Validation of Artificial Intelligence in the Classification of Adolescent Idiopathic Scoliosis and the Compairment to Clinical Manual Handling

Abstract

Objective: The accurate measurement of Cobb angles is crucial for the effective clinical management of patients with adolescent idiopathic scoliosis (AIS). The Lenke classification system plays a pivotal role in determining the appropriate fusion levels for treatment planning. However, the presence of interobserver variability and time-intensive procedures presents challenges for clinicians. The purpose of this study is to compare the measurement accuracy of our developed artificial intelligence measurement system for Cobb angles and Lenke classification in AIS patients with manual measurements to validate its feasibility.

Methods: An artificial intelligence (AI) system measured the Cobb angle of AIS patients using convolutional neural networks, which identified the vertebral boundaries and sequences, recognized the upper and lower end vertebras, and estimated the Cobb angles of the proximal thoracic, main thoracic, and thoracolumbar/lumbar curves sequentially. Accordingly, the Lenke classifications of scoliosis were divided by oscillogram and defined by the AI system. Furthermore, a man-machine comparison (n = 300) was conducted for senior spine surgeons (n = 2), junior spine surgeons (n = 2), and the AI system for the image measurements of proximal thoracic (PT), main thoracic (MT), thoracolumbar/lumbar (TL/L), thoracic sagittal profile T5-T12, bending views PT, bending views MT, bending views TL/L, the Lenke classification system, the lumbar modifier, and sagittal thoracic alignment.

Results: In the AI system, the calculation time for each patient's data was 0.2 s, while the measurement time for each surgeon was 23.6 min. The AI system showed high accuracy in the recognition of the Lenke classification and had high reliability compared to senior doctors (ICC 0.962).

Conclusion: The AI system has high reliability for the Lenke classification and is a potential auxiliary tool for spinal surgeons.

Keywords: Artificial Intelligence; Cobb Angle; Deep Learning; Lenke Classification; Scoliosis; Spine.

FIGURE 1

Logic diagram and process of classification algorithm.

FIGURE 2

Draw points on four vertices of each vertebra for deep learning. The coronal image of whole spine boundary vertices of the vertebra from T1 to L5 were labeled manually (A); sagittal image (B) and its amplification (C). Right and left bending views (D, E).

FIGURE 3

Oscillogram of the endplate slope distribution. The labeled images were used to structure the artificial intelligence endplate slope distribution (A). The oscillogram obtained from the endplate slope distribution (B). The Cobb angles of the proximal thoracic (PT), main thoracic (MT), and thoracolumbar/lumbar (TL/L) curves were identified according to the oscillogram.

FIGURE 4

Oscillograms of Lenke 1 of adolescent idiopathic scoliosis. Coronal Lenke 1 classification image (A). The oscillogram is displayed by sinuoid, and the green part represents PT (B). The PT was less than 25° in the bending view (G, H). The Cobb angle of the main curve was calculated by the difference in the value between the trough and peak in the oscillogram (red) (B). Thoracic sagittal profile T5–T12 (C) in the oscillogram (red) (D).

FIGURE 5

Oscillograms of Lenke 2 of adolescent idiopathic scoliosis. The coronal view of the Lenke 2 classification image (A). The Cobb angle of the structural proximal thoracic (PT) curve was calculated by the difference in the values between the left peak and the trough in the oscillogram, and the main curve was calculated by the trough and right peak (B). The red part represents structural PT and main thoracic (MT) (E, F, G, H). Thoracic sagittal profile T5–T12 (C) in the oscillogram (red) (D).

FIGURE 6

Oscillograms of Lenke 3 of adolescent idiopathic scoliosis. The coronal view of the Lenke 3 classification image (A). The Cobb angle of the structural main thoracic (MT) curve was calculated by the difference in the values between the left trough and the peak in the oscillogram, and the TL/L curve was calculated using the peak and right trough. The tail section of the oscillogram indicates that thoracolumbar/lumbar (TL/L) was not a main bend, and the difference between the peak and trough is less than that of the main curve (B). The red part represents structural MT and TL/L (E, F, G, H). Thoracic sagittal profile T5–T12 (C) in the oscillogram (red) (D).

FIGURE 7

Oscillograms of Lenke 4 of adolescent idiopathic scoliosis. The coronal view of the Lenke 4 classification image (A). The oscillogram of Lenke 4 displayed a sharp broken line, and the value differences between one peak and one trough represent proximal thoracic (PT), main thoracic (MT), and thoracolumbar/lumbar (TL/L), respectively (B). The red part represents structural PT, MT, and TL/L (E, F, G, H). Thoracic sagittal profile T5–T12 (C) in the oscillogram (red) (D).

FIGURE 8

Oscillograms of Lenke 5 of adolescent idiopathic scoliosis. The coronal view of the Lenke 5 classification image (A). The oscillogram of the Lenke 5 curve was characterized by one peak and trough in the posterior segments that showed over 25° in the coronal view (red) (B) and less than 25° in the right bending view (green) (E, F). Thoracic sagittal profile T5–T12 (C) in the oscillogram (red) (D).

FIGURE 9

Oscillograms of Lenke 6 of adolescent idiopathic scoliosis. The coronal view of the Lenke 6 classification image (A). The Cobb angle of the structural main thoracic (MT) curve was calculated by the difference in the values between the left trough and peak in the oscillogram, and the thoracolumbar/lumbar (TL/L) curve was calculated by the peak and right trough. The tail section of the oscillogram represents TL/L as the main bend, in which the difference between the peak and trough is more than the main curve (B). The red part represents structural MT and TL/L (E, F, G, H). Thoracic sagittal profile T5–T12 (C) in the oscillogram (red) (D).