AI-Driven Segmentation and Automated Analysis of the Whole Sagittal Spine from X-ray Images for Spinopelvic Parameter Evaluation
Abstract
:1. Introduction
2. Materials and Methods
2.1. Study Design
2.2. Datasets
2.3. AI Model Architecture
2.4. Spinopelvic Parameter Measurement
- Cervical 2 Incidence (C2I) is the angle between a line connecting the center of the femoral heads (yellow dashed line) to the midpoint of the sacral superior endplate and the line perpendicular to the C2 inferior endplate (brown dashed line).
- Cervical 2 Slope (C2S) denotes the angle between the horizontal line (red dashed line) and the inferior endplates of C2.
- Cervical 2 to 7 Lordosis (C27L) refers to the angle between the inferior endplate of C2 and the inferior endplate of C7.
- Thoracic 1 Incidence (T1I) denotes the angle between a line connecting the center of the femoral heads (yellow dashed line) to the midpoint of the sacral superior endplate and the line perpendicular to the T1 superior endplate (brown dashed line).
- Thoracic 1 Slope (T1S) represents the angle between the horizontal line (red dashed line) and the superior endplates of T1.
- Thoracic Kyphosis (TK) refers to the angle between the superior endplate of T1 and the superior endplate of L1.
- Lumbar 1 Incidence (L1I) is the angle between a line connecting the center of the femoral heads (yellow dashed line) to the midpoint of the sacral superior endplate and the line perpendicular to the L1 superior endplate (brown dashed line).
- Lumbar Lordosis (LL) represents the angle between the superior endplate of L1 and the superior endplate of S1.
- Sacral Slope (SS) is the angle between the sacral superior endplate and the horizontal line (red dashed line).
- Pelvic Tilt (PT) refers to the angle between a line connecting the center of the femoral heads (yellow dashed line) to the midpoint of the sacral superior endplate and the vertical line (blue dashed line).
- Pelvic Incidence (PI) is the angle formed by a line connecting the center of the femoral heads (yellow dashed line) to the midpoint of the sacral superior endplate and the line perpendicular to the sacral superior endplate (brown dashed line). The green lines depict the parallel lines corresponding to each endplate.
- Cervical 2 to 7 Sagittal Vertical Axis (C2-7 SVA) is a parameter that measures the horizontal distance between a plumb line dropped from the center of the C2 vertebral body (blue dashed line) and the posterosuperior corner of the C7 vertebral body, within the spinal segment that includes the second to seventh cervical vertebrae.
- Cervical 7 Sagittal Vertical Axis (C7 SVA) measures the horizontal distance between the posterosuperior corner of S1 and the plumb line dropped from the center of the C7 vertebral body (blue dashed line).
2.5. Graphical User Interface (GUI)
2.6. Intraclass Correlation Coefficient (ICC) and Statistical Analyses
3. Results
3.1. AI Detection and Segmentation
3.2. Angle Measurement
