Skip to main content

Applying a Convolutional Neural Network Based Iterative Algorithm to Automatically Measure Spinal Curvature on Radiographs for Children with Scoliosis

Abstract

Purpose

Accurate measurement of spinal curvature for adolescent idiopathic scoliosis (AIS) is important because it affects treatment decisions. Currently, the Cobb angle measured on a radiograph is the gold standard for spinal curvature assessment. However, manual measurements introduce inter- and intra-observer reliability challenges, and while fully automatic methods have been developed, performance could be improved. This paper reported a new approach using convolutional neural networks (CNNs) and an iterative vertebra location algorithm to calculate the Cobb angle automatically by segmenting the spinal and vertebral boundaries on posteroanterior radiographs.

Methods

Two CNNs for spinal column and vertebra segmentation were trained using 110 and 272 images, respectively. An iterative vertebra location algorithm was developed to localize individual vertebrae in the spinal column for segmentation. To evaluate the accuracy of the automatic Cobb angle measurements calculated from the vertebra segmentations, 100 new radiographs were used. The mean absolute difference (MAD), standard deviation of absolute differences (SD), and percent within clinical acceptance (≤ 5°) between manual and automatic measurements were reported as evaluation metrics.

Results

The MAD ± SD was 2.8° ± 2.8° and 88% of the measurements were within 5° of the manual measurements. The result was comparable to other literature, and this method worked for a wide range of curve severities. The average automatic measurement time per image was 90 s, which is clinically acceptable.

Conclusion

The automatic measurement method based on CNNs provided a comparable accuracy and speed on spinal curvature measurements on radiographs. It could be a valuable tool for reducing clinical workload and measurement variation.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6

References

  1. Weinstein, S., Dolan, L., Cheng, J., Danielsson, A., & Morcuende, J. (2008). Adolescent idiopathic scoliosis. The Lancet, 371(9623), 1527–1537. https://doi.org/10.1016/S0140-6736(08)60658-3

    Article  Google Scholar 

  2. Kotwicki, T. (2008). Evaluation of scoliosis today: Examination, X-rays and beyond. Disability and Rehabilitation, 30(10), 742–751. https://doi.org/10.1080/09638280801889519

    Article  PubMed  Google Scholar 

  3. Gstoettner, M., Sekyra, K., Walochnik, N., Winter, P., Wachter, R., & Bach, C. M. (2007). Inter- and intraobserver reliability assessment of the Cobb angle: Manual versus digital measurement tools. European Spine Journal, 16(10), 1587–1592. https://doi.org/10.1007/s00586-007-0401-3

    Article  PubMed  PubMed Central  Google Scholar 

  4. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. In Proceedings of the 25th International Conference on Neural Information Processing Systems—Volume 1 (pp. 1097–1105).

  5. Ronneberger, O., Fischer, P., & Brox, T. (2015). U-Net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer Assisted Intervention—MICCAI 2015 (pp. 234–241). Munich, Germany.

  6. Horng, M. H., Kuok, C. P., Fu, M. J., Lin, C. J., & Sun, Y. N. (2019). Cobb angle measurement of spine from X-Ray images using convolutional neural network. Computational and Mathematical Methods in Medicine. https://doi.org/10.1155/2019/6357171

    Article  PubMed  PubMed Central  Google Scholar 

  7. Fu, X., Yang, G., Zhang, K., Xu, N., & Wu, J. (2020). An automated estimator for Cobb angle measurement using multi-task networks. Neural Computing and Applications. https://doi.org/10.1007/s00521-020-05533-y

    Article  Google Scholar 

  8. Sardjono, T. A., Wilkinson, M. H. F., Veldhuizen, A. G., van Ooijen, P. M. A., Purnama, K. E., & Verkerke, G. J. (2013). Automatic Cobb angle determination from radiographic images. Spine, 38(20), E1256-1262. https://doi.org/10.1097/BRS.0b013e3182a0c7c3

    Article  PubMed  Google Scholar 

  9. Kim, K. C., Yun, H. S., Kim, S., & Seo, J. K. (2020). Automation of spine curve assessment in frontal radiographs using deep learning of vertebral-tilt vector. IEEE Access, 8, 84618–84630. https://doi.org/10.1109/ACCESS.2020.2992081

    Article  Google Scholar 

  10. Pan, Y., Chen, Q., Chen, T., Wang, H., Zhu, X., Fang, Z., & Lu, Y. (2019). Evaluation of a computer-aided method for measuring the Cobb angle on chest X-rays. European Spine Journal, 28(12), 3035–3043. https://doi.org/10.1007/s00586-019-06115-w

    Article  PubMed  Google Scholar 

  11. Wu, H., Bailey, C., Rasoulinejad, P., & Li, S. (2018). Automated comprehensive adolescent idiopathic scoliosis assessment using MVC-Net. Medical Image Analysis, 48, 1–11. https://doi.org/10.1016/j.media.2018.05.005

    Article  PubMed  Google Scholar 

  12. Galbusera, F., Niemeyer, F., Wilke, H. J., Bassani, T., Casaroli, G., Anania, C., Costa, F., Brayda-Bruno, M., & Sconfienza, L. M. (2019). Fully automated radiological analysis of spinal disorders and deformities: A deep learning approach. European Spine Journal, 28(5), 951–960. https://doi.org/10.1007/s00586-019-05944-z

    Article  PubMed  Google Scholar 

  13. Kingma, D., & Ba, J. (2015). Adam: A method for stochastic optimization. In 3rd International conference for learning representations.

  14. Al Arif, S. M. M. R., Knapp, K., & Slabaugh, G. (2018). Shape-Aware deep convolutional neural network for vertebrae segmentation. In B. Glocker, J. Yao, T. Vrtovec, A. Frangi, & G. Zheng (Eds.), Computational methods and clinical applications in musculoskeletal imaging (pp. 12–24). Springer.

    Chapter  Google Scholar 

  15. Karimi, D., & Salcudean, S. E. (2020). Reducing the hausdorff distance in medical image segmentation with convolutional neural networks. IEEE Transactions on Medical Imaging, 39(2), 499–513. https://doi.org/10.1109/TMI.2019.2930068

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

The authors would like to acknowledge the financial support of the Innovation grant from the Women and Children’s Health Research Institute and the Discovery Grant from the Natural Science and Engineering Research Council of Canada.

Author information

Authors and Affiliations

Authors

Contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by SS, JW and EL. The first draft of the manuscript was written by SS and JW and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Edmond Lou.

Ethics declarations

Conflict of interest

The authors have no conflicts of interest to declare that are relevant to the content of this article.

Ethical Approval

Ethics approval was granted from the University of Alberta Health Research Ethics Board (Pro00102044—Chart Review).

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Sigurdson, S., Wong, J., Reformat, M. et al. Applying a Convolutional Neural Network Based Iterative Algorithm to Automatically Measure Spinal Curvature on Radiographs for Children with Scoliosis. J. Med. Biol. Eng. (2022). https://doi.org/10.1007/s40846-022-00712-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s40846-022-00712-9

Keywords

  • Convolutional neural network
  • Machine learning
  • Radiograph
  • Scoliosis
  • Spinal curvature