Quantitative Radiomic Features from Computed Tomography Can Predict Pancreatic Cancer up to 36 Months Before Diagnosis

Abstract

Objectives: Pancreatic cancer (PC) is the 3rd leading cause of cancer deaths. We aimed to detect early changes on computed tomography (CT) images associated with pancreatic ductal adenocarcinoma (PDAC) based on quantitative imaging features (QIF). Methods: Adults 18+ years of age diagnosed with PDAC in 2008-2018 were identified. Their CT scans 3 months-3 years prior to the diagnosis date were matched to up to two scans of controls. Pancreas was automatically segmented using a previously developed algorithm. 111 QIF were extracted. The dataset was randomly split for training/validation. Neighborhood and principal component analyses were applied to select the most important features. Conditional support vector machine was used to develop prediction algorithms. The computer labels were compared with manually reviewed CT images 2-3 years prior to the index date in 19 cases and 19 controls. Results: 227 scans from cases (stages: 35% I-II, 44% III-IV, 21% unknown) and 554 matched scans of healthy controls were included (average age 71 years; 51% females). In the validation dataset, accuracy measures were 94%-95%, and area under the curve (AUC) measures were 0.98-0.99. Sensitivity, specificity, positive predictive value, and negative predictive values were in the ranges of 88-91%, 96-98%, 91-95%, and 94-96%. QIF on CT examinations within 2-3 years prior to index date also had very high predictive accuracy (accuracy 95-98%; AUC 0.99-1.00). The QIF-based algorithm outperformed manual re-review of images for determination of PDAC-risk. Conclusions: QIF can accurately predict PDAC on CT imaging and represent promising biomarkers for early detection of pancreatic cancer.