3.3. Graphical User Interface (GUI)
3.4. Intraclass Correlation Coefficient (ICC)
4. Discussion
5. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Kwon, J.W.; Moon, S.H.; Park, S.Y.; Park, S.J.; Park, S.R.; Suk, K.S.; Kim, H.S.; Lee, B.H. Lumbar Spinal Stenosis: Review Update 2022. Asian Spine J. 2022, 16, 789–798. [Google Scholar] [CrossRef] [PubMed]
- Kalidindi, K.K.V.; Sangondimath, G.; Bansal, K.; Vishwakarma, G.; Chhabra, H.S. Introduction of a Novel “Segmentation Line” to Analyze the Variations in Segmental Lordosis, Location of the Lumbar Apex, and Their Correlation with Spinopelvic Parameters in Asymptomatic Adults. Asian Spine J. 2022, 16, 502–509. [Google Scholar] [CrossRef] [PubMed]
- Tartara, F.; Garbossa, D.; Armocida, D.; Di Perna, G.; Ajello, M.; Marengo, N.; Bozzaro, M.; Petrone, S.; Giorgi, P.D.; Schiro, G.R.; et al. Relationship between lumbar lordosis, pelvic parameters, PI-LL mismatch and outcome after short fusion surgery for lumbar degenerative disease. Literature review, rational and presentation of public study protocol: RELApSE study (registry for evaluation of lumbar artrodesis sagittal alignEment). World Neurosurg. X 2023, 18, 100162. [Google Scholar] [CrossRef] [PubMed]
- Scheer, J.K.; Tang, J.A.; Smith, J.S.; Acosta, F.L., Jr.; Protopsaltis, T.S.; Blondel, B.; Bess, S.; Shaffrey, C.I.; Deviren, V.; Lafage, V.; et al. Cervical spine alignment, sagittal deformity, and clinical implications: A review. J. Neurosurg. Spine 2013, 19, 141–159. [Google Scholar] [CrossRef] [PubMed]
- Barrey, C.; Jund, J.; Perrin, G.; Roussouly, P. Spinopelvic alignment of patients with degenerative spondylolisthesis. Neurosurgery 2007, 61, 981–986, discussion 986. [Google Scholar] [CrossRef] [PubMed]
- Knott, P.T.; Mardjetko, S.M.; Techy, F. The use of the T1 sagittal angle in predicting overall sagittal balance of the spine. Spine J. 2010, 10, 994–998. [Google Scholar] [CrossRef] [PubMed]
- Vialle, R.; Levassor, N.; Rillardon, L.; Templier, A.; Skalli, W.; Guigui, P. Radiographic analysis of the sagittal alignment and balance of the spine in asymptomatic subjects. J. Bone Jt. Surg. Am. 2005, 87, 260–267. [Google Scholar] [CrossRef] [PubMed]
- Vrtovec, T.; Ibragimov, B. Spinopelvic measurements of sagittal balance with deep learning: Systematic review and critical evaluation. Eur. Spine J. 2022, 31, 2031–2045. [Google Scholar] [CrossRef] [PubMed]
- Joshi, R.S.; Haddad, A.F.; Lau, D.; Ames, C.P. Artificial Intelligence for Adult Spinal Deformity. Neurospine 2019, 16, 686–694. [Google Scholar] [CrossRef] [PubMed]
- Orosz, L.D.; Bhatt, F.R.; Jazini, E.; Dreischarf, M.; Grover, P.; Grigorian, J.; Roy, R.; Schuler, T.C.; Good, C.R.; Haines, C.M. Novel artificial intelligence algorithm: An accurate and independent measure of spinopelvic parameters. J. Neurosurg. Spine 2022, 37, 893–901. [Google Scholar] [CrossRef] [PubMed]
