Pancreas
The pancreas plays crucial roles in digestion and blood sugar regulation. It produces digestive enzymes and hormones, like insulin and glucagon. Proper pancreas function is essential for overall health and managing conditions like diabetes.
- Home
- Research
- Pancreas

Research in Biomedical Imaging Analysis
Accurate Pancreas Segmentation

Dynamic Linear Transformer for 3D Biomedical Image Segmentation
Accurate pancreas segmentation is important for precise diagnostics, treatment planning, and research. It allows for better identification of abnormalities, such as tumors or cysts. Improved accuracy can lead to more effective treatments and better patient outcomes. Furthermore, it contributes to the development of advanced techniques in medical imaging and artificial intelligence. In this work, we proposed one "Dynamic Linear Transformer" structure to accurately segment pancreas.
Precise Pancreas Diagnoisis

Radiomics Boosts Deep Learning Model for IPMN Classification
Intraductal Papillary Mucinous Neoplasm (IPMN) cysts are premalignant pancreas lesions, and they can progress into pancreatic cancer. Therefore, detecting and stratifying their risk level is of ultimate importance for effective treatment planning and disease control. However, this is a highly challenging task because of the diverse and irregular shape, texture, and size of the IPMN cysts as well as the pancreas. In this study, we proposed a novel computer-aided diagnosis pipeline for IPMN risk classification from multi-contrast MRI scans. Paper and code links to be published.
Interpretable Systems

Robust Visual Explanation for Radiology AI Applications with Information Bottleneck
(RSNA 2022)
Visual explanation methods provide insight into the decision-making process for neural networks. Determining the
impacts of each pixel in the input on the classifier decision has huge importance to understand the behavior of the
classifiers. Most of the gradient-based attribution methods produce results that are independent of model performance
which cannot be used for reliable explanation. In this study, we analyzed the Information Bottleneck Attribution
approach on medical images and showed that it outperforms the Grad-CAM.
Funding Resources
NIH/NCI #R01 CA246704-01
Cyst-X: Interpretable Deep Learning Based Risk Stratification of Pancreatic Cystic Tumors
The goal of this effort is to develop novel interpretable AI methods to determine the risk status of pre-cancerous pancreatic cysts at early stages using radiology screening (MRI).