Abstract
The automatic segmentation of the pancreatic cyst lesion (PCL) is essential for the automated diagnosis of pancreatic cyst lesions on endoscopic ultrasonography (EUS) images. In this study, we proposed a deep-learning approach for PCL segmentation on EUS images. We employed the Attention U-Net model for automatic PCL segmentation. The Attention U-Net was compared with the Basic U-Net, Residual U-Net, and U-Net++ models. The Attention U-Net showed a better dice similarity coefficient (DSC) and intersection over union (IoU) scores than the other models on the internal test. Although the Basic U-Net showed a higher DSC and IoU scores on the external test than the Attention U-Net, there was no statistically significant difference. On the internal test of the cross-over study, the Attention U-Net showed the highest DSC and IoU scores. However, there was no significant difference between the Attention U-Net and Residual U-Net or between the Attention U-Net and U-Net++. On the external test of the cross-over study, all models showed no significant difference from each other. To the best of our knowledge, this is the first study implementing segmentation of PCL on EUS images using a deep-learning approach. Our experimental results show that a deep-learning approach can be applied successfully for PCL segmentation on EUS images.
Highlights
The pancreas is located behind the stomach and regulates blood sugar levels by secreting related hormones to the digestive system
The deep learning-based pancreatic cyst lesion (PCL) segmentation results are presented
We performed an internal test of the cross-over study to evaluate the deep learningbased PCL segmentation on the endoscopic ultrasonography (EUS) images
Summary
The pancreas is located behind the stomach and regulates blood sugar levels by secreting related hormones to the digestive system. Nguon et al developed a deep learning-based CAD system for the differentiation of subtypes of PCL, which are MCN and SCN (Serous cystic neoplasm) [13] They achieved an accuracy of up to 82.75% and an area under the receiver operating characteristic (AUROC) score of 0.88. Zhang et al used deep-learning algorithms for the detection of pancreas location and recognition of EUS station [15] They conducted pancreas segmentation related to anatomical structures, such as the abdominal aorta, pancreatic body, pancreatic tail, confluence, pancreatic head from the stomach, and pancreatic head from the descending part of the duodenum; they did not study pathological lesions like PCL. Tonozuka et al developed a deep learning-based CAD system for pancreatic cancer detection using a 7-layer convolutional neural network (CNN) from a single-institution dataset [16] They achieved an AUROC score of 92.4% for the validation set and 94.0% for the test set. This study implemented automatic PCL segmentation on EUS images based on a deep-learning approach
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have