Identifying suicide attempts, ideation, and non-ideation in major depressive disorder from structural MRI data using deep learning

Jinlong Hu,Yangmin Huang,Xiaojing Zhang,Bin Liao,Gangqiang Hou,Ziyun Xu,Shoubin Dong,Ping Li
DOI: https://doi.org/10.1016/j.ajp.2023.103511
IF: 13.89
2023-02-12
Asian Journal of Psychiatry
Abstract:The present study aims to identify suicide risks in major depressive disorders (MDD) patients from structural MRI (sMRI) data using deep learning. In this paper, we collected the sMRI data of 288 MDD patients, including 110 patients with suicide ideation (SI), 93 patients with suicide attempts (SA), and 85 patients without suicidal ideation or attempts (NS). And we developed interpretable deep neural network models to classify patients in three tasks including SA-versus-SI, SA-versus-NS, and SI-versus-NS, respectively. Furthermore, we interpreted the models by extracting the important features that contributed most to the classification, and further discussed these features or ROI/brain regions.
psychiatry
What problem does this paper attempt to address?