A5054888342- Medical Image Segmentation Techniques
- Advanced Neuroimaging Techniques and Applications
- Advanced MRI Techniques and Applications
- Functional Brain Connectivity Studies
- Advanced Neural Network Applications
- AI in cancer detection
- Medical Imaging Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Medical Imaging and Analysis
- Fetal and Pediatric Neurological Disorders
- Alzheimer's disease research and treatments
- Domain Adaptation and Few-Shot Learning
- Crystal Structures and Properties
- Anatomy and Medical Technology
- Dementia and Cognitive Impairment Research
- Brain Tumor Detection and Classification
- Microwave Dielectric Ceramics Synthesis
- Chemical Synthesis and Characterization
- Advanced Image and Video Retrieval Techniques
- Cell Image Analysis Techniques
- MRI in cancer diagnosis
- Epilepsy research and treatment
- Neonatal and fetal brain pathology
- X-ray Diffraction in Crystallography
- Image Retrieval and Classification Techniques
Massachusetts General Hospital
2015-2025
University College London
2016-2025
Athinoula A. Martinos Center for Biomedical Imaging
2015-2025
Harvard University
2014-2025
Massachusetts Institute of Technology
2019-2025
Yale University
2025
Geneva College
2024
Harvard University Press
2024
University of Maryland, College Park
2023
Centro Científico Tecnológico - Santa Fe
2023
Automated analysis of MRI data the subregions hippocampus requires computational atlases built at a higher resolution than those that are typically used in current neuroimaging studies. Here we describe construction statistical atlas hippocampal formation subregion level using ultra-high resolution, ex vivo MRI. Fifteen autopsy samples were scanned 0.13 mm isotropic (on average) customized hardware. The images manually segmented into 13 different substructures protocol specifically designed...
Automatic whole-brain extraction from magnetic resonance images (MRI), also known as skull stripping, is a key component in most neuroimage pipelines. As the first element chain, its robustness critical for overall performance of system. Many stripping methods have been proposed, but problem not considered to be completely solved yet. systems literature good on certain datasets (mostly they were trained/tuned on), fail produce satisfactory results when acquisition conditions or study...
The human thalamus is a brain structure that comprises numerous, highly specific nuclei. Since these nuclei are known to have different functions and be connected areas of the cerebral cortex, it great interest for neuroimaging community study their volume, shape connectivity in vivo with MRI. In this study, we present probabilistic atlas thalamic built using ex MRI scans histological data, as well application segmentation. was manual delineation 26 on serial histology 12 whole thalami from...
In this paper we present a method to segment four brainstem structures (midbrain, pons, medulla oblongata and superior cerebellar peduncle) from 3D brain MRI scans. The segmentation relies on probabilistic atlas of the its neighboring structures. To build atlas, combined dataset 39 scans with already existing manual delineations whole 10 in which were manually labeled protocol that was specifically designed for study. resulting can be used Bayesian framework novel Thanks generative nature...
Despite advances in data augmentation and transfer learning, convolutional neural networks (CNNs) difficultly generalise to unseen domains. When segmenting brain scans, CNNs are highly sensitive changes resolution contrast: even within the same MRI modality, performance can decrease across datasets. Here we introduce SynthSeg, first segmentation CNN robust against contrast resolution. SynthSeg is trained with synthetic sampled from a generative model conditioned on segmentations. Crucially,...
Quantitative analysis of magnetic resonance imaging (MRI) scans the brain requires accurate automated segmentation anatomical structures. A desirable feature for such methods is to be robust against changes in acquisition platform and protocol. In this paper we validate performance a algorithm designed meet these requirements, building upon generative parametric models previously used tissue classification. The method tested on four different datasets acquired with scanners, field strengths...
