Improving segmentation reliability of multi-scanner brain images using a generative adversarial network

Kai Niu; Xueyan Li; Li Zhang; Zhensong Yan; Wei Yu; Peipeng Liang; Yan Wang; Ching-Po Lin; Huimao Zhang; Chunjie Guo; Kuncheng Li; Tianyi Qian

doi:10.21037/qims-21-653

Improving segmentation reliability of multi-scanner brain images using a generative adversarial network

Quant Imaging Med Surg. 2022 Mar;12(3):1775-1786. doi: 10.21037/qims-21-653.

Authors

Kai Niu¹, Xueyan Li^{2

3}, Li Zhang⁴, Zhensong Yan⁵, Wei Yu⁵, Peipeng Liang⁶, Yan Wang⁷, Ching-Po Lin^{8

9}, Huimao Zhang⁴, Chunjie Guo^#⁴, Kuncheng Li^#¹⁰, Tianyi Qian⁵

Affiliations

¹ Department of Otorhinolaryngology Head and Neck Surgery, the First Hospital of Jilin University, Changchun, China.
² State Key Laboratory of Integrated Optoelectronics, College of Electronic Science and Engineering, Jilin University, Changchun, China.
³ College of Electronic Science and Engineering, Peng Cheng Laboratory, Shenzhen, China.
⁴ Department of Radiology, the First Hospital of Jilin University, Changchun, China.
⁵ AI Lab, QuantMind, Beijing, China.
⁶ School of Psychology, Capital Normal University, Beijing, China.
⁷ Key Laboratory of Symbol Computation & Knowledge Engineering, Ministry of Education, College of Computer Science & Technology, Jilin University, Changchun, China.
⁸ Neurological Research Center, the First Hospital of Jilin University, Changchun, China.
⁹ Institute of Neuroscience, Yang-Ming University, Taipei 112, Taiwan, China.
¹⁰ Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing, China.

^# Contributed equally.

Abstract

Background: Magnetic resonance (MR) images generated by different scanners generally have inconsistent contrast properties, making it difficult to perform a combined quantitative analysis of images from a range of scanners. In this study, we aimed to develop an automatic brain image segmentation model to provide a more reliable analysis of MR images taken with different scanners.

Methods: The spatially localized atlas network tiles-27 (SLANT-27) deep learning model was used to train the automatic segmentation module, based on a multi-center dataset of 1,917 three-dimensional (3D) T1-weighted MR images. Subsequently, a framework called Qbrain, consisting of a new generative adversarial network (GAN) image transfer module and the SLANT-27 segmentation module, was developed. Another 3D T1-weighted MRI interscan dataset of 48 participants who were scanned in 3 MRI scanners (1.5T Siemens Avanto, 3T Siemens Trio Tim, and 3T Philips Ingenia) on the same day was used to train and test the Qbrain model. Volumetric T1-weighted images were processed with Qbrain, SLANT-27, and FreeSurfer (FS). The automatic segmentation reliability across the scanners was assessed using test-retest variability (TRV).

Results: The reproducibility of different segmentation methods across scanners showed a consistent trend in the greater reliability and robustness of QBrain compared to SLANT-27 which, in turn, showed greater reliability and robustness compared to FS. Furthermore, when the GAN image transfer module was added, the mean segmentation error of the TRV of the 3T Siemens vs. 1.5T Siemens, the 3T Philips vs. 1.5T Siemens, and the 3T Siemens vs. 3T Philips scanners was reduced by 1.57%, 2.01%, and 0.56%, respectively. In addition, the segmentation model improved intra-scanner variability (0.9-1.67%) compared with that of FS (2.47-4.32%).

Conclusions: The newly developed QBrain method combined with GAN image transfer module and a SLANT-27 segmentation module was shown to improve the reliability of whole-brain automatic structural segmentation results across multiple scanners, thus representing a suitable alternative quantitative method of comparative brain tissue analysis for individual patients.

Keywords: Magnetic resonance imaging (MRI); brain, segmentation; deep learning; generative adversarial network (GAN).