Development and validation of a semi-automated measurement tool for calculating consistent and reliable surface metrics describing cosmesis in Adolescent Idiopathic Scoliosis

www.nature.com/scientificreports OPEN Development and validation of a semi‑automated measurement tool for calculating consistent and reliable surface metrics describing cosmesis in Adolescent Idiopathic Scoliosis Sinduja Suresh 1*, Pasan Perera 1, Maree T. Izatt 1,2, Robert D. Labrom 1,2, Geoffrey N. Askin 1,2 & J. Paige Little 1 Adolescent Idiopathic Scoliosis (AIS) is a 3D spine deformity that also causes ribcage and torso distortion. While clinical metrics are important for monitoring disorder progression, patients are often most concerned about their cosmesis. The aim of this study was to automate the quantification of AIS cosmesis metrics, which can be measured reliably from patient-specific 3D surface scans (3DSS). An existing database of 3DSS for pre-operative AIS patients treated at the Queensland Children’s Hospital was used to create 30 calibrated 3D virtual models. A modular generative design algorithm was developed on the Rhino-Grasshopper software to measure five key AIS cosmesis metrics from these models—shoulder, scapula and hip asymmetry, torso rotation and head-pelvis shift. Repeat cosmetic measurements were calculated from user-selected input on the Grasshopper graphical interface. InterClass-correlation (ICC) was used to determine intra- and inter-user reliability. Torso rotation and head-pelvis shift measurements showed excellent reliability (> 0.9), shoulder asymmetry measurements showed good to excellent reliability (> 0.7) and scapula and hip asymmetry measurements showed good to moderate reliability (> 0.5). The ICC results indicated that experience with AIS was not required to reliably measure shoulder asymmetry, torso rotation and headpelvis shift, but was necessary for the other metrics. This new semi-automated workflow reliably characterises external torso deformity, reduces the dependence on manual anatomical landmarking, and does not require bulky/expensive equipment. Adolescent Idiopathic Scoliosis (AIS) is the most common type of s coliosis1 and has been reported to affect 0.93–12% of adolescents (10–18 years old) w orldwide2–4, depending on the region. It is a complex three dimensional spine deformity characterised by a lateral curvature of the spine, loss of normal sagittal curves and is accompanied by a rotation of the spine and r ibcage5–7. This results in an externally observable torso distortion which affects normal posture8. While clinical metrics describing AIS like the Cobb angle9 and Angle of Torso Rotation (ATR) or rib hump10 are important for surveillance, management and treatment of the condition, what is often most important to the patients (and their families) is their cosmesis, or “how they look”11. Cosmetic improvement has been shown to directly influence patient quality of life1213, and correcting the aesthetics of the torso is therefore a prime consideration when treating AIS14. Consequently, significant effort has been invested into developing metrics that characterise the posterior surface of the torso. The quantification of surface topography (ST) in AIS has evolved from the subjective scoring of diagrams and photographs15–22 to objective methods like the Moiré technique23,24, rasterstereography25–31, photogrammetry32–34, 1 Biomechanics and Spine Research Group (BSRG), Centre for Biomedical Technologies (CBT) at the Centre for Children’s Health Research (CCHR), School of Mechanical Medical and Process Engineering, Queensland University of Technology, Brisbane, Australia. 2Orthopaedics Department, Queensland Children’s Hospital (QCH), Brisbane, Australia. *email: Scientific Reports | (2023) 13:5574 | https://doi.org/10.1038/s41598-023-32614-4 1 Vol.:(0123456789) www.nature.com/scientificreports/ depth sensors35,36 and infrared t hermography37 that generate cosmesis metrics and indices. Limitations of current methods include complicated patient positioning and the use of ambiguous anatomical landmarks which are the major sources of measurement errors38. Further, advanced instruments for surface topography measurements are either too bulky to install, too expensive to purchase and maintain, require trained technicians to operate, involve indirect/complex/specific interpretations, or all the above38–40. It is important to note that while an increase in measurement accuracy is always desired, not every clinic or hospital can easily adopt these technologies into regular practice. A review of existing literature highlights the need for a 360° 3D cosmesis assessment method that is simple, accurate, reliable, automation friendly and easily translatable into the clinical pathway. As a step in this direction, 3D surface scanning (3DSS) has garnered significant interest within the scoliosis community in recent years. Accuracy, reliability and validity have been investigated for full body s canners41–43, multi-component scanning systems44–51 and handheld scanners52–54 and the resultant reconstructed 3D Computer Aided Design (CAD) models have begun to be used to assess AIS cosmesis. Of these, portable handheld scanners are of particular interest as they have a high geometric a ccuracy53 and can potentially be widely and easily adopted into regular clinical practice and telehealth initiatives. So far, 3DSS data from handheld scanners have predominantly been used to perform markerless symmetry analysis of the torso. This is done by estimating a sagittal plane of symmetry on the torso, separating the halves, creating mirrored models, and then performing a 3D deviation analysis on the superimposed original and mirrored halves46–48,52. While markerless symmetry analysis is an adequate method for identifying and tracking overall AIS progression, there is still scope for the quantification of individual metrics of spine deformity that are currently mostly visually assessed by treating clinicians. This study aims to make a major contribution to this new and growing area of AIS research by using 3DSS to make reliable objective measurements of five key AIS cosmetic features. To this end, we have developed a user-friendly semi-automated measurement tool to standardise and streamline this process. The reliability study presented here tests the agility of the tool and ascertains how much experience is required to use it effectively to assess trunk deformity parameters in AIS. Methods Preliminary data processing. An existing database of 3DSS data for AIS patients treated at the Queensland Children’s Hospital Spine Clinic (QCHSC) was used to create 30 calibrated pre-operative patient-specific 3D virtual models. The dataset included patients with Lenke curve types (type 1 = 17, type 2 = 1, type 3 = 5, type 4 = 1, type 5 = 5, type 6 = 1) and both sexes (male = 2, female = 28). The mean age was 14.3 years (range 11.3–17 years), mean Body Mass Index (BMI) was 18.08 kg/m2 (range 14.4–25.6 kg/m2), mean Cobb angle of the major curve was 63° (range 38–105°) and mean Angle of Torso Rotation (ATR) of the major curve was 20° (range 12–38°). (...truncated)