Despite the crucial role of hypothalamus in regulation human body, neuroimaging studies this structure and its nuclei are scarce. Such scarcity partially stems from lack automated segmentation tools, since manual delineation suffers scalability reproducibility issues. Due to small size image contrast vicinity, is difficult has been long neglected by widespread packages like FreeSurfer or FSL. Nonetheless, recent advances deep machine learning enabling us tackle problems with high accuracy....
The development of automated tools for brain morphometric analysis in infants has lagged significantly behind analogous adults. This gap reflects the greater challenges this domain due to: 1) a smaller-scaled region interest, 2) increased motion corruption, 3) regional changes geometry to heterochronous growth, and 4) variations contrast properties corresponding ongoing myelination other maturation processes. Nevertheless, there is great need image-processing quantify differences between...
We introduce a strategy for learning image registration without acquired imaging data, producing powerful networks agnostic to contrast introduced by magnetic resonance (MRI). While classical methods accurately estimate the spatial correspondence between images, they solve an optimization problem every new pair. Learning-based techniques are fast at test time but limited registering images with contrasts and geometric content similar those seen during training. propose remove this dependency...
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...
Every year, millions of brain MRI scans are acquired in hospitals, which is a figure considerably larger than the size any research dataset. Therefore, ability to analyze such could transform neuroimaging research. Yet, their potential remains untapped since no automated algorithm robust enough cope with high variability clinical acquisitions (MR contrasts, resolutions, orientations, artifacts, and subject populations). Here, we present SynthSeg + , an AI segmentation suite that enables...
Reconstructing 3D MR volumes from multiple motion-corrupted stacks of 2D slices has shown promise in imaging moving subjects, e. g., fetal MRI. However, existing slice-to-volume reconstruction methods are time-consuming, especially when a high-resolution volume is desired. Moreover, they still vulnerable to severe subject motion and image artifacts present acquired slices. In this work, we NeSVoR, resolution-agnostic method, which models the underlying as continuous function spatial...
Thermal decomposition of boehmite and bayerite has been studied by X-ray diffraction NMR IR spectroscopies. Coordination site distortion Al polyhedra have estimated 27Al spectroscopy while optical properties transitional aluminas, η-, γ-, δ-, θ-Al2O3 were derived from near-normal specular reflectance technique. In the low-temperature phases, aluminum vacancies are located in tetrahedral positions γ-Al2O3 η-Al2O3 they distributed at random between octahedral positions. addition, a small...
The activation energy involved in the motion of Li+ ions along conduction channels NASICON framework has been determined from electrical conductivity measurements samples composition LiM2(PO4)3 and LiMM'(PO4)3, where M M' are Ge, Ti, Sn, Hf, all compounds belonging to space group R3̄c. Two lithium sites, M1 M2, inside channels, can be distinguished. sites connected through triangular bottlenecks oxygen atoms, size bottleneck estimated refined simulated structures for each composition. plot...
The hippocampal formation is a complex, heterogeneous structure that consists of number distinct, interacting subregions. Atrophy these subregions implied in variety neurodegenerative diseases, most prominently Alzheimer's disease (AD). Thanks to the increasing resolution MR images and computational atlases, automatic segmentation becoming feasible MRI scans. Here we introduce generative model for dedicated longitudinal relies on subject-specific atlases. segmentations scans at different...
Volume deficits of the hippocampus in schizophrenia have been consistently reported. However, is anatomically heterogeneous; it remains unclear whether certain portions are affected more than others schizophrenia. In this study, we aimed to determine volume confined specific subfields and measure subfield trajectories over course illness. Magnetic resonance imaging scans were obtained from Data set 1: 155 patients with (mean duration illness 7 years) 79 healthy controls, 2: an independent...
Subcortical structures play a critical role in brain function. However, options for assessing electrophysiological activity these are limited. Electromagnetic fields generated by neuronal subcortical can be recorded non-invasively using magnetoencephalography (MEG) and electroencephalography (EEG). signals much weaker than those due to cortical activity. In addition, we show here that it is difficult resolve sources, because distributed explain the MEG EEG patterns deep sources. We then...