Accurate Intervertebral Disc Segmentation Approach Based on Deep Learning

Yu-Kai Cheng; Chih-Lung Lin; Yi-Chi Huang; Guo-Shiang Lin; Zhen-You Lian; Cheng-Hung Chuang

doi:10.3390/diagnostics14020191

Accurate Intervertebral Disc Segmentation Approach Based on Deep Learning

Diagnostics (Basel). 2024 Jan 16;14(2):191. doi: 10.3390/diagnostics14020191.

Authors

Yu-Kai Cheng¹, Chih-Lung Lin^{2

3}, Yi-Chi Huang⁴, Guo-Shiang Lin⁵, Zhen-You Lian⁶, Cheng-Hung Chuang⁶

Affiliations

¹ Department of Neurosurgery, China Medical University Hospital, Taichung 404, Taiwan.
² Department of Neurosurgery, Asia University Hospital, Taichung 413, Taiwan.
³ Department of Occupational Therapy, Asia University, Taichung 413, Taiwan.
⁴ Department of Radiology, Asia University Hospital, Taichung 413, Taiwan.
⁵ Department of Computer Science and Information Engineering, National Chin-Yi University of Technology, Taichung 411, Taiwan.
⁶ Department of Artificial Intelligence and Computer Engineering, National Chin-Yi University of Technology, Taichung 411, Taiwan.

Abstract

Automatically segmenting specific tissues or structures from medical images is a straightforward task for deep learning models. However, identifying a few specific objects from a group of similar targets can be a challenging task. This study focuses on the segmentation of certain specific intervertebral discs from lateral spine images acquired from an MRI scanner. In this research, an approach is proposed that utilizes MultiResUNet models and employs saliency maps for target intervertebral disc segmentation. First, a sub-image cropping method is used to separate the target discs. This method uses MultiResUNet to predict the saliency maps of target discs and crop sub-images for easier segmentation. Then, MultiResUNet is used to segment the target discs in these sub-images. The distance maps of the segmented discs are then calculated and combined with their original image for data augmentation to predict the remaining target discs. The training set and test set use 2674 and 308 MRI images, respectively. Experimental results demonstrate that the proposed method significantly enhances segmentation accuracy to about 98%. The performance of this approach highlights its effectiveness in segmenting specific intervertebral discs from closely similar discs.

Keywords: deep learning; distance transformation; intervertebral disc segmentation; u-net.

Grants and funding

This research was funded by the National Science and Technology Council, Taiwan, under Grant No. NSTC 111-2221-E-167-037, the National Chin-Yi University of Technology, Taiwan, under Grant No. NCUT 24-R-AI-016, and the trilateral collaboration project between Asia University, Asia University Hospital, and China Medical University Hospital, Taichung, Taiwan, under Grant No. ASIA-110-CMUH-08.