A5066569995- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Blind Source Separation Techniques
- Neural dynamics and brain function
- Digital Imaging for Blood Diseases
- Fault Detection and Control Systems
- Biomedical Text Mining and Ontologies
- Colorectal Cancer Screening and Detection
- Cell Image Analysis Techniques
- Generative Adversarial Networks and Image Synthesis
- Anesthesia and Sedative Agents
- Speech and Audio Processing
- EEG and Brain-Computer Interfaces
- Gene expression and cancer classification
- Control Systems and Identification
- Computational Drug Discovery Methods
- Anesthesia and Neurotoxicity Research
- Hearing Loss and Rehabilitation
- Sparse and Compressive Sensing Techniques
- Advanced Adaptive Filtering Techniques
- Music and Audio Processing
- Gaussian Processes and Bayesian Inference
- Modular Robots and Swarm Intelligence
- Medical Imaging and Analysis
- Artificial Intelligence in Healthcare and Education
Brigham and Women's Hospital
2022-2025
Harvard University
2021-2024
Massachusetts General Hospital
2021-2024
Broad Institute
2022-2024
Dana-Farber Cancer Institute
2022-2024
Mass General Brigham
2024
Columbia University
2024
Health Education North West
2023
NHS England
2023
Massachusetts Institute of Technology
2015-2022
Human tissue, which is inherently three-dimensional (3D), traditionally examined through standard-of-care histopathology as limited two-dimensional (2D) cross-sections that can insufficiently represent the tissue due to sampling bias. To holistically characterize histomorphology, 3D imaging modalities have been developed, but clinical translation hampered by complex manual evaluation and lack of computational platforms distill insights from large, high-resolution datasets. We present...
Supervised deep learning has gained significant attention for speech enhancement recently. The state-of-the-art methods perform the task by a ratio/binary mask that is applied to mixture in time-frequency domain produce clean speech. Despite great performance single-channel setting, these frameworks lag multichannel setting as majority of a) fail exploit available spatial information fully, and b) still treat architecture black box which may not be well-suited audio processing. This paper...
State-dependent activity of locus ceruleus (LC) neurons has long suggested a role for noradrenergic modulation arousal. However, <i>in vivo</i> insights into arousal circuitry have been constrained by the fundamental inaccessibility human brain invasive studies. Functional magnetic resonance imaging (fMRI) studies performed during site-specific pharmacological manipulations levels may be used to study circuitry. Dexmedetomidine is an anesthetic that alters level selectively targeting α2...
Foundation models are reshaping computational pathology by enabling transfer learning, where pre-trained on vast datasets can be adapted for downstream diagnostic, prognostic, and therapeutic response tasks. Despite these advances, foundation still limited in their ability to encode the entire gigapixel whole-slide images without additional training often lack complementary multimodal data. Here, we introduce Threads, a slide-level model capable of generating universal representations any...
Sparse Bayesian learning (SBL) is a powerful framework for tackling the sparse coding problem while also providing uncertainty quantification. The most popular inference algorithms SBL exhibit prohibitively large computational costs high-dimensional problems due to need maintain covariance matrix. To resolve this issue, we introduce new method accelerating – named covariance-free expectation maximization (CoFEM) that avoids explicit computation of CoFEM solves multiple linear systems...
Advances in digitizing tissue slides and the fast-paced progress artificial intelligence, including deep learning, have boosted field of computational pathology. This holds tremendous potential to automate clinical diagnosis, predict patient prognosis response therapy, discover new morphological biomarkers from images. Some these intelligence-based systems are now getting approved assist diagnosis; however, technical barriers remain for their widespread adoption integration as a research...
Electroencephalography (EEG)-based motor imagery (MI) has potential applications in diverse fields including rehabilitation, drone control, and virtual reality. However, its practical use is hindered by low generalization performance decoding brain signals, primarily due to the subject-dependency of EEG signals. Although multitask autoencoder (MTAE) techniques have recently been used mitigate this issue, these approaches encounter an imbalance problem between loss functions with different...
Spatial transcriptomics (ST) enables interrogating the molecular composition of tissue with ever-increasing resolution, depth, and sensitivity. However, costs, rapidly evolving technology, lack standards have constrained computational methods in ST to narrow tasks small cohorts. In addition, underlying morphology as reflected by H&E-stained whole slide images (WSIs) encodes rich information often overlooked studies. Here, we introduce HEST-1k, a collection 1,108 spatial transcriptomic...
Abstract Early identification of drug toxicity is essential yet challenging in development. At the preclinical stage, assessed with histopathological examination tissue sections from animal models to detect morphological lesions. To complement this analysis, toxicogenomics increasingly employed understand mechanism action compound and ultimately identify lesion-specific safety biomarkers for which vitro assays can be designed. However, existing works that aim correlates expression changes...
We introduce FCUBE, a cloud-based framework that enables machine learning researchers to contribute their learners its community-shared repository. FCUBE exploits data parallelism in lieu of algorithmic parallelization allow users efficiently tackle large problems automatically. It passes random subsets generated via resampling multiple it executes simultaneously and then combines model predictions with simple fusion technique. is an example what we have named Bring Your Own Learner model....
Multimodal survival methods combining gigapixel histology whole-slide images (WSIs) and transcriptomic profiles are particularly promising for patient prognostication stratification. Current approaches involve tokenizing the WSIs into smaller patches (>10,000 patches) transcriptomics gene groups, which then integrated using a Transformer predicting outcomes. However, this process generates many tokens, leads to high memory requirements computing attention complicates post-hoc...
In drug development, assessing the toxicity of candidate compounds is crucial for successfully transitioning from preclinical research to early-stage clinical trials. Drug safety typically assessed using animal models with a manual histopathological examination tissue sections characterize dose-response relationship compound - time-intensive process prone inter-observer variability and predominantly involving tedious review cases without abnormalities. Artificial intelligence (AI) methods in...