Clinical validation of a novel system enabling non-radiation spinal deformity prediction using a combination of artificial intelligence and depth sensing technologies

Dr Zhang, Teng   (Principal Investigator (PI))

Dr Wong Kenneth Kwan Yee   (Co-Investigator)

Dr Cheung Jason Pui Yin   (Co-Investigator)

24

2021-10-01

100000

Clinical validation of a novel system enabling non-radiation spinal deformity prediction using a combination of artificial intelligence and depth sensing technologies

Artificial intelligence, Deformity, Diagnosis, Non-radiation, Portable, Screening

Others - Medicine, Dentistry and Health

08192266

Health and Medical Research Fund - Mini Grant

2020

On-going

Objectives: To validate our newly developed artificial intelligence (AI) system for objective appearance assessment and spinal alignments measurement, using high-resolution images acquired by an optical camera with concurrent depth-sensors (RGB-D). Hypothesis: Our AI-system using high-resolution optical images with concurrent depth-sensing technologies can objectively assess the appearance and spine alignment of patients with spinal deformity. Details include 1) depth values of the surface landmarks auto-detected via our AI-system on an RGB-D image can assess and predict the appearance alterations of deformity patients; 2) the X-ray synthesized using our AI-system from the RGB-D image can predict the Cobb angles (CAs) and deformity severity. Design and subjects: A prospective trial with six-month follow-ups to evaluate the efficacy of our system in assessing deformed spines. The subjects are spinal deformity patients. Instruments: An optical camera with depth sensors to take high-resolution RGB-D images of the patient’s back. Interventions: No additional interventions required. Main outcome measures: The primary outcomes include 1.1) pelvic imbalance calculated via the angle between the line connecting the auto-detected posterior superior iliac spines and the horizon, 1.2) rotation of the thoracolumbar spine calculated via the depth difference between the inferior scapular angles, and 1.3) central shift calculated by the horizontal distance between C7 and coccyx. The secondary outcomes include, 2.1) auto-classifications of deformity severities, and 2.2) the predicted CAs. Data analysis: Five outcomes will be compared with values assessed by spine specialists. Expected results: Our novel AI-system can objectively assess the appearance of deformity patients and predict its severity.