Prediction of Fusion Rod Curvature Angles in Posterior Scoliosis Correction Using Artificial Intelligence

Document Type : RESEARCH PAPER

Authors

1 Department of Biomedical Engineering, Amirkabir University of Technology, Tehran, Iran

2 Department of Mechanical Engineering, Sharif University of Technology, Tehran, Iran

3 Bone and Joint Reconstruction Research Center, Shafa Orthopedic Hospital, Iran University of Medical Sciences, Tehran, Iran

Abstract

Objectives: This study aimed to estimate post-operative rod angles in both concave and convex sides 
of scoliosis curvature in patients who had undergone posterior surgery, using neural networks and 
support vector machine (SVM) algorithms.
Methods: Radiographs of 72 scoliotic individuals were obtained to predict post-operative rod angles at all fusion 
levels (all spinal joints fused by rods). Pre-operative radiographical indices and pre-operatively resolved net joint 
moments of the apical vertebrae were employed as inputs for neural networks and SVM with biomechanical 
modeling using inverse dynamics analysis. Various group combinations were considered as inputs, based on the 
number of pre-operative angles and moments. Rod angles on both the concave and convex sides of the Cobb angle 
were considered as outputs. To assess the outcomes, root mean square errors (RMSEs) were evaluated between 
actual and predicted rod angles.
Results: Among eight groups with various combinations of radiographical and biomechanical parameters (such as 
Cobb, kyphosis, and lordosis, as well as joint moments), RMSEs of groups 4 (with seven radiographical angles in 
each case, which is greater in quantity) and 5 (with four radiographical angles and one biomechanical moment in 
each case, which is the least possible number of inputs with both radiographical and biomechanical parameters) 
were minimum, particularly in prediction of the concave rod kyphosis angle (errors were 5.5° and 6.3° for groups 4 
and 5, respectively). Rod lordosis angles had larger estimation errors than rod kyphosis ones.
Conclusion: Neural networks and SVM can be effective techniques for the post-operative estimation of rod angles 
at all fusion levels to assist surgeons with rod bending procedures before actual surgery. However, since rod lordosis 
fusion levels vary widely across scoliosis cases, it is simpler to predict rod kyphosis angles, which is more essential 
for surgeons.
 Level of evidence: IV

