AI for Pancreatic Diseases at Northwestern: A Look at the Machine and Hybrid Intelligence Lab
AI Update | Elif Keles, Deputy Director of Augmented Imaging for CISAM and Ulas Bagci, PhD | bagcilab.com
The Machine and Hybrid Intelligence Lab was founded at Northwestern in early 2021 when Dr. Ulas Bagci moved to Northwestern Radiology. His lab creates innovative Artificial Intelligence (AI) solutions for healthcare problems spanning from pancreatic cancer to lung cancer, prostate cancer, liver cirrhosis, infectious and inflammatory lung diseases, and others. Bagci and his team go beyond conventional AI algorithms and try to build a trustworthy bridge between the engineering world and clinical sciences. In this article, we focus on one of the groundbreaking studies that Dr. Bagci, and his senior clinical research associate Dr. Elif Keles are leading titled “AI for Pancreatic Diseases”. Pancreatic cancer is a type of cancer that is often aggressive and difficult to treat. There is currently no widely recommended screening test for pancreatic cancer that has been shown to reduce the mortality rate from this disease effectively. This is because pancreatic cancer often does not produce symptoms until it has progressed to an advanced stage, at which point the cancer may have already spread beyond the pancreas. Some tests may include imaging tests or endoscopic procedures. However, their effectiveness in detecting early-stage pancreatic cancer is not apparent. Early detection and treatment may improve the chances of a better outcome. The team at the Machine and Hybrid Intelligence Lab are working on the early detection of pancreatic cancer from pancreatic cysts. Efforts towards the team’s larger goal have been funded by a NIH R01 titled “Cyst-X: Interpretable Deep Learning Based Risk Stratification of Pancreatic Cystic Tumors” (5R01CA246704-03). This grant, awarded to Northwestern University, is a multi-center study at the national and international level, with collaborations from NYU Langone, Mayo Clinic (Florida, Rochester, Arizona), Erasmus Medical Center- Rotterdam, and Allegheny Health Network. The overall goal of this project is to create a new diagnostic tool called Cyst-X to detect pancreatic cysts from radiologic (MRI) images first, then characterize them to predict their likelihood of being aggressive in the future. Meanwhile, the lab continues collecting multi-modal data from pancreas MRIs and CT scans, pathology, cytology results, and demographic data from our collaborating centers. Already, they have developed a cutting-edge AI algorithm for pancreas MRI analysis and have trained over 500 MRI scans. All pancreas images were classified by the deep learning normal, low-grade, and high- grade cyst cases. Current segmentation and classification results were analyzed to create new evaluation metrics and radiomics in daily clinical routines. Next, the team will further enhance the initially developed AI-based risk stratification model with newly collected data and information to characterize pancreatic cysts better. The Machine and Hybrid Intelligence Lab will act as the AI Core Lab for the IMMINENT and DREAM studies. They are primarily working on making new machine learning algorithms for MRI analysis of the pancreas in both health and disease. As an AI Core Lab, they will create systems that use AI to make predictions for
MRI analysis of the pancreas. Specifically, these algorithms will be able to be explained in the decision space and understood in the model construction space. DREAM (Design and Rationale for the Use of Magnetic Resonance Imaging Biomarkers to Predict Diabetes after Acute Pancreatitis in the Diabetes RElated to Acute Pancreatitis and Its Mechanisms) is another part of the project. In collaboration with the University of Illinois at Chicago and the other PIs of the IMMINENT ((Imaging Morphology of Pancreas in Diabetic Patients following Acute Pancreatitis) study, the team at Northwestern will coordinate the technical and clinical parts of the project. More specifically, they will connect clinical sciences and artificial intelligence for the IMMINENT study's imaging and image analysis. IMMINENT is under the T1DAPC - The Type 1 Diabetes in Acute Pancreatitis Consortium. T1DAPC was granted in 2020 by the National Institute of Diabetes and Digestive and Kidney Diseases (19093//3U01DK127384-02S1). It consists of ten clinical centers and one data coordination center. It aims to investigate Type 1 diabetes and other kinds of diabetes manifesting during or after one or more episodes of acute pancreatitis. To achieve this, T1DAPC's primary research will follow up with patients recently diagnosed with acute pancreatitis to see how many patients develop diabetes. Most recently, the Lab also began work as an AI Center under the CPDPC (Chronic Pancreatitis, Diabetes, and Pancreatic Cancer) Consortium due to their AI efforts for pancreatic diseases,. They will develop new Trustable AI Based Risk Assessment for Patients with Chronic Pancreatitis PROspective Evaluation of Chronic Pancreatitis for EpidEmiologic and Translational StuDies (PROCEED): Rationale and Study Design From the Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer.
The Read | Volume 6 | Page 6
Department of Radiology
Made with FlippingBook - Online magazine maker