A5089640308- Medical Image Segmentation Techniques
- Advanced Neuroimaging Techniques and Applications
- Advanced MRI Techniques and Applications
- Medical Imaging and Analysis
- Medical Imaging Techniques and Applications
- Anatomy and Medical Technology
- Cell Image Analysis Techniques
- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Functional Brain Connectivity Studies
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Generative Adversarial Networks and Image Synthesis
- Optical Coherence Tomography Applications
- Image Retrieval and Classification Techniques
- Advanced Fluorescence Microscopy Techniques
- 3D Shape Modeling and Analysis
- Digital Image Processing Techniques
- Advanced Vision and Imaging
- MRI in cancer diagnosis
- Brain Tumor Detection and Classification
- Image and Signal Denoising Methods
- Advanced Image and Video Retrieval Techniques
- Cerebrovascular and Carotid Artery Diseases
- Robotics and Sensor-Based Localization
University College London
2018-2025
Athinoula A. Martinos Center for Biomedical Imaging
2021-2024
Harvard University
2021-2024
Massachusetts General Hospital
2021-2024
Wellcome Centre for Human Neuroimaging
2018-2022
National Hospital for Neurology and Neurosurgery
2019-2022
CEA Paris-Saclay
2016-2020
Université Paris-Saclay
2016-2020
Commissariat à l'Énergie Atomique et aux Énergies Alternatives
2015-2020
Centre National de la Recherche Scientifique
2016-2020
Image registration is a fundamental medical image analysis task, and wide variety of approaches have been proposed. However, only few studies comprehensively compared on range clinically relevant tasks. This limits the development methods, adoption research advances into practice, fair benchmark across competing approaches. The Learn2Reg challenge addresses these limitations by providing multi-task data set for comprehensive characterisation deformable algorithms. A continuous evaluation...
Every year, millions of brain magnetic resonance imaging (MRI) scans are acquired in hospitals across the world. These have potential to revolutionize our understanding many neurological diseases, but their morphometric analysis has not yet been possible due anisotropic resolution. We present an artificial intelligence technique, "SynthSR," that takes clinical MRI with any MR contrast (T1, T2, etc.), orientation (axial/coronal/sagittal), and resolution turns them into high-resolution T1...
Statistical Parametric Mapping (SPM) is an integrated set of methods for testing hypotheses about the brain's structure and function, using data from imaging devices. These are implemented in open source software package, SPM, which has been continuous development more than 30 years by international community developers. This paper reports release SPM 25.01, a major new version that incorporates novel analysis methods, optimisations existing as well improved practices science development.
Most existing algorithms for automatic 3D morphometry of human brain MRI scans are designed data with near-isotropic voxels at approximately 1 mm resolution, and frequently have contrast constraints as well-typically requiring T1-weighted images (e.g., MP-RAGE scans). This limitation prevents the analysis millions acquired large inter-slice spacing in clinical settings every year. In turn, inability to quantitatively analyze these hinders adoption quantitative neuro imaging healthcare, also...
Magnetic resonance imaging (MRI) is the standard tool to image human brain
Brain cells are arranged in laminar, nuclear, or columnar structures, spanning a range of scales. Here, we construct reliable cell census the frontal lobe human cerebral cortex at micrometer resolution magnetic resonance imaging (MRI)-referenced system using innovative and analysis methodologies. MRI establishes macroscopic reference coordinate laminar cytoarchitectural boundaries. Cell counting is obtained with digital stereological approach on 3D reconstruction cellular from custom-made...
Abstract Purpose Brain maps of the MRI parameters R 2 * and magnetic susceptibility ( χ ) enable investigation microscopic tissue changes associated with brain disease in patient populations. However, are computed from gradient-echo data acquired at multiple echo times affected by cardiac-induced physiological noise. In this study, we introduce ISME – a sampling strategy that minimizes level noise . Methods Cardiac-induced causes exponential-like effects on decay signal magnitude vary across...
Accurate labeling of specific layers in the human cerebral cortex is crucial for advancing our understanding neurodevelopmental and neurodegenerative disorders. Building on recent advancements ultra-high-resolution ex vivo MRI, we present a novel semi-supervised segmentation model capable identifying supragranular infragranular MRI with unprecedented precision. On dataset consisting 17 whole-hemisphere scans at 120 $\mu $m, propose Multi-resolution U-Nets framework that integrates global...
