The following is a summary of “Development of a Diagnostic Artificial Intelligence Tool for Lateral Lymph Node Metastasis in Advanced Rectal Cancer,” published in the December 2023 issue of Gastroenterology by Ozaki, et al.
With or without chemoradiation treatment, metastatic lateral lymph node removal can help people with rectal cancer live longer. However, the best MRI factors for finding tumors in side lymph nodes are still unknown. For a study, researchers sought to create an artificial intelligence (AI) tool that uses deep learning to diagnose lateral lymph node spread in people who had major surgery and lateral lymph node removal for rectal cancer. Twenty-nine people with rectal cancer who had major surgery and lateral lymph node removal at fifteen hospitals took part in the study. They were split into three groups: training (n = 139), test (n = 17), and evaluation (n = 53).
In the neoadjuvant treatment group, pictures taken before treatment were called baseline pictures, and pictures taken after treatment were called presurgery pictures. In the group that had upfront surgery, pictures taken before surgery were labeled as baseline images and images taken before surgery. They used baseline and pre-surgery pictures to build two types of artificial intelligence by feeding the patches from these images into ResNet-18. Then, they tested how well they could diagnose the problem. In total, 124 patients only had surgery, 52 had preoperative chemotherapy, and 33 had chemoradiotherapy. In the training group, 2,418 lateral lymph nodes were removed. In the test group, 279 were removed. And in the evaluation group, 850 were removed.
The rates of metastasis were 2.8%, 0.7%, and 3.7%, in that order. The precision-recall area under the curve for the baseline picture in the confirmation group was 0.870, and for the presurgery image, it was 0.963. Images taken at baseline and before surgery were both good at finding tumors in lateral lymph nodes, but images taken before surgery were more accurate than images taken at baseline. An AI tool could be used to diagnose lateral lymph node spread accurately.
Source: journals.lww.com/dcrjournal/abstract/2023/12000/development_of_a_diagnostic_artificial.11.aspx