Segmentation of Peripheral Nerves From Magnetic Resonance Neurography: A Fully-Automatic, Deep Learning-Based Approach.

Fabian Balsiger,Lorenz Grunder,Olivier Scheidegger,Benedikt Wagner,Carolin Steindel,Mauricio Reyes,Marwan El-Koussy,Mirjam Arn,Waldo Valenzuela

doi:10.3389/fneur.2018.00777

Fabian Balsiger, Lorenz Grunder + Show 7 more

Open Access

https://doi.org/10.3389/fneur.2018.00777

Copy DOI

Abstract

Diagnosis of peripheral neuropathies relies on neurological examinations, electrodiagnostic studies, and since recently magnetic resonance neurography (MRN). The aim of this study was to develop and evaluate a fully-automatic segmentation method of peripheral nerves of the thigh. T2-weighted sequences without fat suppression acquired on a 3 T MR scanner were retrospectively analyzed in 10 healthy volunteers and 42 patients suffering from clinically and electrophysiologically diagnosed sciatic neuropathy. A fully-convolutional neural network was developed to segment the MRN images into peripheral nerve and background tissues. The performance of the method was compared to manual inter-rater segmentation variability. The proposed method yielded Dice coefficients of 0.859 ± 0.061 and 0.719 ± 0.128, Hausdorff distances of 13.9 ± 26.6 and 12.4 ± 12.1 mm, and volumetric similarities of 0.930 ± 0.054 and 0.897 ± 0.109, for the healthy volunteer and patient cohorts, respectively. The complete segmentation process requires less than one second, which is a significant decrease to manual segmentation with an average duration of 19 ± 8 min. Considering cross-sectional area or signal intensity of the segmented nerves, focal and extended lesions might be detected. Such analyses could be used as biomarker for lesion burden, or serve as volume of interest for further quantitative MRN techniques. We demonstrated that fully-automatic segmentation of healthy and neuropathic sciatic nerves can be performed from standard MRN images with good accuracy and in a clinically feasible time.

Highlights

Current state-of-the-art to diagnose and monitor the effects of potentially available treatments for peripheral neuropathies relies on clinical examination and electrodiagnostic studies (EDX)
The proposed method was successful in segmenting peripheral nerves with and without lesions with good accuracy both in healthy volunteers and patients suffering from sciatic neuropathy
The peripheral nerves in our images can be considered as small structures with a volume fraction of 0.143 ± 0.049 %

Summary

Introduction

Current state-of-the-art to diagnose and monitor the effects of potentially available treatments for peripheral neuropathies relies on clinical examination and electrodiagnostic studies (EDX). Quantitative assessment of peripheral nerves from MRN typically proceeds either by assessment of CSA or by the identification of regions of interest in which abnormal signal behavior or quantitative MR parameters are further analyzed In both cases, a segmentation of the peripheral nerve at interest is typically performed manually. Computer-assisted segmentation of peripheral nerves from MRN images has been addressed by Felisaz et al [15] They proposed a semi-automatic method to compartmentalize the tibial nerve in micro-neurography images and showed the potential of computer-assisted segmentation by associating peripheral neuropathy with decreased fascicularto-nerve volume ratio, increased nerve volume, and increased CSA [16]. Fully- or semi-automatic computer-assisted segmentation of peripheral nerves with a large extremities coverage may be an important future step to obtain quantitative imaging outcome measures at a larger scale for disease-specific diagnosis and treatment monitoring of peripheral neuropathies

Objectives

Methods

Conclusion