Annotation-Efficient Deep Learning Model for Pancreatic Cancer Diagnosis and Classification Using CT Images: A Retrospective Diagnostic Study

Abstract

Simple Summary In computer-assisted diagnostics for pancreatic cancer, attributes featuring irregular contours and indistinct boundaries on CT images present challenges in acquiring high-quality annotations. In response to this issue, we have devised an innovative self-supervised learning algorithm, engineered to enhance the differentiation of malignant and benign lesions. This innovation obviates the necessity for radiologist intervention, thus facilitating the precise classification of pancreatic cancer. By employing a pseudo-lesion segmentation self-supervised learning model, which capitalizes on automatically generated high-quality training data, we have managed to significantly elevate the performance of both convolutional neural network-based and transformer-based deep learning models. The aim of this study was to develop a novel deep learning (DL) model without requiring large-annotated training datasets for detecting pancreatic cancer (PC) using computed tomography (CT) images. This retrospective diagnostic study was conducted using CT images collected from 2004 and 2019 from 4287 patients diagnosed with PC. We proposed a self-supervised learning algorithm (pseudo-lesion segmentation (PS)) for PC classification, which was trained with and without PS and validated on randomly divided training and validation sets. We further performed cross-racial external validation using open-access CT images from 361 patients. For internal validation, the accuracy and sensitivity for PC classification were 94.3% (92.8-95.4%) and 92.5% (90.0-94.4%), and 95.7% (94.5-96.7%) and 99.3 (98.4-99.7%) for the convolutional neural network (CNN) and transformer-based DL models (both with PS), respectively. Implementing PS on a small-sized training dataset (randomly sampled 10%) increased accuracy by 20.5% and sensitivity by 37.0%. For external validation, the accuracy and sensitivity were 82.5% (78.3-86.1%) and 81.7% (77.3-85.4%) and 87.8% (84.0-90.8%) and 86.5% (82.3-89.8%) for the CNN and transformer-based DL models (both with PS), respectively. PS self-supervised learning can increase DL-based PC classification performance, reliability, and robustness of the model for unseen, and even small, datasets. The proposed DL model is potentially useful for PC diagnosis.