TY - GEN
T1 - Deep Learning-Based Quantification of Lumbar Disc Herniation on High-Resolution Magnetic Resonance Imaging
AU - Ding, Lizhong
AU - Chen, Wenxin
AU - Wu, Qina
AU - Zhu, Hongdian
AU - Li, Changsheng
AU - Zhao, Jing
AU - Duan, Xingguang
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.
PY - 2026
Y1 - 2026
N2 - This paper focuses on the segmentation of lumbar spine high-resolution magnetic resonance imaging. We designed an end-to-end deep learning-based model for automatic segmentation. Additionally, our method includes the automatic quantification of the herniated disc by incorporating acquisition parameters. We collected high-resolution magnetic resonance imaging from 17 patients with lumbar disc herniation. Based on deep learning techniques, we designed a convolutional neural network that accepts 3D inputs. The segmentation results of our model show high similarity to those obtained through manual segmentation, with the mean dice coefficient exceeding 0.8. By incorporating the acquisition parameters of magnetic resonance imaging, the automatic quantification method has an accuracy comparable to that of manual segmentation. Our research demonstrates that the designed deep learning model can reliably extract key features and reconstruct critical structures. Our method has the potential to be allowed for potential routine reporting in the clinical setting.
AB - This paper focuses on the segmentation of lumbar spine high-resolution magnetic resonance imaging. We designed an end-to-end deep learning-based model for automatic segmentation. Additionally, our method includes the automatic quantification of the herniated disc by incorporating acquisition parameters. We collected high-resolution magnetic resonance imaging from 17 patients with lumbar disc herniation. Based on deep learning techniques, we designed a convolutional neural network that accepts 3D inputs. The segmentation results of our model show high similarity to those obtained through manual segmentation, with the mean dice coefficient exceeding 0.8. By incorporating the acquisition parameters of magnetic resonance imaging, the automatic quantification method has an accuracy comparable to that of manual segmentation. Our research demonstrates that the designed deep learning model can reliably extract key features and reconstruct critical structures. Our method has the potential to be allowed for potential routine reporting in the clinical setting.
KW - deep learning
KW - high-resolution magnetic resonance imaging
KW - lumbar disc herniation
KW - medical image segmentation
UR - https://www.scopus.com/pages/publications/105030934997
U2 - 10.1007/978-981-95-6730-0_11
DO - 10.1007/978-981-95-6730-0_11
M3 - Conference contribution
AN - SCOPUS:105030934997
SN - 9789819567294
T3 - Communications in Computer and Information Science
SP - 153
EP - 164
BT - Advanced Computational Intelligence and Intelligent Informatics - 9th International Workshop, IWACIII 2025, Proceedings
A2 - Ma, Hongbin
A2 - Xin, Bin
A2 - She, Jinhua
A2 - Dai, Yaping
PB - Springer Science and Business Media Deutschland GmbH
T2 - 9th International Workshop on Advanced Computational Intelligence and Intelligent Informatics, IWACIII 2025
Y2 - 31 October 2025 through 4 November 2025
ER -