Overview
The investigators aim to use artificial intelligence (AI) to help clinicians in diagnosing and assessing spinal deformities.
Description
Background Spinal deformity is a prevalent spinal disorder in both paediatric and adult populations. The spine alignment need to be quantitively assessed for further treatment planning. However, the current practice requires spine surgeons to manually place landmarks of endplates and key vertebrae. The process is laborious and prone to inter- and intra-rater variance. Thus, the investigators have developed an AI-powered spine alignment assessment system (AlignProCARE) to facilitate clinicians in fast, accurate and consistent analytical results.
The investigators aim to test and improve the performance of the spine alignment auto-analysis in all patients with spinal deformities in multiple centers including Malaysia, China, and Japan
- Objectives
-
- prospectively test the alignment assessment of patients' spinal deformities with whole spine X-rays (both PA and lateral) and nude back image with the assessment via AlignProCARE.
- Collect 500 labeled deformity radiographs and nude back images in both PA and lateral views per center. 150 patients need to be followed up with radiographs and nude back photos collected (all parameters measured again).
- Use transfer learning to update the current AlignProCARE for scoliosis analysis to form AlignProCARE+.
4 Qualitatively analyse the AlignProCARE+ using an independent dataset.
Eligibility
Inclusion Criteria:
- Idiopathic scoliosis, adult deformity (spondylolisthesis, idiopathic kyphosis, kyphoscoliosis, lordoscoliosis)
Exclusion Criteria:
- Refusal for imaging, postoperative patients