Meeting Banner
Abstract #3768

A MRI-based model using Graph Convolutional Network combine Nomogram to predict Microvascular invasion diagnosis

Yang Zhou1, Ziqian Zhang1, Wenjuan Zhao1, Xinxin Wang1, Kun Wang2, Kuan Luan2, Jianxiu Lian3, and Mengchao Shi3
1Harbin Medical University Cancer Hospital, Harbin 150010, Harbin, China, 2Harbin Engineering University, Harbin 150001, Harbin, China, 3Philips Healthcare, Beijing, China


Accurate preoperative assessment of microvascular invasion (MVI) can help clinicians choose more reasonable treatment options, reduce the recurrence of HCC patients after surgical treatment, and improve the survival rate of patients. In this study, Graph convolutional network (GCN) was used to build a preoperative diagnostic model for MVI for mining the correlation between radiomic features. The results revealed that the value of the predicted MVI nomogram established was 0.884 in the validation when the radiographic characteristics of the patients were combined with graph convolutional network Score(GS).

This abstract and the presentation materials are available to members only; a login is required.

Join Here