Convolutional Neural Network–Based Automated Segmentation of the Spinal Cord and Contusion Injury: Deep Learning Biomarker Correlates of Motor Impairment in Acute Spinal Cord Injury

D.B. McCoy; S.M. Dupont; C. Gros; J. Cohen-Adad; R.J. Huie; A. Ferguson; X. Duong-Fernandez; L.H. Thomas; V. Singh; J. Narvid; L. Pascual; N. Kyritsis; M.S. Beattie; J.C. Bresnahan; S. Dhall; W. Whetstone; J.F. Talbott

doi:10.3174/ajnr.A6020

Abstract

BACKGROUND AND PURPOSE: Our aim was to use 2D convolutional neural networks for automatic segmentation of the spinal cord and traumatic contusion injury from axial T2-weighted MR imaging in a cohort of patients with acute spinal cord injury.

MATERIALS AND METHODS: Forty-seven patients who underwent 3T MR imaging within 24 hours of spinal cord injury were included. We developed an image-analysis pipeline integrating 2D convolutional neural networks for whole spinal cord and intramedullary spinal cord lesion segmentation. Linear mixed modeling was used to compare test segmentation results between our spinal cord injury convolutional neural network (Brain and Spinal Cord Injury Center segmentation) and current state-of-the-art methods. Volumes of segmented lesions were then used in a linear regression analysis to determine associations with motor scores.

RESULTS: Compared with manual labeling, the average test set Dice coefficient for the Brain and Spinal Cord Injury Center segmentation model was 0.93 for spinal cord segmentation versus 0.80 for PropSeg and 0.90 for DeepSeg (both components of the Spinal Cord Toolbox). Linear mixed modeling showed a significant difference between Brain and Spinal Cord Injury Center segmentation compared with PropSeg (P < .001) and DeepSeg (P < .05). Brain and Spinal Cord Injury Center segmentation showed significantly better adaptability to damaged areas compared with PropSeg (P < .001) and DeepSeg (P < .02). The contusion injury volumes based on automated segmentation were significantly associated with motor scores at admission (P = .002) and discharge (P = .009).

CONCLUSIONS: Brain and Spinal Cord Injury Center segmentation of the spinal cord compares favorably with available segmentation tools in a population with acute spinal cord injury. Volumes of injury derived from automated lesion segmentation with Brain and Spinal Cord Injury Center segmentation correlate with measures of motor impairment in the acute phase. Targeted convolutional neural network training in acute spinal cord injury enhances algorithm performance for this patient population and provides clinically relevant metrics of cord injury.

ABBREVIATIONS:

BASICseg: Brain and Spinal Cord Injury Center segmentation
CNN: convolutional neural network
SC: spinal cord
SCI: spinal cord injury
SCT: Spinal Cord Toolbox

Footnotes

Disclosures: Charley Gros—RELATED: Grant: IVADO Labs.* Adam Ferguson—RELATED: Consulting Fee or Honorarium: University of Texas Medical Branch; Burke Neurological Institute, Weill Cornell School of Medicine, Comments: I have received honoraria for talks at medical schools; Support for Travel to Meetings for the Study or Other Purposes: University of Texas Medical Branch; Burke Neurological Institute, Weill Cornell School of Medicine; European Neurotrauma Summer School; International Symposium on Neural Regeneration; National Neurotrauma Symposium, Comments: I travel to give talks; Fees for Participation in Review Activities such As Data Monitoring Boards, Statistical Analysis, Endpoint Committees, and the Like: National Institutes of Health study section; Swiss National Science Foundation; Wings for Life Foundation; UNRELATED: Grants/Grants Pending: National Institutes of Health; Veterans Administration; Department of Energy; Wings for Life Foundation; Craig H. Neilsen Foundation, Comments: I serve as Principal Investigator or Co-Investigator on grant applications to federal agencies related to this work.* Vineeta Singh—RELATED: Grant: Craig H. Neilsen Foundation grant.* Lisa Pascual—RELATED: Grant: Department of Defense, Comments: “Early Critical Care Decisions and Outcomes after Spinal Cord Injury.”* Michael S. Beattie—RELATED: Grant: Department of Defense Congressionally Directed Medical Research Programs Spinal Cord Injury Research Program, Comments: Department of Defense award to support the TRACK-SCI study. I am Principal Investigator.* Jacqueline C. Bresnahan—RELATED: Grant: Department of Defense.* William Whetstone—RELATED: Grant: Department of Defense.* Jason F. Talbott—UNRELATED: Expert Testimony: Tindall Bennett & Shoup; RELATED: Grant: Department of Defense Grant SC120259. Russel J. Huie—RELATED: Grant: Department of Defense Grant SC120259.* Leigh Thomas—RELATED: Grant: Department of Defense; Craig H. Neilsen Foundation*; UNRELATED: Employment: University of California, San Francisco. Xuan Duong Fernandez—RELATED: Grant: Department of Defense, Craig H. Neilsen Foundation*; UNRELATED: Employment: University of California, San Francisco. *Money paid to the institution.
This work was supported by the Craig H. Neilsen Foundation and the Department of Defense grant SC120259.