ARTIFICIAL INTELLIGENCE TOOLS IN POST-PROCESSING ANALYSIS OF NEUROIMAGING: A REVIEW OF THE LAST DECADE (2016–2026)

Post-processing of neuroimaging data — the segmentation, quantification, and volumetric or perfusion analysis performed after an MRI or CT scan is acquired — has traditionally relied on manual or semi-automated work by radiologists and neurologists. Over the past decade, artificial intelligence (AI), and particularly deep learning, has been increasingly integrated into this stage of the neuro-diagnostic workflow. This review synthesizes peer-reviewed research published between 2016 and 2026 on AI-based post-processing tools across four major areas of neurology: multiple sclerosis lesion segmentation, Alzheimer’s disease and dementia classification/volumetry, acute ischemic stroke perfusion and hemorrhage analysis, and brain tumor segmentation/volumetry. For each area, the responsible authors, publication year, AI methodology, target disease, study methods, and reported results are summarized. A dedicated section then compares manual versus AI-automated image analysis across these same conditions, drawing on studies that directly measured agreement, reproducibility, and time efficiency between the two approaches. Overall, the literature indicates that deep learning-based post-processing tools achieve accuracy and reproducibility that is comparable to, and in several structured tasks (e.g., hippocampal volumetry, tumor sub-region quantification) more consistent than, manual expert analysis, while substantially reducing analysis time and inter-rater variability.