Diagnostic imaging and functional outcome in patients with diffuse glioma grades 2 and 3. From presentation to the 12-month post-operative follow-up

Abstract

The overall aim of this thesis was to contribute to the improvement of diagnostic imaging and functional outcome practices in adult patients with diffuse glioma grades 2 or 3. At the preoperative stage our studies assessed the use of semi-automatic quantitative imaging methods, and at consecutive follow-ups the studies provided systematic data on patients’ functional outcome measurements. In all four studies the included patients were adults (over 18 years of age) and had a histopathological diagnosis indicating a diffuse glioma grades 2 or 3. Study I, reporting on a voxel-based quantitative method, showed that in the studied cohort tumor anatomical distribution was linked to the tumor molecular status. Study II, comparing methods of classification of complications in neurosurgery, observed that the newer classification system showed a small shift towards classifying complications as of higher order. Study III, describing individual level data, found that a considerable subgroup of patients reported unfavorable changes in global health status and fatigue score. Study IV, evaluating the usefulness of machine-based deep learning, showed that the machine-based deep learning model failed to deliver additional gain over the baseline clinical model in predicting IDH mutations with high specificity. In conclusion, our studies provided quantitative evidence on the use of semi-automatic imaging methods at the preoperative stage. Also, our studies raised awareness over the importance of individual level functional outcome assessment. Altogether, these studies have demonstrated that in an uncurable disease characterized by low annual incidence and prolonged survival times, research integration into clinical workflows and collaborative efforts across multiple centers hold a significant promise for enhancing diagnostic imaging and functional outcome practices.