Keywords

Main Subjects

References

1. Sharifmoradi K, Naderi A, Saljoghiyan P. The Effect of Boston 
Brace on Lower Limb and L5-S1 Joint Contact Forces during 
Walking in Patients with Idiopathic Scoliosis. Scientific 
journal of Ilam University of medical sciences. 2017; 
25(3):90-99.
2. Luković V, Ćuković S, Milošević D, Devedžić G. An ontologybased module of the information system ScolioMedIS for 3D 
digital diagnosis of adolescent scoliosis. Comput Methods 
Programs Biomed. 2019:178:247-263. doi: 
10.1016/j.cmpb.2019.06.027.
3. Arima H, Hasegawa T, Yamato Y, et al. Clinical Outcomes and 
Complications of Corrective Fusion Surgery Down to L4, L5, 
and the Pelvis for Adult Scoliosis in Patients Younger than 50 
Years. Spine Surg Relat Res. 2022; 6(5):518-525. doi: 
10.22603/ssrr.2021-0220. 
4. Salmingo RA, Tadano S, Fujisaki K, Abe Y, Ito M. Relationship 
of forces acting on implant rods and degree of scoliosis 
correction. Clin Biomech (Bristol, Avon).2013; 28(2):122-8. 
doi: 10.1016/j.clinbiomech.2012.12.001.
5. Yang JH, Suh SW, Chang D-G. Comparison of surgical 
correction rates between titanium and cobalt-chrome-alloy 
as rod materials in adolescent idiopathic scoliosis. Sci Rep. 
2020; 10(1):10053. doi: 10.1038/s41598-020-66975-x. 
6. Wang W, Baran GR, Betz RR, Samdani AF, Pahys JM, Cahill PJ. 
The use of finite element models to assist understanding and 
treatment for scoliosis: a review paper. Spine Deform.2014; 
2(1):10-27. doi: 10.1016/j.jspd.2013.09.007. 
7. Salmingo RA, Tadano S, Abe Y, Ito M. Influence of implant rod 
curvature on sagittal correction of scoliosis deformity. Spine J. 
2014; 14(8):1432-9. doi: 10.1016/j.spinee.2013.08.042.
8. Le Navéaux F, Aubin C-E, Parent S, O Newton P, Labelle H. 3D 
rod shape changes in adolescent idiopathic scoliosis 
instrumentation: how much does it impact correction? Eur 
Spine J. 2017; 26(6):1676-1683. doi: 10.1007/s00586-017-
4958-1. 
9. Courvoisier A, Cebrian A, Simon J, et al. Virtual Scoliosis 
Surgery Using a 3D-Printed Model Based on Biplanar 
Radiographs. Bioengineering (Basel).2022; 9(9):469. doi: 
10.3390/bioengineering9090469. 
10. Shah K, Gadiya A, Shah M, et al. Does Three-dimensional 
printed patient-specific templates add benefit in revision 
surgeries for complex pediatric kyphoscoliosis deformity 
with sublaminar wires in situ? A clinical study. Asian Spine 
J.2021; 15(1):46-53. doi: 10.31616/asj.2019.0021.
11. Ghandhari H, Mahabadi MA, Nikouei F, et al. The role of 
spinopelvic parameters in clinical outcomes of spinal 
osteotomies in patients with sagittal imbalance. Arch Bone Jt 
Surg. 2018; 6(4):324-330. 
12. Solla F, Ilharreborde B, Blondel B, et al. Can Lumbopelvic 
Parameters Be Used to Predict Thoracic Kyphosis at all Ages? 
A National Cross-Sectional Study. Global Spine J. 2024; 
14(4):1116-1124. doi: 10.1177/21925682221134039.
13. Hu B, Wang L, Song Y, Yang X, Liu L, Zhou C. Postoperative 
proximal junctional kyphosis correlated with thoracic inlet 
angle in Lenke 5c adolescent idiopathic scoliosis patients 
following posterior surgery. BMC Musculoskelet Disord.2022; 
23(1):919. doi: 10.1186/s12891-022-05868-8. 
14. Junaid J. Prediction of Scoliosis Curve Correction Using Apical 
Fulcrum Bending Radiographs in Adolescent Idiopathic 
Scoliosis (AIS). Pakistan Journal of Medicine and Dentistry. 
2021; 10(3):47-53. 
15. Tokala DP, Nelson IW, Mehta JS, Powell R, Grannum S, 
Hutchinson MJ. Prediction of scoliosis curve correction using 
pedicle screw constructs in AIS: A comparison of fulcrum 
bend radiographs and traction radiographs under general 
anesthesia. Global Spine J.2018 (7):676-682. doi: 
10.1177/2192568218763147. 
16. Sudo H, Tachi H, Kokabu T, et al. In vivo deformation of 
anatomically pre-bent rods in thoracic adolescent idiopathic 
scoliosis. Sci Rep. 2021; 11(1):12622. doi: 10.1038/s41598-
021-92187-y. 
17. Kokabu T, Kanai S, Kawakami N, et al. An algorithm for using 
deep learning convolutional neural networks with three 
dimensional depth sensor imaging in scoliosis detection. 
Spine J. 2021; 21(6):980-987. doi: 
10.1016/j.spinee.2021.01.022.
18. Vergari C, Skalli W, Gajny L. A convolutional neural network 
to detect scoliosis treatment in radiographs. Int J Comput 
Assist Radiol Surg. 2020; 15(6):1069-1074. doi: 
