5 papers
Input-Dependent Fisher Information for Local Sensitivity Analysis of Medical Image Classifiers
Sourya Sengupta. Mark A. Anastasio
Deep neural networks have achieved strong performance in medical image classification, but often work like black-box. Commonly used post-hoc interpretation methods often provide he…
Observer-Usable Information as a Task-specific Image Quality Metric
Changjie Lu, Sourya Sengupta, Hua Li +1
Objective, task-based measures of image quality (IQ) have been widely advocated for assessing and optimizing medical imaging technologies. Besides signal detection theory-based mea…
Synthetic Volumetric Data Generation Enables Zero-Shot Generalization of Foundation Models in 3D Medical Image Segmentation
Satrajit Chakrabarty, Sourya Sengupta, Gopal Avinash +1
Foundation models such as Segment Anything Model 2 (SAM 2) exhibit strong generalization on natural images and videos but perform poorly on medical data due to differences in appea…
On the Utility of Virtual Staining for Downstream Applications as it relates to Task Network Capacity
Sourya Sengupta, Jianquan Xu, Phuong Nguyen +3
Virtual staining, or in-silico-labeling, has been proposed to computationally generate synthetic fluorescence images from label-free images by use of deep learning-based image-to-i…
SynthFM: Training Modality-agnostic Foundation Models for Medical Image Segmentation without Real Medical Data
Sourya Sengupta, Satrajit Chakrabarty, Keerthi Sravan Ravi +2
Foundation models like the Segment Anything Model (SAM) excel in zero-shot segmentation for natural images but struggle with medical image segmentation due to differences in textur…