Multiscale and multiperception feature learning for pancreatic lesion detection based on noncontrast CT.

Pancreatic cancer is one of the most malignant tumours, demonstrating a poor prognosis and nearly identically high mortality and morbidity, mainly because of the difficulty of early diagnosis and timely treatment for localized stages. It is clinically important to develop a noncontrast CT (NCCT)-based pancreatic lesion detection model that could serve as an intelligent tool for diagnosing pancreatic cancer early. However, abdominal NCCT presents low contrast intensities for the pancreas and its lesions and complex anatomical structures, developing such a model is challenging. Therefore, we design a multiscale and multiperception (MSMP) feature learning network with ResNet50 coupled with a feature pyramid network (FPN) as the backbone for strengthening feature expressions. We added multiscale atrous convolutions to expand different receptive fields, contextual attention to perceive contextual information, and channel and spatial attention to focus on important channels and spatial regions, respectively. The MSMP network then acts as a feature extractor for proposing an NCCT-based pancreatic lesion detection model with image patches covering the pancreas as its input; Faster R-CNN is employed as the detection method for accurately detecting pancreatic lesions. By using the new MSMP network as a feature extractor, our model outperforms the conventional object detection algorithms in terms of the recall (75.40% and 90.95%), precision (40.84% and 68.21%), F1 score (52.98% and 77.96%), F2 score (64.48% and 85.26%) and Ap50 (53.53% and 70.14%) at the image and patient levels, respectively. The good performance of our new model implies that MSMP can mine NCCT imaging features for detecting pancreatic lesions from complex backgrounds well. The proposed detection model is expected to be developed as an intelligent method for the early detection of pancreatic cancer.