10.1007/s11548-020-02173-4.  19. Phan P, Mezghani N, Wai EK, de Guise J, Labelle H. Artificial 
neural networks assessing adolescent idiopathic scoliosis: 
comparison with Lenke classification. Spine J. 2013; 
13(11):1527-33. doi: 10.1016/j.spinee.2013.07.449.
20. Mezghani N, Phan P, Mitiche A, Labelle H, De Guise JA. A 
Kohonen neural network description of scoliosis fused 
regions and their corresponding Lenke classification. Int J 
Comput Assist Radiol Surg. 2012;7(2):257-64. doi: 
10.1007/s11548-011-0667-0.
21. Rak M, Steffen J, Meyer A, Hansen C, Tönnies KD. Combining 
convolutional neural networks and star convex cuts for fast 
whole spine vertebra segmentation in MRI. Comput Methods 
Programs Biomed. 2019:177:47-56. doi: 
10.1016/j.cmpb.2019.05.003.
22. Nozawa K, Maki S, Furuya T, et al. Magnetic resonance image 
segmentation of the compressed spinal cord in patients with 
degenerative cervical myelopathy using convolutional neural 
networks. Int J Comput Assist Radiol Surg. 2023; 18(1):45-54. 
doi: 10.1007/s11548-022-02783-0.
23. Galbusera F, Casaroli G, Bassani T. Artificial intelligence and 
machine learning in spine research. JOR Spine. 2019; 
2(1):e1044. doi: 10.1002/jsp2.1044.
24. Yang D, Lee T, Lai K, et al. Semi-automatic method for presurgery scoliosis classification on X-ray images using Bending 
Asymmetry Index. Int J Comput Assist Radiol Surg. 2022; 
17(12):2239-2251. doi: 10.1007/s11548-022-02740-x.
25. Peng L, Lan L, Xiu P, et al. Prediction of proximal junctional 
kyphosis after posterior scoliosis surgery with machine 
learning in the Lenke 5 adolescent idiopathic scoliosis 
patient. Front Bioeng Biotechnol. 2020:8:559387. doi: 
10.3389/fbioe.2020.559387.
26. Abedi R, Fatouraee N, Bostanshirin M, Arjmand N, Ghandhari 
H. Prediction of Post-operative Clinical Indices in Scoliosis 
Correction Surgery Using an Adaptive Neuro-fuzzy Interface 
System. Arch Bone Jt Surg. 2023; 11(4):241-247. doi:
10.22038/ABJS.2022.66559.3176.
27. Garg B, Mehta N, Bansal T, Malhotra R. EOS® imaging: 
Concept and current applications in spinal disorders. J Clin 
Orthop Trauma. 2020; 11(5):786-793. doi: 
10.1016/j.jcot.2020.06.012. 
28. Melhem E, Assi A, El Rachkidi R, Ghanem I. EOS® biplanar Xray imaging: concept, developments, benefits, and limitations. 
J Child Orthop. 2016; 10(1):1-14. doi: 10.1007/s11832-016-
0713-0. 
29. Schmid S, Burkhart KA, Allaire BT, Grindle D, Anderson DE. 
Musculoskeletal full-body models including a detailed 
thoracolumbar spine for children and adolescents aged 6–
18 years. J Biomech. 2020:102:109305. doi: 
10.1016/j.jbiomech.2019.07.049.
30. Schmid S, Connolly L, Moschini G, Meier ML, Senteler M. Skin 
marker-based subject-specific spinal alignment modeling: A 
feasibility study. J Biomech. 2022:137:111102. doi: 
10.1016/j.jbiomech.2022.111102.
31. Salmingo R, Tadano S, Fujisaki K, Abe Y, Ito M. Corrective 
force analysis for scoliosis from implant rod deformation. Clin 
Biomech (Bristol, Avon). 2012; 27(6):545-50. doi: 
10.1016/j.clinbiomech.2012.01.004. 
32. Wang X, Boyer L, Le Naveaux F, Schwend RM, Aubin C-E. How 
does differential rod contouring contribute to 3-dimensional 
correction and affect the bone-screw forces in adolescent 
idiopathic scoliosis instrumentation? Clin Biomech (Bristol, 
Avon).2016:39:115-121. doi: 
10.1016/j.clinbiomech.2016.10.002.
33. Kamal Z, Rouhi G, Arjmand N, Adeeb S. A stability-based 
model of a growing spine with adolescent idiopathic scoliosis: 
A combination of musculoskeletal and finite element 
approaches. Med Eng Phys. 2019:64:46-55. doi: 
10.1016/j.medengphy.2018.12.015.
34. Skov ST, Li H, Hansen ES, et al. New growth rod concept 
provides three dimensional correction, spinal growth, and 
preserved pulmonary function in early-onset scoliosis. Int 
Orthop. 2020; 44(9):1773-1783. doi: 10.1007/s00264-020-
04604-y.
35. Lechner R, Putzer D, Dammerer D, Liebensteiner M, Bach C, 
Thaler M. Comparison of two-and three-dimensional 
measurement of the Cobb angle in scoliosis. Int Orthop. 2017; 
41(5):957-962. doi: 10.1007/s00264-016-3359-0.
The Archives of Bone and Joint Surgery
Volume 12, Issue 7
July 2024
Pages 494-505

Files

History

  • Receive Date: 12 December 2023
  • Revise Date: 05 May 2024
  • Accept Date: 27 April 2024

Share

How to cite

Statistics