Search

Objectives: Different machine learning algorithms (MLAs) for automated segmentation of gliomas have been reported in the literature. Automated segmentation of different tumor characteristics can be of added value for the diagnostic work-up and treatment planning. The purpose of this study was to provide an overview and meta-analysis of different MLA methods., Methods: A systematic literature review and meta-analysis was performed on the eligible studies describing the segmentation of gliomas. Meta-analysis of the performance was conducted on the reported dice similarity coefficient (DSC) score of both the aggregated results as two subgroups (i.e., high-grade and low-grade gliomas). This study was registered in PROSPERO prior to initiation (CRD42020191033)., Results: After the literature search (n = 734), 42 studies were included in the systematic literature review. Ten studies were eligible for inclusion in the meta-analysis. Overall, the MLAs from the included studies showed an overall DSC score of 0.84 (95% CI: 0.82-0.86). In addition, a DSC score of 0.83 (95% CI: 0.80-0.87) and 0.82 (95% CI: 0.78-0.87) was observed for the automated glioma segmentation of the high-grade and low-grade gliomas, respectively. However, heterogeneity was considerably high between included studies, and publication bias was observed., Conclusion: MLAs facilitating automated segmentation of gliomas show good accuracy, which is promising for future implementation in neuroradiology. However, before actual implementation, a few hurdles are yet to be overcome. It is crucial that quality guidelines are followed when reporting on MLAs, which includes validation on an external test set., Key Points: • MLAs from the included studies showed an overall DSC score of 0.84 (95% CI: 0.82-0.86), indicating a good performance. • MLA performance was comparable when comparing the segmentation results of the high-grade gliomas and the low-grade gliomas. • For future studies using MLAs, it is crucial that quality guidelines are followed when reporting on MLAs, which includes validation on an external test set., (© 2021. The Author(s).)

Background: The pedunculopontine nucleus (PPN) has been studied as a possible target for deep brain stimulation (DBS) for Parkinson's disease (PD). However, identifying the PPN can be challenging as the PPN is poorly visualized on conventional or even high-resolution MR scans. From histological studies it is known that the PPN is surrounded by major white matter tracts, which could function as possible anatomical landmarks., Methods: This study aimed to localize the PPN using 7T magnetic resonance (MR) imaging and diffusion tensor imaging (DTI) of its white matter borders in one post-mortem brain. Histological validation of the same specimen was performed. The PPN was segmented in both spaces, after which the two masks were compared using the Dice Similarity Index (DSI). The DSI compared the similarity of two samples on an inter-individual level and validated the MR findings. The error in distance between the center of the two 3D segmentations was measured by use of the Euclidean distance., Results: The PPN can be found in between the superior cerebellar peduncle and the medial lemniscus on both the FA-maps of the DTI images and the histological sections. The histological transverse sections showed to be superior to recognize the PPN (DSI: 1.0). The DTI images have a DSI of 0.82. The overlap-masks of both spaces showed a DSI of 0.32, whereas the concatenation-masks of both spaces showed a remarkable overlap, a DSI of 0.94. Euclidean distance of the overlap- and concatenation-mask in the two spaces showed to be 1.29 mm and 1.59 mm, respectively., Conclusion: This study supports previous findings that the PPN can be identified using FA-maps of DTI images. For possible clinical application in DBS localization, in vivo validation of the findings of our study is needed.