Research

This page collects my current research publications and preprints across computer vision, medical imaging, multimodal learning, and clinically oriented deep learning. The work spans chest X-ray classification, long-tailed recognition, lung CT segmentation, histopathology image analysis, echocardiography, retinal OCT, and responsible AI topics such as LLM privacy.

Featured research pages include CXR-LT, TARU-Net, CIPS-Net, ECG-Free Echo, and Lung Digital Twin.

Publications

† Equal contribution

Ongoing Research Projects

  • HuMAR - Working on developing efficient and scalable text-instructed vision-language model for multimodal and multitasking Human Centric detection.
  • Novel Segmentation and Denoising Architectures - Working on developing of novel self supervised models for multitasking and rigorous experiments on various OCT datasets.
  • Bone Cancer Detection - Generaly Whole Genome Sequencing is the gold standard for detection costing $6k per patient, so we are working on a method to detect the bone cancer from the H&E whole slide images.
  • Whole Genome Doubling - WGD is one of the somatic events of cancer, detecting it takes very long time using sequencing, hence we are developing a methdology on detecting the WGD from the H&E WSI's.

Services and Contributions

Research Competitions