we start learning “easy” examples and move towards “hard”, the model can reach a better performance, even with fewer data. The evaluation is performed on the classification of a clinical dataset of about 1000 X-ray images. Our results show that, compared to class-uniform and random strategies, the proposed medical knowledge-based curriculum, performs up to 15% better in terms of accuracy, achieving the performance of experienced trauma surgeons.

Fetal cortical surface atlas parcellation based on growth patterns.

Abstract: Defining anatomically and functionally meaningful parcellation maps on cortical surface atlases is of great importance in surface-based neuroimaging analysis. The conventional cortical parcellation maps are typically defined based on anatomical cortical folding landmarks in adult surface atlases. However, they are not suitable for fetal brain studies, due to dramatic differences in brain size, shape, and properties between adults and fetuses. To address this issue, we propose a novel data-driven method for parcellation of fetal cortical surface atlases into distinct regions based on the dynamic "growth patterns" of cortical properties (e.g., surface area) from a population of fetuses. Our motivation is that the growth patterns of cortical properties indicate the underlying rapid changes of microstructures, which determine the molecular and functional principles of the cortex. Thus, growth patterns are well suitable for defining distinct cortical regions in development, structure, and function. To comprehensively capture the similarities of cortical growth patterns among vertices, we construct two complementary similarity matrices. One is directly based on the growth trajectories of vertices, and the other is based on the correlation profiles of vertices' growth trajectories in relation to a set of reference points. Then, we nonlinearly fuse these two similarity matrices into a single one, which can better capture both their common and complementary information than by simply averaging them. Finally, based on this fused similarity matrix, we perform spectral clustering to divide the fetal cortical surface atlases into distinct regions. By applying our method on 25 normal fetuses from 26 to 29 gestational weeks, we construct age-specific fetal cortical surface atlases equipped with biologically meaningful parcellation maps based on cortical growth patterns. Importantly, our generated parcellation maps reveal spatially contiguous, hierarchical and bilaterally relatively symmetric patterns of fetal cortical surface development.

Revealing heterogeneity of brain imaging phenotypes in Alzheimer's disease based on unsupervised clustering of blood marker profiles

Abstract: Alzheimer's disease (AD) affects millions of people and is a major rising problem in health care worldwide. Recent research suggests that AD could have different subtypes, presenting differences in how the disease develops. Characterizing those subtypes could be key to deepen the understanding of this complex disease. In this paper, we used a multivariate, non-supervised clustering method over blood-based markers to find subgroups of patients defined by distinctive blood marker profiles. Our analysis on ADNI database identified 4 possible subgroups, each with a different blood profile. More importantly, we show that subgroups with different profiles have a different relationship between brain phenotypes detected in magnetic resonance imaging and disease condition.

Global and Regional Changes in Cortical Development Assessed by MRI in Fetuses with Isolated Nonsevere Ventriculomegaly Correlate with Neonatal Neurobehavior.

Abstract: BACKGROUND AND PURPOSE: Fetuses with isolated nonsevere ventriculomegaly (INSVM) are at risk of presenting neurodevelopmental delay. However, the currently used clinical parameters are insufficient to select cases with high risk and determine whether subtle changes in brain development are present and might be a risk factor. The aim of this study was to perform a comprehensive evaluation of cortical development in INSVM by magnetic resonance (MR) imaging and assess its association with neonatal neurobehavior. MATERIALS AND METHODS: Thirty-two INSVM fetuses and 29 healthy controls between 26–28 weeks of gestation were evaluated using MR imaging. We compared sulci and fissure depth, cortical maturation grading of specific areas and sulci and volumes of different brain regions obtained from 3D brain reconstruction of cases and controls. Neonatal outcome was assessed by using the Neonatal Behavioral Assessment Scale at a mean of 4 ± 2 weeks after birth. RESULTS: Fetuses with INSVM showed less profound and underdeveloped sulcation, including the Sylvian fissure (mean depth: controls 16.8 ± 1.9 mm, versus INSVM 16.0 ± 1.6 mm; P = .01), and reduced global cortical grading (mean score: controls 42.9 ± 10.2 mm, versus INSVM: 37.8 ± 9.9 mm; P = .01). Fetuses with isolated nonsevere ventriculomegaly showed a mean global increase of gray matter volume (controls, 276.8 ± 46.0 ×10 mm3, versus INSVM 277.5 ± 49.3 ×10 mm3, P = .01), but decreased mean cortical volume in the frontal lobe (left: controls, 53.2 ± 8.8 ×10 mm3, versus INSVM 52.4 ± 5.4 ×10 mm3; P = CONCLUSIONS: INSVM fetuses showed differences in cortical development, including regions far from the lateral ventricles, that are associated with neonatal neurobehavior. These results suggest the possible use of these parameters to identify cases at higher risk of altered neurodevelopment.

Global Planar Convolutions for improved context aggregation in Brain Tumor Segmentation

Abstract: In this work, we introduce the Global Planar Convolution module as a building-block for fully-convolutional networks that aggregates global information and, therefore, enhances the context perception capabilities of segmentation networks in the context of brain tumor segmentation. We implement two baseline architectures (3D UNet and a residual version of 3D UNet, ResUNet) and present a novel architecture based on these two architectures, ContextNet, that includes the proposed Global Planar Convolution module. We show that the addition of such module eliminates the need of building networks with several representation levels, which tend to be over-parametrized and to showcase slow rates of convergence. Furthermore, we provide a visual demonstration of the behavior of GPC modules via visualization of intermediate representations. We finally participate in the 2018 edition of the BraTS challenge with our best performing models, that are based on ContextNet, and report the evaluation scores on the validation and the test sets of the challenge.

