Automatic brain lesion segmentation on standard magnetic resonance images: a scoping review

Abstract

ObjectivesMedical image analysis practices face challenges that can potentially be addressed with algorithm-based segmentation tools. In this study, we map the field of automatic MR brain lesion segmentation to understand the clinical applicability of prevalent methods and study designs, as well as challenges and limitations in the field.DesignScoping review.SettingThree databases (PubMed, IEEE Xplore and Scopus) were searched with tailored queries. Studies were included based on predefined criteria. Emerging themes during consecutive title, abstract, methods and whole-text screening were identified. The full-text analysis focused on materials, preprocessing, performance evaluation and comparison.ResultsOut of 2990 unique articles identified through the search, 441 articles met the eligibility criteria, with an estimated growth rate of 10% per year. We present a general overview and trends in the field with regard to publication sources, segmentation principles used and types of lesions. Algorithms are predominantly evaluated by measuring the agreement of segmentation results with a trusted reference. Few articles describe measures of clinical validity.ConclusionsThe observed reporting practices leave room for improvement with a view to studying replication, method comparison and clinical applicability. To promote this improvement, we propose a list of recommendations for future studies in the field.

ResectVol: A tool to automatically segment and characterize lacunas in brain images

“…Although many studies have focused on segmenting brain lesions, 10 , 16 to the best of our knowledge, not as many have specifically targeted brain lacunas. Gau and colleagues 11 claim to be the first group to have done so.…”

“…Furthermore, available methods of automatic or semiautomatic segmentation of the volume of the lacuna do not provide information on which brain anatomic regions were resected. 10 In this study, we present and validate a tool developed by our group that automatically delineates and provides a 3D mask of the surgical lacuna, calculates the volume of the tissue resected, and identifies which brain structures were removed. This tool will facilitate and enhance our ability to evaluate surgical resections in detail and generate a range of possibilities to analyze the area resected and its relationship with other neuroimaging modalities and surgical outcomes.…”

“…Furthermore, available methods of automatic or semi‐automatic segmentation of the volume of the lacuna do not provide information on which brain anatomic regions were resected. 10 …”

Deep learning for automatic brain tumour segmentation on MRI: evaluation of recommended reporting criteria via a reproduction and replication study

“…3 Consequently, assessing which model design choices determine the empirical gains is challenging. [3][4][5] Critics have also pointed out that scientific reporting of study designs has often been insufficient, and that the analysis of results tends to be biased towards authors' desired outcomes. 4 6 7 These issues present critical challenges to realising the potential of artificial intelligence and translating promising scientific algorithms into reliable and trusted clinical decision support tools.…”

“…However, their outstanding performance comes at the cost of high complexity and inherent variability in model performance 3. Consequently, assessing which model design choices determine the empirical gains is challenging 3–5. Critics have also pointed out that scientific reporting of study designs has often been insufficient, and that the analysis of results tends to be biased towards authors’ desired outcomes 4 6 7.…”