May 2025: We are very excited to share our new article published in Scientific Reports. This paper discussed an AI model developed by our group for a fully-automated morphologically normal volumetric pancreas segmentation on portal venous phase CTs. The AI-based pancreas segmentation is a widely studied problem. However since AI is data-centric, we have trained an AI model that used the largest reported dataset of 3031 CTs and corresponding reference segmentations that were meticulously curated by our expert radiologists. As expected, the model showed the highest reported volumetric overlap or dice score of 0.94 on an internal test subset (n = 452). Moreover, the generalizability of the model was further validated on a multi-institutional public dataset (n = 585) where the model achieved a dice of 0.96. This model paves our way for creating a fully-automated and highly accurate CT-based pancreas segmentation model that alleviates the need of manual intervention of the radiologists while segmenting pancreas which is subject to an unavoidable inter/intra-reader variability. Thanks to our Radiologist colleagues for the dataset and time worth spent for the curation.
Here is the link to the paper
https://pmc.ncbi.nlm.nih.gov/articles/PMC12084540/
February 2025: Presented my works on AI-based early pancreatic cancer detection and fully automated pancreatic ductal adenocarcinoma (PDAC) segmentation at the Society of Abdominal Radiology (SAR) 2025 annual meeting held in Tucson, Arizona.
December 2024: Our two new recent review articles highlighted recent progresses made in the pancreatic AI-research domain starting from early pancreatic cancer detection to prognostication. Early pancreatic cancer detection is especially a challenging problem and we have initially shown the prospect of a radiomics-based machine learning model (REDMOD) based on pre-diagnostic CTs (CTs acquired 3-36 months before the actual clinical diagnosis) to detect early pancreatic ductal adenocarcinoma (PDA), a common form of pancreatic cancer. We have also discussed other seminal works related to early detection. Additional applications of AI such as volumetric pancreas and PDA segmentations as well as survival analysis were also explored.
https://link.springer.com/article/10.1007/s00261-024-04775-x
https://www.liebertpub.com/doi/abs/10.1089/aipo.2024.0032
December 2024: Presenting my work on AI-based early pancreatic cancer detection at American Pancreatic Association (APA) Annual Meeting in Maui, Hawaii.
October 2024: Our two abstracts received oral presentation at Society of Abdominal Radiology (SAR) Annual Meeting to be held in Tucson, AZ in Feb 2025.
August 2024: Our recent review article was published in AI in Precision Oncology. This article showcases some of the recent progresses made in the AI-domain for pancreatic cancer diagnosis and management (https://www.liebertpub.com/doi/abs/10.1089/aipo.2024.0032) .
April 2024: Received a pilot research grant from Mayo Clinic Comprehensive Cancer Center. This will help kick-off the AI-based prognostication project for pancreatic cancer patients.
January, 2024: Glad to share our another article on the assessment of the ML model for early pancreatic cancer detection against various simulated perturbations added to the predictive radiomic features (https://pubmed.ncbi.nlm.nih.gov/38175255/ ).
December 2023: Presented my AI-based fully-automated pancreatic tumor segmentation work at Radiological Society of North America Annual Meeting, Chicago.
August 2023: Glad to share our recent AI article that targeted a nnUNet-based deep learning algorithm for pancreatic tumor segmentation. The latter is a challenging problem while we showed that the segmentation performance could be significantly improved if the tumors are tightly localized first using user-defined bounding boxes. More works need to be done for generating a fully automated version (https://www.sciencedirect.com/science/article/pii/S142439032300145X?via%3Dihub) .
December 2022: Presented my AI-based semi-automated pancreatic tumor segmentation work at Radiological Society of North America Annual Meeting, Chicago.
November, 2022: Our another ML paper (co-first author) is now online. This proof-of-concept study shows a positive signal that type 2 diabetes can be detected through CT Radiomics-based ML model. (https://link.springer.com/article/10.1007/s00261-022-03668-1)
September, 2022: Received an academic rank of Assistant Professor of Radiology by Mayo Clinic School of Medicine
August 2022: My Gastroenterology paper received multiple encouraging editorials and news mentions!!
Editorials
News Mentions
https://theimagingwire.com/2022/07/17/the-case-for-pancreatic-cancer-radiomics/
https://medicalxpress.com/news/2022-07-ai-prediagnostic-cts-pancreatic-cancer.html
July, 2022: My recent machine learning (ML) paper published in Gastroenterology (IF: 33.88) shows the potential of radiomics-based ML model for detecting early pancreatic cancer. (https://www.sciencedirect.com/science/article/pii/S0016508522007284?via%3Dihub)
July 2022: Presented our work at SIIM-CMIMI at Baltimore, MD.
March, 2022: Promoted to Senior Scientist
December 2021: Presented my AI-based early pancreatic cancer detection work at Radiological Society of North America Annual Meeting, Chicago.
February , 2021: Joined Mayo Clinic, Rochester, MN as Medical Imaging Scientist