Radiomics-based models can detect pancreatic cancer well before clinical diagnosis

A recent analysis published in Gastroenterology highlighted the effectiveness of radiomics-based machine learning models in detecting pancreatic cancer in the prediagnostic stage of its development. 

After testing numerous machine learning models, Mukherjee et al. found that a radiomics-based model yielded the highest diagnostic accuracy for predicting who would develop pancreatic cancer in the three to 36 months following their abdominal CT scans. In comparison to the radiologists who interpreted the exams, all four machine learning models yielded higher AUCs. 

These models also appeared to be effective across multiple CT scanners and slice thicknesses—an accomplishment that the co-authors of a new paper describe as an important feature of the study. 

“Clinical applications of radiomics have been inhibited by limited reproducibility and generalizability across sites, scanners, and protocols, but the authors of this study took convincing steps to overcome these limitations,” explained the AJR co-authors Michael H. Rosenthal, MD, PhD, and Khoschy Schawkat, MD, PhD, both of Brigham and Women's Hospital

Some of these steps included manual labeling of the pancreas, a “rigorous” feature extraction using an open-source package and the removal of features that were strongly dependent on slice thickness. 

The researchers of the study also included a sample set not used for training, in addition to completing validations using an internal test set and an external pancreas CT dataset from the NIH. Rosenthal and Schawkat expressed that these steps increase their confidence in the positive performances observed for the models, adding that this makes the models more generalizable to other sites and populations. 

“In pancreatic cancer, opportunistic screening could identify individuals at sufficiently high risk to warrant active screening, as is currently performed for high-risk families,” Rosenthal and Schawkat suggested. 

The authors did, however, note that opportunistic screenings increase the likelihood of false positives, and although the radiomics-based models could certainly improve care in high-risk patients, radiologists must remain the as the “gatekeepers” to appropriately assess risks. 

The AJR article can be found here, the Gastroenterology study abstract here.

Hannah murhphy headshot

In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She joined Innovate Healthcare in 2021 and has since put her unique expertise to use in her editorial role with Health Imaging.

Around the web

CMS has more than doubled the CCTA payment rate from $175 to $357.13. The move, expected to have a significant impact on the utilization of cardiac CT, received immediate praise from imaging specialists.

The newly cleared offering, AutoChamber, was designed with opportunistic screening in mind. It can evaluate many different kinds of CT images, including those originally gathered to screen patients for lung cancer. 

AI-enabled coronary plaque assessments deliver significant value, according to late-breaking data presented at TCT. These AI platforms have gained considerable momentum in recent months, receiving expanded Medicare coverage in addition to a new Category I CPT code.

Trimed Popup
Trimed Popup