Assistant Professor (Tenure-Track), Department of Computer Science, University of South Dakota |
IEEE Senior Member | Top 2% Scientist in AI & Biomedical Engineering (Stanford–Elsevier Ranking)
I develop human-centered, trustworthy AI systems that learn to see, interpret, and reason like clinicians, with the goal of making expert-level diagnosis faster, fairer, and universally accessible. As an Assistant Professor of Computer Science at the University of South Dakota and Senior AI Scientist at Ther-AI LLC, I lead pioneering research at the intersection of computer vision, biomedical informatics, and clinical practice. My work spans multimodal AI integration, large-scale medical dataset curation, and the development of energy-efficient models deployable on low-end devices—pushing the boundaries of real-world impact in healthcare. My journey began with a deep commitment to solving medical problems through algorithmic intelligence. Trained in computer science and biomedical engineering, I have led innovations across gastrointestinal imaging (esophagus, IBD, ERCP, colonoscopy, and video capsule endoscopy), thoracic and abdominal radiology (liver, lung, pancreas, prostate), and radiation therapy planning. To overcome the longstanding bottleneck of data scarcity in endoscopy, I spearheaded the creation of some of the world’s largest and most-used public datasets—HyperKvasir, KvasirCapsule, and PolypDB—now cited and benchmarked globally. My segmentation models, including ColonSegNet, ResUNet++, and DoubleUNet, are widely adopted in academia and industry. ColonSegNet has been integrated into NVIDIA Clara Holoscan, a milestone demonstrating the clinical scalability of my work. As a PI, I lead multi-institutional collaborations and NIH/NSF-aligned initiatives aimed at building explainable AI tools that radiologists can trust—blending transformers, Mamba architectures, language-vision models, and human-in-the-loop AI into high-performance diagnostic systems. With over 125 peer-reviewed publications and 7,800+ citations, my research has shaped the global discourse on AI for medical imaging. Recognized as a Top 2% Scientist Worldwide (Stanford/Elsevier), I’ve received multiple honors including the Meta Paper with Code Award, IEEE Distinguished R&D Award, and Best Paper Awards at IEEE and IAPR. I actively mentor students, lead open science efforts, and organize international medical AI workshops to democratize innovation. Driven by a vision of equitable, AI-augmented healthcare, I continue to design technologies that not only diagnose but also empower clinicians, reduce the burden, and save lives.
At the heart of our lab is a singular mission:
To build intelligent systems that can see, reason, and learn like clinicians—advancing medicine for every patient, everywhere. We believe that the next breakthroughs in healthcare will come not from replacing doctors, but from empowering them with AI that is transparent, trustworthy, and human-centered.
Our work sits at the cutting edge of deep learning, computer vision, language modeling, and multimodal intelligence, with a relentless focus on solving real-world problems in medical imaging, diagnostics, and decision support.
From GI endoscopy and oncology to radiology and radiation therapy, we are building the tools that redefine what’s possible in clinical AI.
The future of healthcare is intelligent.
We’re building it—one dataset, one model, and one breakthrough at a time.
I teach a range of undergraduate and graduate courses in computer science, with a strong focus on data science, AI, and systems design. Below are the core subjects I’ve led or actively teach:
Email: debeshjha@gmail.com