RESUMO
Optic never fibers in the visual pathway play significant roles in vision formation. Damages of optic nerve fibers are biomarkers for the diagnosis of various ophthalmological and neurological diseases; also, there is a need to prevent the optic nerve fibers from getting damaged in neurosurgery and radiation therapy. Reconstruction of optic nerve fibers from medical images can facilitate all these clinical applications. Although many computational methods are developed for the reconstruction of optic nerve fibers, a comprehensive review of these methods is still lacking. This paper described both the two strategies for optic nerve fiber reconstruction applied in existing studies, i.e., image segmentation and fiber tracking. In comparison to image segmentation, fiber tracking can delineate more detailed structures of optic nerve fibers. For each strategy, both conventional and AI-based approaches were introduced, and the latter usually demonstrates better performance than the former. From the review, we concluded that AI-based methods are the trend for optic nerve fiber reconstruction and some new techniques like generative AI can help address the current challenges in optic nerve fiber reconstruction.
RESUMO
BACKGROUND: Segmentation is a crucial and necessary step in diffusion tensor imaging (DTI) analysis of the cervical spinal cord. In existing studies, different diffusive metric maps [B0, fractional anisotropy (FA) and axial diffusivity (AD) maps] have been involved in the segmentation of tissues of the cervical spinal cord. The selection of a diffusive metric map for segmentation may affect the accuracy of segmentation and then affect the validity and effectiveness of the extracted diffusive features. However, there are few discussions on this problem. Therefore, this study would like to examine the effect of segmentation based on different diffusive metric maps for DTI analysis of the cervical spinal cord. METHODS: Twenty-nine healthy subjects and thirty patients with cervical spondylotic myelopathy (CSM) were finally included in this study. All subjects accepted DTI scanning at cervical levels from C2 to C7/T1. For healthy subjects, all cervical levels were included for analysis; while, for each patient, only one compressed cervical level was included. After DTI scanning, DTI metrics including B0, FA, AD, radial diffusivity (RD) and mean diffusivity (MD) were calculated. The evaluation was performed to B0, FA and AD maps from two aspects. First, the accuracy of segmentation was evaluated via a comparison between segmentation based on each diffusive metric map and segmentation based on an average image, which was acquired by averaging B0, FA, AD, RD and MD maps. The segmentation was achieved by a semi-automatic segmentation process, and the similarity between two segmentation results was denoted by the intersection of the union (IOU). Second, the diversity of extracted diffusive features was equalized as their performance in the classification of image pixels of different regions of interest (ROIs) and then was evaluated by mutual information (MI) and area under the curve (AUC). One-way ANOVA and Bonferroni's post hoc tests were applied to compare the evaluation results. RESULTS: One-way ANOVA suggested that there were differences (P<0.001) in IOU, MI and AUC values among the three diffusive metric maps for both healthy subjects and patients. The post-hoc tests further indicated that FA performed the best (P<0.001), i.e., the most substantial accuracy of segmentation and the highest diversity in extracted diffusive features. CONCLUSIONS: Different evaluation results had been observed for segmentation based on different diffusive metric maps, suggesting the necessity of selection of diffusive metric maps for segmentation in DTI analysis of the cervical spinal cord. Moreover, FA map is suggested for segmentation due to its best performance in the evaluation.