- IEEE. 2017 IEEE International Conference on Computer Vision ICCV 2017, Venice, Italy, 22–29 October 2017; IEEE Piscataway: Piscataway, NJ, USA, 2017. [Google Scholar]
- He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar]
- Lin, T.-Y.; Dollár, P.; Girshick, R.; He, K.; Hariharan, B.; Belongie, S. Feature pyramid networks for object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 21–26 July 2017; pp. 2117–2125. [Google Scholar]
- Long, J.; Shelhamer, E.; Darrell, T. Fully convolutional networks for semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 7–12 June 2015; pp. 3431–3440. [Google Scholar]
- Ren, S.; He, K.; Girshick, R.; Sun, J. Faster r-cnn: Towards real-time object detection with region proposal networks. Adv. Neural Inf. Process. Syst. 2015, 28. [Google Scholar] [CrossRef] [PubMed]
- Kingma, D.P.; Ba, J. Adam: A method for stochastic optimization. arXiv 2014, arXiv:1412.6980. [Google Scholar]
- Fujimori, T.; Suzuki, Y.; Takenaka, S.; Kita, K.; Kanie, Y.; Kaito, T.; Ukon, Y.; Watabe, T.; Nakajima, N.; Kido, S. Development of artificial intelligence for automated measurement of cervical lordosis on lateral radiographs. Sci. Rep. 2022, 12, 15732. [Google Scholar] [CrossRef] [PubMed]
- Galbusera, F.; Niemeyer, F.; Wilke, H.-J.; Bassani, T.; Casaroli, G.; Anania, C.; Costa, F.; Brayda-Bruno, M.; Sconfienza, L.M. Fully automated radiological analysis of spinal disorders and deformities: A deep learning approach. Eur. Spine J. 2019, 28, 951–960. [Google Scholar] [CrossRef] [PubMed]
- Yeh, Y.-C.; Weng, C.-H.; Huang, Y.-J.; Fu, C.-J.; Tsai, T.-T.; Yeh, C.-Y. Deep learning approach for automatic landmark detection and alignment analysis in whole-spine lateral radiographs. Sci. Rep. 2021, 11, 7618. [Google Scholar] [CrossRef] [PubMed]
- Lee, C.S.; Kang, S.S. Spino-pelvic parameters in adult spinal deformities. J. Korean Orthop. Assoc. 2016, 51, 9–29. [Google Scholar] [CrossRef]
- Kang, K.-B.; Kim, Y.J.; Muzaffar, N.; Yang, J.-H.; Kim, Y.B.; Yeo, E.-D. Changes of sagittal spinopelvic parameters in normal Koreans with age over 50. Asian Spine J. 2010, 4, 96. [Google Scholar] [CrossRef] [PubMed]
Comparison of Measurements | C2I | C27L | C2S | T1S | C2-7 SVA | C7 SVA | T1I | TK | L1I | LL | PT | SS | PI | Time (s) |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Specialist vs. Resident—Correlation | 0.672 | 0.861 | 0.836 | 0.656 | 0.962 | 0.832 | 0.583 | 0.641 | 0.871 | 0.623 | 0.717 | 0.291 | 0.492 | 0.57 |
Specialist vs. Resident—R2 | 0.451 | 0.741 | 0.698 | 0.431 | 0.926 | 0.692 | 0.34 | 0.411 | 0.759 | 0.388 | 0.515 | 0.085 | 0.242 | 0.325 |
Specialist vs. Resident—Mean Error | 7.98 | 6.204 | 5.143 | 6.49 | 2.282 | 9.594 | 8.816 | 8.49 | 6.143 | 10.327 | 5.918 | 8.286 | 9.429 | 184.531 |
Specialist vs. Resident—STD of Error | 6.498 | 3.959 | 3.362 | 4.59 | 2.862 | 16.507 | 6.886 | 5.588 | 5.067 | 7.372 | 4.526 | 7.682 | 6.402 | 143.664 |
Specialist vs. AI—Correlation | 0.92 | 0.955 | 0.913 | 0.836 | 0.978 | 0.922 | 0.848 | 0.794 | 0.96 | 0.79 | 0.844 | 0.685 | 0.794 | 0.258 |