Image-Based 3D Characterization of Abdominal Aortic Aneurysm Deformation After Endovascular Aneurysm Repair

Abstract: An abdominal aortic aneurysm (AAA) is a focal dilation of the abdominal aorta, that if not treated, tends to grow and may rupture. The most common treatment for AAAs is the endovascular aneurysm repair (EVAR), which requires that patients undergo Computed Tomography Angiography (CTA)-based post-operative lifelong surveillance due to the possible appearance of complications. These complications may again lead to AAA dilation and rupture. However, there is a lack of advanced quantitative image-analysis tools to support the clinicians in the follow-up. Currently, the approach is to evaluate AAA diameter changes along time to infer the progress of the patient and the post-operative risk of AAA rupture. An increased AAA diameter is usually associated with a higher rupture risk, but there are some small AAAs that rupture, whereas other larger aneurysms remain stable. This means that the diameter-based rupture risk assessment is not suitable for all the cases, and there is increasing evidence that the biomechanical behavior of the AAA may provide additional valuable information regarding the progression of the disease and the risk of rupture. Hence, we propose a promising methodology for post-operative CTA time-series registration and subsequent aneurysm biomechanical strain analysis. From these strains, quantitative image-based descriptors are extracted using a principal component analysis of the tensile and compressive strain fields. Evaluated on 22 patients, our approach yields a mean area under the curve of 88.6% when correlating the strain-based quantitative descriptors with the long-term patient prognosis. This suggests that the strain information directly extracted from the CTA images is able to capture the biomechanical behavior of the aneurysm without relying on finite element modeling and simulation. Furthermore, the extracted descriptors set the basis for possible future imaging biomarkers that may be used in clinical practice. Apart from the diameter, these biomarkers may be used to assess patient prognosis and to enable informed decision making after an EVAR intervention, especially in difficult uncertain cases.

Machine Learning for Clinical Decision-Making: Challenges and Opportunities

Abstract: 1 August Pi i Sunyer Biomedical Research Institute (IDIBAPS), Barcelona, Spain 2 University Pompeu Fabra, Department of Information and Communication Technologies, Barcelona, Spain 3 University of Zagreb School of Medicine, Department of Cardiovascular Diseases, Zagreb, Croatia 4 Joint Research Centre, European Commission, Brussels, Belgium 5 Computer Science Department, Intelligent Data Science and Artificial Intelligence (IDEAI) Research Center, Universitat Politecnica de Catalunya, Barcelona, Spain 6 Wales Heart Research Institute, School of Medicine, Cardiff University, Cardiff, UK 7 ICREA, Barcelona, Spain

Three-Dimensional Face Reconstruction from Uncalibrated Photographs: Application to Early Detection of Genetic Syndromes

Abstract: Facial analysis from photography supports the early identification of genetic syndromes, but clinically-acquired uncalibrated images suffer from image pose and illumination variability. Although 3D photography overcomes some of the challenges of 2D images, 3D scanners are not typically available. We present an optimization method for 3D face reconstruction from uncalibrated 2D photographs of the face using a novel statistical shape model of the infant face. First, our method creates an initial estimation of the camera pose for each 2D photograph using the average shape of the statistical model and a set of 2D facial landmarks. Second, it calculates the camera pose and the parameters of the statistical model by minimizing the distance between the projection of the estimated 3D face in the image plane of each camera and the observed 2D face geometry. Using the reconstructed 3D faces, we automatically extract a set of 3D geometric and appearance descriptors and we use them to train a classifier to identify facial dysmorphology associated with genetic syndromes. We evaluated our face reconstruction method on 3D photographs of 54 subjects (age range 0–3 years), and we obtained a point-to-surface error of 2.01 \( \pm \) 0.54%, which was a significant improvement over 2.98 \( \pm \) 0.64% using state-of-the-art methods (p < 0.001). Our classifier detected genetic syndromes from the reconstructed 3D faces from the 2D photographs with 100% sensitivity and 92.11% specificity.

Re-Identification and Growth Detection of Pulmonary Nodules without Image Registration Using 3D Siamese Neural Networks

Abstract: Lung cancer follow-up is a complex, error prone, and time consuming task for clinical radiologists. Several lung CT scan images taken at different time points of a given patient need to be individually inspected, looking for possible cancerogenous nodules. Radiologists mainly focus their attention in nodule size, density, and growth to assess the existence of malignancy. In this study, we present a novel method based on a 3D siamese neural network, for the re-identification of nodules in a pair of CT scans of the same patient without the need for image registration. The network was integrated into a two-stage automatic pipeline to detect, match, and predict nodule growth given pairs of CT scans. Results on an independent test set reported a nodule detection sensitivity of 94.7%, an accuracy for temporal nodule matching of 88.8%, and a sensitivity of 92.0% with a precision of 88.4% for nodule growth detection.

Automatic Segmentation Of the Placenta and its Peripheral Vasculature in Volumetric Ultrasound for TTTS Fetal Surgery

Abstract: Twin-to-twin transfusion syndrome is a serious condition that can affect pregnancies when identical twins share the placenta. In these cases, abnormal placental vessel connections (anastomoses) cause an uneven blood distribution between the babies. Ultrasound (US) enormously facilitates the assessment of these cases, but placenta segmentation is still a challenging task due to artifacts and high variability in its position, orientation, shape and appearance. We propose for the first time a fully-automated framework to achieve an accurate segmentation of the placenta and its peripheral vasculature in 3D US. A conditional Generative Adversarial Network is used to automatically identify the placenta. Afterwards, the entire vasculature is extracted using Modified Spatial Kernelized Fuzzy C-Means and Markov Random Fields. The method is tested on singleton and twin pregnancies from 15 to 38 gestational weeks, achieving a mean Dice coefficient of 0.75 ± 0.12 and 0.70 ± 0.14 for the placenta and its vessels.