Deep learning analysis of MRI accurately detects early-stage perihilar cholangiocarcinoma in patients with primary sclerosing cholangitis.
Authors
Affiliations (13)
Affiliations (13)
- Department of Radiology, Mayo Clinic, Rochester, Minnesota, USA.
- Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, Minnesota, USA.
- Department of Biostatistics, Mayo Clinic, Rochester, Minnesota, USA.
- Department of Clinical Science, Haraldsplass Deaconess Hospital, University of Bergen, Bergen, Norway.
- Department of Transplantation Medicine, Division of Surgery, Inflammatory Medicine and Transplantation, Norwegian PSC Research Center, Oslo University, Oslo, Norway.
- Department of Gastroenterology, Akershus University Hospital, Lørenskog, Norway.
- Institute of Clinical Medicine, University of Oslo, Oslo, Norway.
- Department of Radiology, Haukeland University Hospital, Bergen, Norway.
- Department of Diagnostic Imaging, Akershus University Hospital, Lørenskog, Norway.
- Department of Radiology, Oslo University Hospital, Oslo, Norway.
- Toronto Centre for Liver Disease, University Health Network and Department of Medicine, University of Toronto, Toronto, Ontario, Canada.
- Toronto Centre for Liver Disease, University Health Network and Department of Radiology, University of Toronto, Toronto, Ontario, Canada.
- Division of Transplantation Surgery, Mayo Clinic, Rochester, Minnesota, USA.
Abstract
Among those with primary sclerosing cholangitis (PSC), perihilar cholangiocarcinoma (pCCA) is often diagnosed at a late stage and is a leading source of mortality. Detection of pCCA in PSC when curative action can be taken is challenging. Our aim was to create a deep learning model that analyzed MRI to detect early-stage pCCA and compare its diagnostic performance with expert radiologists. We conducted a multicenter, international, retrospective cohort study involving adults with large duct PSC who underwent contrast-enhanced MRI. Senior abdominal radiologists reviewed the images. All patients with pCCA had early-stage cancer and were registered for liver transplantation. We trained a 3D DenseNet-121 model, a form of deep learning, using MRI images and assessed its performance in a separate test cohort. The study included 398 patients (training cohort n=150; test cohort n=248). pCCA was present in 230 individuals (training cohort n=64; test cohort n=166). In the test cohort, the respective performances of the model compared to the radiologists were: sensitivity 87.9% versus 50.0%, p <0.001; specificity 84.1% versus 100.0%, p <0.001; area under receiving operating curve 86.0% versus 75.0%, p <0.001. Even when a mass was absent, the model had a higher sensitivity for pCCA than radiologists (91.6% vs. 50.6%, p <0.001) and maintained good specificity (84.1%). The 3D DenseNet-121 MRI model effectively detects early-stage pCCA in PSC patients. Compared to expert radiologists, the model missed fewer cases of cancer.