Albert Hong1, Neha Jaswal1, Lindsey Westover2, Eric C Parent3, Marc Moreau4, Douglas Hedden4, Samer Adeeb1. 1. Department of Civil and Environmental Engineering, University of Alberta, Edmonton, AB, Canada. 2. Department of Mechanical Engineering, University of Alberta, Edmonton, AB, Canada. 3. Department of Physical Therapy, Faculty of Rehabilitation Medicine, University of Alberta, Edmonton, AB, Canada. 4. Department of Surgery, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, AB, Canada.
Abstract
STUDY DESIGN: A validation study. OBJECTIVE: The aim of this study was to independently validate the diagnostic accuracy of surface topography (ST) classification trees to identify curve severity and progression using a new sample of data in participants with adolescent idiopathic scoliosis (AIS). SUMMARY OF BACKGROUND DATA: Radiographs for diagnosing and monitoring AIS involve harmful radiation exposure repeated at successive clinical visits. Classification trees using a novel ST technique have been proposed to determine curve severity and progression noninvasively that could be used to monitor scoliosis. METHODS: Forty-five adolescents with AIS treated nonoperatively, with ST scans and radiographs at baseline and follow-up (1 year later), were recruited from a scoliosis clinic. The Cobb angle (CA) from radiographs determined curve severity as mild (10° < CA < 25°) or moderate/severe (CA ≥ 25°) and progression as an increase >5°.ST scans were analyzed to calculate the best plane of symmetry and associated deviation color map. Root mean squares and maximum deviation were calculated for each area of asymmetry. ST measurements were analyzed using two published decision trees developed to maximize sensitivity and negative predictive value. Curves were classified as mild or moderate/severe and curve progression was predicted. Accuracy statistics were calculated to evaluate performance. RESULTS: For curve severity, sensitivity and specificity were 95% and 35%, respectively. Negative and positive predictive values were 90% and 53%, respectively, with an accuracy of 61%. For curve progression, sensitivity and specificity were 73% and 44%, respectively. Negative and positive predictive values were 83% and 30%, respectively, with an accuracy of 51%. Assuming that mild and nonprogressive curves would not require an x-ray, the use of ST decision trees could eliminate 31% of x-rays. CONCLUSION: Decision trees showed strong negative predictive values and sensitivity suggesting it may be possible to safely use ST asymmetry analysis with validated decision trees to reduce x-rays in patients with mild and nonprogressive curves. LEVEL OF EVIDENCE: 2.
STUDY DESIGN: A validation study. OBJECTIVE: The aim of this study was to independently validate the diagnostic accuracy of surface topography (ST) classification trees to identify curve severity and progression using a new sample of data in participants with adolescent idiopathic scoliosis (AIS). SUMMARY OF BACKGROUND DATA: Radiographs for diagnosing and monitoring AIS involve harmful radiation exposure repeated at successive clinical visits. Classification trees using a novel ST technique have been proposed to determine curve severity and progression noninvasively that could be used to monitor scoliosis. METHODS: Forty-five adolescents with AIS treated nonoperatively, with ST scans and radiographs at baseline and follow-up (1 year later), were recruited from a scoliosis clinic. The Cobb angle (CA) from radiographs determined curve severity as mild (10° < CA < 25°) or moderate/severe (CA ≥ 25°) and progression as an increase >5°.ST scans were analyzed to calculate the best plane of symmetry and associated deviation color map. Root mean squares and maximum deviation were calculated for each area of asymmetry. ST measurements were analyzed using two published decision trees developed to maximize sensitivity and negative predictive value. Curves were classified as mild or moderate/severe and curve progression was predicted. Accuracy statistics were calculated to evaluate performance. RESULTS: For curve severity, sensitivity and specificity were 95% and 35%, respectively. Negative and positive predictive values were 90% and 53%, respectively, with an accuracy of 61%. For curve progression, sensitivity and specificity were 73% and 44%, respectively. Negative and positive predictive values were 83% and 30%, respectively, with an accuracy of 51%. Assuming that mild and nonprogressive curves would not require an x-ray, the use of ST decision trees could eliminate 31% of x-rays. CONCLUSION: Decision trees showed strong negative predictive values and sensitivity suggesting it may be possible to safely use ST asymmetry analysis with validated decision trees to reduce x-rays in patients with mild and nonprogressive curves. LEVEL OF EVIDENCE: 2.