Specialist vs. AI—R2 | 0.847 | 0.912 | 0.833 | 0.699 | 0.956 | 0.85 | 0.72 | 0.631 | 0.921 | 0.623 | 0.713 | 0.469 | 0.631 | 0.066 |
Specialist vs. AI—Mean Error | 4.112 | 3.625 | 3.806 | 3.773 | 2.112 | 5.219 | 5.583 | 6.484 | 3.703 | 8.789 | 3.577 | 6.036 | 5.689 | 736.245 |
Specialist vs. AI—STD of Error | 2.88 | 2.423 | 3.266 | 2.812 | 2.005 | 11.806 | 4.335 | 4.634 | 2.844 | 5.739 | 3.578 | 4.299 | 5.249 | 112.654 |
Mean (Second) | Mean Error | Standard Error | |
---|---|---|---|
Resident | 904.15 | 204.97 | 28.27 |
Specialist | 732.4 | 105.1 | 16.1 |
AI | 3.73 | 0.73 | 0.1 |
Variable | Specialist vs. Resident | CI95% | p | Specialist vs. AI | CI95% | p |
---|---|---|---|---|---|---|
ICC | ICC | |||||
C2I | 0.67 | [0.48, 0.80] | <0.01 | 0.92 | [0.86, 0.95] | <0.01 |
C2S | 0.83 | [0.70, 0.90] | <0.01 | 0.9 | [0.83, 0.94] | <0.01 |
C27L | 0.85 | [0.75, 0.92] | <0.01 | 0.95 | [0.92, 0.97] | <0.01 |
T1I | 0.58 | [0.36, 0.74] | <0.01 | 0.84 | [0.73, 0.91] | <0.01 |
T1S | 0.61 | [0.36, 0.77] | <0.01 | 0.83 | [0.72, 0.9] | <0.01 |
TK | 0.62 | [0.42, 0.77] | <0.01 | 0.74 | [0.45, 0.87] | <0.01 |
L1I | 0.87 | [0.78, 0.92] | <0.01 | 0.95 | [0.92, 0.97] | <0.01 |
LL | 0.62 | [0.42, 0.77] | <0.01 | 0.76 | [0.57, 0.86] | <0.01 |
SS | 0.27 | [−0.0, 0.50] | 0.027 | 0.62 | [0.34, 0.79] | <0.01 |
PT | 0.69 | [0.49, 0.82] | <0.01 | 0.85 | [0.74, 0.91] | <0.01 |
PI | 0.49 | [0.24, 0.68] | <0.01 | 0.78 | [0.63, 0.87] | <0.01 |
C2-7 SVA | 0.96 | [0.92, 0.98] | <0.01 | 0.98 | [0.96, 0.99] | <0.01 |
C7 SVA | 0.79 | [0.65, 0.88] | <0.01 | 0.92 | [0.86, 0.95] | <0.01 |
Time(s) | 0.31 | [−0.07, 0.61] | <0.01 | 0 | [−0.01, 0.01] | 0.491 |
Variable | Specialist vs. Resident | Specialist vs. AI |
---|---|---|
C2I | ||
C2S | ||
C27L | ||
T1I | ||
T1S | ||
TK | ||
L1I | ||
LL | ||
SS | ||
PT | ||
PI | ||
C2-7 SVA | ||
C7 SVA | ||
Time (s) |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Song, S.-Y.; Seo, M.-S.; Kim, C.-W.; Kim, Y.-H.; Yoo, B.-C.; Choi, H.-J.; Seo, S.-H.; Kang, S.-W.; Song, M.-G.; Nam, D.-C.; et al. AI-Driven Segmentation and Automated Analysis of the Whole Sagittal Spine from X-ray Images for Spinopelvic Parameter Evaluation. Bioengineering 2023, 10, 1229. https://doi.org/10.3390/bioengineering10101229
Song S-Y, Seo M-S, Kim C-W, Kim Y-H, Yoo B-C, Choi H-J, Seo S-H, Kang S-W, Song M-G, Nam D-C, et al. AI-Driven Segmentation and Automated Analysis of the Whole Sagittal Spine from X-ray Images for Spinopelvic Parameter Evaluation. Bioengineering. 2023; 10(10):1229. https://doi.org/10.3390/bioengineering10101229
Chicago/Turabian StyleSong, Sang-Youn, Min-Seok Seo, Chang-Won Kim, Yun-Heung Kim, Byeong-Cheol Yoo, Hyun-Ju Choi, Sung-Hyo Seo, Sung-Wook Kang, Myung-Geun Song, Dae-Cheol Nam, and et al. 2023. "AI-Driven Segmentation and Automated Analysis of the Whole Sagittal Spine from X-ray Images for Spinopelvic Parameter Evaluation" Bioengineering 10, no. 10: 1229. https://doi.org/10.3390/bioengineering10101229