1. McGraw, Mark

Article Content

Researchers at Cedars-Sinai Medical Center have developed an artificial intelligence (AI)-driven tool that they believe can accurately predict which patients will develop pancreatic cancer based on what CT scan images revealed years before diagnosis. The investigators shared findings from their study using the tool in the journal Cancer Biomarkers, where they outlined their conclusions and how the tool might help prevent deaths related to one of the most difficult-to-treat cancers (2022; doi: 10.3233/CBM-210273).

Pancreatic Cancer. P... - Click to enlarge in new windowPancreatic Cancer. Pancreatic Cancer

Early-stage diagnosis of pancreatic ductal adenocarcinoma (PDAC) is "challenging due to the lack of specific diagnostic biomarkers," the authors wrote. "However, stratifying individuals at high risk of PDAC, followed by monitoring their health conditions on a regular basis, has the potential to allow diagnosis at early stages."


Describing AI as "the primary choice to perform prediction modeling for several cancers," the authors noted that AI-based systems "almost always surpass traditional manual approaches." Advances in AI "offer an enormous range of automated tools for extracting precise measurements of biomarkers and organs, unveiling complex features and quantifying tissue characteristics," the researchers wrote, adding that radiomic analysis, machine learning, and deep learning "are considered the most reliable and frequently utilized AI approaches for prediction modeling."


And, while imaging has not been explicitly used to predict pancreatic cancer, recent advancements in image acquisition devices, imaging processing techniques, and image analysis tools can help make imaging "the foundation for a sophisticated prediction system for PDAC."


Abdominal pain "is the single-most common reason that Americans visit the emergency room, where an abdominal CT scan is usually performed," noted Debiao Li, PhD, Director of the Biomedical Imaging Research Institute and Professor of Biomedical Sciences and Imaging at Cedars-Sinai, as well as senior author of the study.


"Even though most scans don't show any signs of cancer visible to the naked eyes of radiologists, some subjects eventually develop PDAC in the next few years," she added. "These pre-diagnostic CT images provide critical morphological information associated with biological changes at the pre-cancer or early-cancer stage, which can be extracted using AI to predict PDAC risk. Therefore, the objective of the proposed project is to uncover unique features in pre-diagnostic images using AI and develop PDAC prediction model based on these features."


Study Details

For their study, the researchers sought to stratify high-risk individuals for PDAC by identifying features in pre-diagnostic abdominal CT scans. A set of CT features-potentially predictive of PDAC-was identified in the analysis of 4,000 raw radiomic parameters extracted from pancreases in pre-diagnostic scans. The team of investigators reviewed electronic medical records to identify patients who were diagnosed with pancreatic cancer within the last 15 years and had undergone CT scans 6 months to 3 years before their diagnosis. CT images for these patients were considered normal at the time they were taken.


The naive Bayes classifier was developed for automatic classification of CT scans of the pancreas with high risk for PDAC. The authors relied on a set of 108 retrospective CT scans-36 scans from each healthy control, pre-diagnostic and diagnostic group-from 72 subjects for the study. Model development was performed on 66 multiphase CT scans, with external validation performed on 42 venous-phase CT scans.


Overall, the AI system achieved an average classification of 86 percent on the external dataset, with the authors concluding that "radiomic analysis of abdominal CT scans can unveil, quantify, and interpret micro-level changes in the pre-diagnostic pancreas and can efficiently assist in the stratification of high-risk individuals for PDAC."


The diagnosis of pancreatic cancer "occurs late when therapies are not very effective," noted study co-author Stephen Pandol, MD, Director of Basic and Translational Pancreas Research and Program Director of the Gastroenterology Fellowship Program at Cedars-Sinai. "We know that earlier diagnosis will lead to better outcomes, including survival from the cancer. [And] we know from experience that CT scans are done before the diagnosis for a variety of reasons. But the diagnosis is not made from those CT scans. So, we hypothesized that the CT scans done prior have signals that can be picked up using AI methodologies so that early cancer can be detected."


Ultimately, "we envision that the methods we develop will be used on all abdominal CT scans to identify patients who either have early pancreatic cancer or the risk of pancreatic cancer," Pandol said. "This identification will lead to improved clinical care pathways and survival."


With further development and validation using a much larger dataset, the AI tool "can be used to identify individuals with high risk for pancreatic cancer in the next few years based on an abdominal CT scan," Li stated. "They will be followed up closely with imaging examinations or biopsy, which may allow detection of cancer early and surgical intervention while the tumors are still resectable, thus saving lives and increasing survival rate for pancreatic cancer."


Mark McGraw is a contributing writer.