Automatic whole-brain segmentation analysis for diagnosing Sepsis-Associated Encephalopathy

Motivation: Sepsis-associated encephalopathy (SAE) is a severe neurological complication of sepsis, and its early diagnosis is difficult due to the lack of specific biomarkers. Goal(s): This study aims to assess the utility of FreeSurfer-based MRI brain segmentation to improve the diagnosis precision of SAE. Approach: 3D T1-weighted MRI scans on 22 SAE patients and 35 matched healthy controls. 546 brain regions were analyzed to detect structural changes associated with SAE. Results: Significant structural differences were observed in regions such as the hippocampus, thalamus, corpus callosum. Among the three diagnostic models developed, Model 3 exhibited the best predictive performance, with an AUC of 0.971. Impact: Whole brain segmentation using FreeSurfer, combined with artificial intelligence, offers a promising non-invasive method for diagnosing SAE. This approach could facilitate earlier detection and intervention, ultimately improving patient outcomes.