Serial sectioning optical coherence tomography (OCT) enables accurate volumetric reconstruction of several cubic centimeters human brain samples. We aimed to identify anatomical features the ex vivo brain, such as intraparenchymal blood vessels and axonal fiber bundles, from OCT data in 3D, using intrinsic contrast.We developed an automatic processing pipeline enable characterization microvascular network samples.We demonstrated extraction down a 20 μm diameter filtering strategy followed by...
Motion artifacts can negatively impact diagnosis, patient experience, and radiology workflow especially when a recall is required. Detecting motion while the still in scanner could potentially improve reduce costs by enabling immediate corrective action. We demonstrate clinical k-space dataset that using cross-correlation between adjacent phase-encoding lines detect directly from raw multi-shot multi-slice scans. train split-attention residual network to examine performance predicting...
Serial sectioning Optical Coherence Tomography (sOCT) is a high-throughput, label free microscopic imaging technique that becoming increasingly popular to study post-mortem neurovasculature. Quantitative analysis of the vasculature requires highly accurate segmentation; however, sOCT has low signal-to-noise-ratio and displays wide range contrasts artifacts depend on acquisition parameters. Furthermore, labeled data scarce extremely time consuming generate. Here, we leverage synthetic...
We present open-source tools for three-dimensional (3D) analysis of photographs dissected slices human brains, which are routinely acquired in brain banks but seldom used quantitative analysis. Our can: (1) 3D reconstruct a volume from the and, optionally, surface scan; and (2) produce high-resolution segmentation into 11 regions per hemisphere (22 total), independently slice thickness. can be as substitute ex vivo magnetic resonance imaging (MRI), requires access to an MRI scanner, scanning...
In biomedical research, cell analysis is important to assess physiological and pathophysiological information. Virtual microscopy offers the unique possibility study compositions of tissues at a cellular scale. However, images acquired such high spatial resolution are massive, contain complex information, therefore difficult analyze automatically. this article, we address problem individualization size-varying touching neurons in optical two-dimensional (2-D) images. Our approach based on...
Abstract To validate a simultaneous analysis tool for the brain and cervical cord embedded in statistical parametric mapping (SPM) framework, we compared trauma‐induced macro‐ microstructural changes spinal injury (SCI) patients to controls. The findings were with results obtained from existing processing tools that assess separately. A probabilistic brain‐spinal template (BSC) was generated using generative semi‐supervised modelling approach. incorporated into pre‐processing pipeline of...
Segmentation of brain magnetic resonance images (MRI) into anatomical regions is a useful task in neuroimaging. Manual annotation time consuming and expensive, so having fully automated general purpose segmentation algorithm highly desirable. To this end, we propose patched-based labell propagation approach based on generative model with latent variables. Once trained, our Factorisation-based Image Labelling (FIL) able to label target variety image contrasts. We compare the effectiveness...
Purpose Inter‐scan motion is a substantial source of error in estimation methods based on multiple volumes, for example, variable flip angle (VFA), and can be expected to increase at 7T where fields are more inhomogeneous. The established correction scheme does not translate since it requires body coil reference. Here we introduce two alternatives that outperform the method. Since they compute relative sensitivities do require images. Theory proposed use coil‐combined magnitude images obtain...
We present open-source tools for 3D analysis of photographs dissected slices human brains, which are routinely acquired in brain banks but seldom used quantitative analysis. Our can: (i) reconstruct a volume from the and, optionally, surface scan; and (ii) produce high-resolution segmentation into 11 regions per hemisphere (22 total), independently slice thickness. can be as substitute ex vivo magnetic resonance imaging (MRI), requires access to an MRI scanner, scanning expertise,...
Accurate labeling of specific layers in the human cerebral cortex is crucial for advancing our understanding neurodevelopmental and neurodegenerative disorders. Leveraging recent advancements ultra-high resolution
This paper presents a framework for automatically learning shape and appearance models medical (and certain other) images. The algorithm was developed with the aim of eventually enabling distributed privacy-preserving analysis brain image data, such that shared information (shape basis functions) may be passed across sites, whereas latent variables encode individual images remain secure within each site. These are proposed as features data mining applications. approach is demonstrated...