Do current automated tractography methods hold up in tumour and epilepsy pathology? A comparison of four methods with expert manual tractography

Motivation: Assess feasibility of using automated tractography methods to accurately segment white matter tracts in paediatric tumour and epilepsy surgery patients with challenging lesion-related imaging characteristics. Goal(s): Evaluate performance of four state-of-the-art automated tractography methods, each utilising different segmentation techniques, to determine if they can produce results comparable with manual tractography. Approach: We compared each automated tractography methods against expert-based manual tractography by examining dice similarity, and analysed impact of lesion location to automated tract regions that produced false positive and false negative results. Results: Automated tractography results were unable to reproduce the same anatomical accuracy of tracts created manually by an expert. Impact: This study uniquely highlights the inaccuracies of using automated tractography for complex paediatric tumour and epilepsy surgery, particularly near lesions. It emphasises the need for standardised tract definitions and appropriate surgical case data for model training to enhance accuracy.