In the framework of large deformation diffeomorphic metric mapping (LDDMM), we present a practical methodology to integrate prior knowledge about registered shapes in regularizing metric. Our goal is perform rich anatomical shape comparisons from volumetric images with mathematical properties offered by LDDMM framework. We first notion characteristic scale at which image features are deformed. then propose compare variations multi-scale fashion, i.e., several scales simultaneously. this...

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> Within-subject analysis in fMRI essentially addresses two problems, the <emphasis emphasistype="boldital">detection</emphasis> of brain regions eliciting evoked activity and emphasistype="boldital">estimation</emphasis> underlying dynamics. In Makni <etal/>, 2005 2008, a detection-estimation framework has been proposed to tackle these problems jointly, since they are connected one another....

We studied normal and tumorous three-dimensional (3D) microvascular networks in primate rat brain. Tissues were prepared following a new preparation technique intended for high-resolution synchrotron tomography of networks. The resulting 3D images with spatial resolution less than the minimum capillary diameter permit complete description entire vascular network volumes as large tens cubic millimeters. structural properties investigated by several multiscale methods such fractal...

This paper describes the use of a new 3D high-resolution imaging technique dedicated to functional vessels for systematic quantitative study angiogenesis in primate cortex. We present method which permits, using synchrotron X-ray micro-tomography imaging, identification micro-vascular components as well their automatic numerical digitalization and extraction from very large image analysis post-treatments. is used analyze various levels organization postnatal modifications. Comparing newborn-...

We present a new algorithm which merges discontinuities in 3-D images of tubular structures presenting undesirable gaps. The application the proposed method is mainly associated to large microvascular networks. In order recover real network topology, we need fill gaps between closest discontinuous vessels. presented this paper aims at achieving goal. This based on skeletonization segmented followed by tensor voting method. It permits merge most common kinds found robust, easy use, and...

Real-time ultrasound image acquisition is a pivotal resource in the medical community, spite of its limited quality. This poses challenges to registration methods, particularly those driven by intensity values. We address these difficulties novel diffeomorphic technique for tumor tracking series 2-D liver ultrasound. Our method has two main characteristics: 1) each voxel described three features: intensity, local phase, and phase congruency; 2) we compute set forces from either information...

Abstract Sex differences in behavioral and neural characteristics can be caused by cultural influences but also sex-based neurophysiological sensorimotor features. Since signal-response systems influence decision-making, cooperative collaborative behaviors, the anatomical or physiological bases for any difference sensory mechanisms are important to explore. Here, we use uniform scaling nonparametric representations of human cochlea, main organ hearing that imprints its adult-like morphology...

Applications based on Machine Learning models have now become an indispensable part of the everyday life and professional world. A critical question then recently arised among population: Do algorithmic decisions convey any type discrimination against specific groups population or minorities? In this paper, we show importance understanding how a bias can be introduced into automatic decisions. We first present mathematical framework for fair learning problem, specifically in binary...

Predicting tumor growth and its response to therapy remains a major challenge in cancer research strongly relies on models. In this paper, we introduce, calibrate, verify novel image-driven reaction-diffusion model of avascular growth. The allows for proliferation, death spread cells, accounts nutrient distribution hypoxia. It is constrained by longitudinal time series dynamic contrast-enhancement-MRI images. Tumor specific parameters are estimated from two early points used predict the...

Abstract Key questions in paleoneurology concern the timing and emergence of derived cerebral features within human lineage. Endocasts are replicas internal table bony braincase that widely used as a proxy for reconstructing timeline hominin brain evolution fossil record. The accurate identification sulci imprints endocasts is critical assessing topographic extension structural organisation cortical regions hominins. High‐resolution imaging techniques combined with established methods based...

Automatic recommendation systems based on deep neural networks have become extremely popular during the last decade. Some of these can, however, be used in applications that are ranked as High Risk by European Commission AI act—for instance, online job candidate recommendations. When Union, commercial such will required to proper statistical properties with regard potential discrimination they could engender. This motivated our contribution. We present a novel optimal transport strategy...

In the context of Alzheimer's disease, two challenging issues are (1) characterization local hippocampal shape changes specific to disease progression and (2) identification mild-cognitive impairment patients likely convert. literature, is usually solved first detect areas potentially related disease. These then considered as an input solve (2). As alternative this sequential strategy, we investigate use a classification model using logistic regression address both simultaneously. The...

An algorithm dedicated to automatic segmentation of breast magnetic resonance images is presented in this paper. Our approach based on a pipeline that includes denoising step and statistical segmentation. The noise removal preprocessing relies an anisotropic diffusion scheme, whereas the conducted through Markov random field model. continuous updating all parameters governing process enables denoising, partial volume effect also addressed during labeling step. To assess relevance, Jaccard...

In the framework of large deformation diffeomorphic metric mapping (LDDMM), we develop a multi-scale theory for diffeomorphism group based on previous works. The purpose paper is (1) to in details variational approach analysis diffeomorphisms, (2) generalise several scales semidirect product representation and (3) illustrate resulting decomposition synthetic real images. We also show that approaches presented other papers mixture kernels are equivalent.

Abstract Our knowledge of human brain evolution primarily relies on the interpretation palaeoneurological evidence. In this context, an endocast or replica inside bony braincase can be used to reconstruct a timeline cerebral changes that occurred during evolution, including in topographic extension and structural organisation cortical areas. These tracked by identifying imprints, particularly sulci. The description these crucial landmarks fossil endocasts is, however, challenging....

In this article, we explore the use of novel neural-network architectures to distinguish natural seepages from artificial slicks in synthetic aperture radar (SAR) images. They exploit a distinctive property seepages, which is their temporal recurrence same geographical area. This information can be captured different SAR images acquired at location over time, but not necessarily regular time frequency. The proposed are then built as specific block layers, efficiently treat unordered...

As part of fMRI data analysis, the pyhrf package provides a set tools for addressing two main issues involved in intra-subject analysis: (1) localization cerebral regions that elicit evoked activity and (2) estimation activation dynamics also known as Hemodynamic Response Function (HRF) recovery. To tackle these problems, implements Joint Detection-Estimation framework (JDE) which recovers parcel-level HRFs embeds an adaptive spatio-temporal regularization scheme maps. With respect to sole...