A5094139993- Radiomics and Machine Learning in Medical Imaging
- Radiation Dose and Imaging
- Advanced X-ray and CT Imaging
- Lung Cancer Diagnosis and Treatment
- Artificial Intelligence in Healthcare and Education
- Cardiac Imaging and Diagnostics
- Medical Imaging Techniques and Applications
- Topic Modeling
- Radiology practices and education
- COVID-19 diagnosis using AI
Massachusetts General Hospital
2024-2025
Harvard University
2024-2025
Lung cancer screening (LCS) reduces mortality and involves vast multimodal data such as text, tables, images. Fully mining big requires multitasking; otherwise, occult but important features may be overlooked, adversely affecting clinical management healthcare quality. Here we propose a medical multimodal-multitask foundation model (M3FM) for three-dimensional low-dose computed tomography (CT) LCS. After curating multitask dataset of 49 types, 163,725 chest CT series, 17 tasks involved in...
Abstract We created and validated an open-access AI algorithm (AIc) for assessing image segmentation patient centering in a multi-body-region, multi-center, multi-scanner study. Our study included 825 head, chest, abdomen-pelvis CT from 275 patients (153 females, 128 males; mean age 67 ± 14 years) scanned at five academic community hospitals. images were processed with the AIc to determine vertical horizontal skull base (head CT), carina (chest L2-L3 disc (abdomen CT). manually measured...
Abstract Importance Automatic generation of the impression section radiology report can help make radiologists efficient and avoid reporting errors. Objective To evaluate relationship, content, accuracy an Powerscribe Smart Impression (PSI) against radiologists’ reported findings (RDF). Design, Setting, Participants The institutional review board approved retrospective study developed trained PSI algorithm (Nuance Communications, Inc.) with 9.8 million reports from multiple sites to generate...
Structured radiology reporting is advantageous for optimizing clinical workflows and patient outcomes. Current LLMs in creating structured reports face the challenges of formatting errors, content hallucinations, privacy leakage concerns when uploaded to external servers. We aim develop an enhanced open-source LLM standardized LCS from free-text descriptions. After institutional IRB approvals, 5,442 de-identified two institutions were retrospectively analyzed. 500 randomly selected evenly...