A5103062314- Privacy-Preserving Technologies in Data
- Radiomics and Machine Learning in Medical Imaging
- Artificial Intelligence in Healthcare and Education
- vaccines and immunoinformatics approaches
- Advanced Neural Network Applications
- Genetics, Bioinformatics, and Biomedical Research
- AI in cancer detection
- Glycosylation and Glycoproteins Research
- Glioma Diagnosis and Treatment
- Motor Control and Adaptation
- Machine Learning in Healthcare
- Tactile and Sensory Interactions
- Cancer Genomics and Diagnostics
- Brain Tumor Detection and Classification
- Heart Failure Treatment and Management
- Cardiac Imaging and Diagnostics
- COVID-19 diagnosis using AI
- Minimally Invasive Surgical Techniques
- Cardiovascular Effects of Exercise
- Cardiovascular Issues in Pregnancy
- Medical Imaging and Analysis
- Action Observation and Synchronization
- Domain Adaptation and Few-Shot Learning
- Rocket and propulsion systems research
- Calcium Carbonate Crystallization and Inhibition
Intel (United States)
2019-2023
Mission College
2020
Naval Medical Center San Diego
2013
University of Milan
1995-2007
University of Pittsburgh
2004
Washington University in St. Louis
2004
Neurosciences Institute
2001-2003
John Jay College of Criminal Justice
2003
Several studies underscore the potential of deep learning in identifying complex patterns, leading to diagnostic and prognostic biomarkers. Identifying sufficiently large diverse datasets, required for training, is a significant challenge medicine can rarely be found individual institutions. Multi-institutional collaborations based on centrally-shared patient data face privacy ownership challenges. Federated novel paradigm data-private multi-institutional collaborations, where model-learning...
Although machine learning (ML) has shown promise in numerous domains, there are concerns about generalizability to out-of-sample data. This is currently addressed by centrally sharing ample, and importantly diverse, data from multiple sites. However, such centralization challenging scale (or even not feasible) due various limitations. Federated ML (FL) provides an alternative train accurate generalizable models, only numerical model updates. Here we present findings the largest FL study...
Abstract Medical artificial intelligence (AI) has tremendous potential to advance healthcare by supporting and contributing the evidence-based practice of medicine, personalizing patient treatment, reducing costs, improving both provider experience. Unlocking this requires systematic, quantitative evaluation performance medical AI models on large-scale, heterogeneous data capturing diverse populations. Here, meet need, we introduce MedPerf, an open platform for benchmarking in domain....
Single-unit activity in area M1 was recorded awake, behaving monkeys during a three-dimensional (3D) reaching task performed virtual reality environment. This study compares motor cortical discharge rate to both the hand's velocity and arm's joint angular velocities. Hand is considered parameter of extrinsic space because it measured Cartesian coordinate system monkey's workspace. Joint intrinsic relative adjacent arm/body segments. In initial analysis, as difference hand position or posture...
A motor illusion was created to separate human subjects' perception of arm movement from their actual during figure drawing. Trajectories constructed cortical activity recorded in monkeys performing the same task showed that represented primary cortex, whereas visualized, presumably perceived, trajectories were found ventral premotor cortex. Perception and action representations can be differentially recognized brain may contained structures.
This manuscript describes the first challenge on Federated Learning, namely Tumor Segmentation (FeTS) 2021. International challenges have become standard for validation of biomedical image analysis methods. However, actual performance participating (even winning) algorithms "real-world" clinical data often remains unclear, as included in are usually acquired very controlled settings at few institutions. The seemingly obvious solution just collecting increasingly more from institutions such...
Convolutional neural network (CNN) models perform state of the art performance on image classification, localization, and segmentation tasks. Limitations in computer hardware, most notably small memory size deep learning accelerator cards, prevent relatively large images, such as those from medical satellite imaging, being processed a whole their original resolution. A fully convolutional topology, U-Net, is typically trained down-sampled images inference resolution, by simply dividing...
Abstract Objective. De-centralized data analysis becomes an increasingly preferred option in the healthcare domain, as it alleviates need for sharing primary patient across collaborating institutions. This highlights consistent harmonized curation, pre-processing, and identification of regions interest based on uniform criteria. Approach. Towards this end, manuscript describes Fe derated T umor S egmentation (FeTS) tool, terms software architecture functionality. Main results. The aim FeTS...
Abstract Background Statins represent a modern mainstay of the drug treatment coronary artery disease and acute syndromes. Reduced aerobic work performance slowed VO 2 kinetics are established features clinical picture post‐myocardial infarction (MI) patients. We tested hypothesis that statin therapy improves exercise in normocholesterolaemic post‐MI Materials methods According to double‐blinded, randomized, crossover placebo‐controlled study design, 18 patients with uncomplicated recent (3...
Using medical imaging as case-study, we demonstrate how Intel-optimized TensorFlow on an x86-based server equipped with 2nd Generation Intel Xeon Scalable Processors large system memory allows for the training of memory-intensive AI/deep-learning models in a scale-up configuration. We believe our work represents first deep neural network having footprint (~ 1 TB) single-node server. recommend this configuration to scientists and researchers who wish develop large, state-of-the-art AI but are...
Medical AI has tremendous potential to advance healthcare by supporting the evidence-based practice of medicine, personalizing patient treatment, reducing costs, and improving provider experience. We argue that unlocking this requires a systematic way measure performance medical models on large-scale heterogeneous data. To meet need, we are building MedPerf, an open framework for benchmarking machine learning in domain. MedPerf will enable federated evaluation which securely distributed...
Deep learning models for semantic segmentation of images require large amounts data. In the medical imaging domain, acquiring sufficient data is a significant challenge. Labeling image requires expert knowledge. Collaboration between institutions could address this challenge, but sharing to centralized location faces various legal, privacy, technical, and data-ownership challenges, especially among international institutions. study, we introduce first use federated multi-institutional...
Abstract BACKGROUND Training deep learning algorithms requires large amounts of data, which is a significant challenge in the medical domain, and particularly neuro-oncology, where ample data can only be found multi-institutional collaborations. The current paradigm for collaborations based on pooled datasets that has always faced privacy, legal, technical, data-ownership concerns. In this study we evaluate hypothesis federated provide method to overcome these concerns facilitate shift...