Supervised segmentation framework for evaluation of diffusion tensor imaging indices in skeletal muscle

Abstract

Diffusion tensor imaging (DTI) is becoming a relevant diagnostic tool to understand muscle disease and map muscle recovery processes following physical activity or after injury. Segmenting all the individual leg muscles, necessary for quantification, is still a time-consuming manual process. The purpose of this study was to evaluate the impact of a supervised semi-automatic segmentation pipeline on the quantification of DTI indices in individual upper leg muscles. Longitudinally acquired MRI datasets (baseline, post-marathon and follow-up) of the upper legs of 11 subjects were used in this study. MR datasets consisted of a DTI and Dixon acquisition. Semi-automatic segmentations for the upper leg muscles were performed using a transversal propagation approach developed by Ogier et al on the out-of-phase Dixon images at baseline. These segmentations were longitudinally propagated for the post-marathon and follow-up time points. Manual segmentations were performed on the water image of the Dixon for each of the time points. Dice similarity coefficients (DSCs) were calculated to compare the manual and semi-automatic segmentations. Bland-Altman and regression analyses were performed, to evaluate the impact of the two segmentation methods on mean diffusivity (MD), fractional anisotropy (FA) and the third eigenvalue (λ 3). The average DSC for all analyzed muscles over all time points was 0.92 ± 0.01, ranging between 0.48 and 0.99. Bland-Altman analysis showed that the 95% limits of agreement for MD, FA and λ 3 ranged between 0.5% and 3.0% for the transversal propagation and between 0.7% and 3.0% for the longitudinal propagations. Similarly, regression analysis showed good correlation for MD, FA and λ 3 (r = 0.99, p < 60; 0.0001). In conclusion, the supervised semi-automatic segmentation framework successfully quantified DTI indices in the upper-leg muscles compared with manual segmentation while only requiring manual input of 30% of the slices, resulting in a threefold reduction in segmentation time.

3D Automated Segmentation of Lower Leg Muscles Using Machine Learning on a Heterogeneous Dataset

“…In addition to that, the accuracy of the volume might not be the most important factor when analyzing clinical parameters such as fat fraction or diffusion parameters. A promising approach already showed diffusion parameters to be consistent comparing manual segmentation and semi-automated segmentation on the upper leg [36]. An interesting question for future studies would be to see the needed accuracy of a 3D labeling technique when analyzing clinical parameters [37].…”

Overview of MR Image Segmentation Strategies in Neuromuscular Disorders

“…The method was initially proposed and validated for the segmentation of the four muscles of the quadriceps femoris group in T 1 -weighted images of 11 healthy subjects ( 58 ). Validation has been then extended for the segmentation of all individual thigh muscles in healthy subjects ( 59 ). Mean DSC scores of 0.90 ± 0.03 was reported with a manual input for 30% of the slices only.…”

Robustness and stability of volume‐based tractography in a multicenter setting

“…Therefore, in a second step, 50 datasets from five different centers were pooled to assess the influence of semiautomatic segmentation approaches on VBT‐derived parameters. Semiautomatic segmentation techniques were successfully applied in healthy individuals and patients with neuromuscular diseases, and reduce the segmentation time significantly 11,12,35 . In this study we assessed the influence of simple, freely available semiautomatic segmentation approaches delineating three or five slices manually on tract properties and the diffusion metrics of healthy calf muscles.…”

“…The most prominent technique is volume‐based analysis, which can be performed using full manual segmentation or semiautomatic segmentation to separate individual muscles 10 . Semiautomatic segmentation techniques are increasingly popular in mDTI analysis because they provide similar accuracy as full manual segmentations and reduce the segmentation time 11,12 . By contrast, classic ROI‐based tractography has to be performed manually and requires a large amount of time and an experienced examiner 9 .…”