1
|
Goldman SN, Hui AT, Choi S, Mbamalu EK, Tirabady P, Eleswarapu AS, Gomez JA, Alvandi LM, Fornari ED. Applications of artificial intelligence for adolescent idiopathic scoliosis: mapping the evidence. Spine Deform 2024; 12:1545-1570. [PMID: 39153073 PMCID: PMC11499369 DOI: 10.1007/s43390-024-00940-w] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 10/21/2023] [Accepted: 07/13/2024] [Indexed: 08/19/2024]
Abstract
PURPOSE Adolescent idiopathic scoliosis (AIS) is a common spinal deformity with varying progression, complicating treatment decisions. Artificial intelligence (AI) and machine learning (ML) are increasingly prominent in orthopedic care, aiding in diagnosis, risk-stratification, and treatment guidance. This scoping review outlines AI applications in AIS. METHODS This study followed PRISMA-ScR guidelines and included articles that reported the development, use, or validation of AI models for treating, diagnosing, or predicting clinical outcomes in AIS. RESULTS 40 full-text articles were included, with most studies published in the last 5 years (77.5%). Common ML techniques were convolutional neural networks (55%), decision trees and random forests (15%), and artificial neural networks (15%). Most AI applications in AIS were for imaging analysis (25/40; 62.5%), focusing on automatic measurement of Cobb angle, and axial vertebral rotation (13/25; 52%) and curve classification/severity (13/25; 52%). Prediction was the second most common application (15/40; 37.5%), with studies predicting curve progression (9/15; 60%), and Cobb angles (9/15; 60%). Only 15 studies (37.5%) reported clinical implementation guidelines for AI in AIS management. 52.5% of studies reported model accuracy, with an average of 85.4%. CONCLUSION This review highlights the applications of AI in AIS care, notably including automatic radiographic analysis, curve type classification, prediction of curve progression, and AIS diagnosis. However, the current lack of clear clinical implementation guidelines, model transparency, and external validation of studied models limits clinician trust and the generalizability and applicability of AI in AIS management.
Collapse
Affiliation(s)
- Samuel N Goldman
- Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY, 10461, USA
| | - Aaron T Hui
- Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY, 10461, USA.
| | - Sharlene Choi
- Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY, 10461, USA
| | - Emmanuel K Mbamalu
- Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY, 10461, USA
| | - Parsa Tirabady
- Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY, 10461, USA
| | - Ananth S Eleswarapu
- Department of Orthopaedics, Montefiore Medical Center, Bronx, NY, 10461, USA
| | - Jaime A Gomez
- Department of Orthopaedics, Montefiore Medical Center, Bronx, NY, 10461, USA
| | - Leila M Alvandi
- Department of Orthopaedics, Montefiore Medical Center, Bronx, NY, 10461, USA
| | - Eric D Fornari
- Department of Orthopaedics, Montefiore Medical Center, Bronx, NY, 10461, USA
| |
Collapse
|
2
|
Li L, Wong MS. The application of machine learning methods for predicting the progression of adolescent idiopathic scoliosis: a systematic review. Biomed Eng Online 2024; 23:80. [PMID: 39118179 PMCID: PMC11308564 DOI: 10.1186/s12938-024-01272-6] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/22/2024] [Accepted: 07/22/2024] [Indexed: 08/10/2024] Open
Abstract
Predicting curve progression during the initial visit is pivotal in the disease management of patients with adolescent idiopathic scoliosis (AIS)-identifying patients at high risk of progression is essential for timely and proactive interventions. Both radiological and clinical factors have been investigated as predictors of curve progression. With the evolution of machine learning technologies, the integration of multidimensional information now enables precise predictions of curve progression. This review focuses on the application of machine learning methods to predict AIS curve progression, analyzing 15 selected studies that utilize various machine learning models and the risk factors employed for predictions. Key findings indicate that machine learning models can provide higher precision in predictions compared to traditional methods, and their implementation could lead to more personalized patient management. However, due to the model interpretability and data complexity, more comprehensive and multi-center studies are needed to transition from research to clinical practice.
Collapse
Affiliation(s)
- Lening Li
- Department of Biomedical Engineering, The Hong Kong Polytechnic University, Hong Kong, China.
| | - Man-Sang Wong
- Department of Biomedical Engineering, The Hong Kong Polytechnic University, Hong Kong, China
| |
Collapse
|
3
|
Samadi B, Raison M, Mahaudens P, Detrembleur C, Achiche S. A preliminary study in classification of the severity of spine deformation in adolescents with lumbar/thoracolumbar idiopathic scoliosis using machine learning algorithms based on lumbosacral joint efforts during gait. Comput Methods Biomech Biomed Engin 2023; 26:1341-1352. [PMID: 36093771 DOI: 10.1080/10255842.2022.2117547] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/08/2021] [Revised: 07/07/2022] [Accepted: 08/22/2022] [Indexed: 11/03/2022]
Abstract
To assess the severity and progression of adolescents with idiopathic scoliosis (AIS), radiography with X-rays is usually used. The methods based on statistical observations have been developed from 3D reconstruction of the trunk or topography. Machine learning has shown great potential to classify the severity of scoliosis on imaging data, generally on X-ray measurements. It is also known that AIS leads to the development of gait disorder. To our knowledge, machine learning has never been tested on spine intervertebral efforts during gait as a radiation-free method to classify the severity of spinal deformity in AIS. Develop automated machine learning algorithms in lumbar/thoracolumbar scoliosis to classify the severity of spinal deformity of AIS based on the lumbosacral joint (L5-S1) efforts during gait. The lumbosacral joint efforts of 30 individuals with lumbar/thoracolumbar AIS were used as distinctive features fed to the machine learning algorithms. Several tests were run using various classification algorithms. The labeling consisted of three classes reflecting the severity of scoliosis i.e. mild, moderate and severe. The ensemble classifier algorithm including k-nearest neighbors, support vector machine, random forest and multilayer perceptron achieved the most promising results, with accuracy scores of 91.4%. This preliminary study shows lumbosacral joint efforts can be used to classify the severity of spinal deformity in lumbar/thoracolumbar AIS. This method showed the potential of being used as an assessment tool to follow-up the progression of AIS as a radiation-free method, alternative to radiography. Future studies should be performed to test the method on other categories of AIS.
Collapse
Affiliation(s)
- Bahare Samadi
- Department of Mechanical Engineering, Polytechnique Montréal, Montreal, QC, Canada
- Technopole in Pediatric Rehabilitation Engineering, Sainte-Justine UHC, Montreal, Canada
| | - Maxime Raison
- Department of Mechanical Engineering, Polytechnique Montréal, Montreal, QC, Canada
- Technopole in Pediatric Rehabilitation Engineering, Sainte-Justine UHC, Montreal, Canada
| | - Philippe Mahaudens
- Service d'orthopédie et de traumatologie de l'appareil locomoteur, Cliniques universitaires Saint-Luc, Brussels, Belgium
- Secteur des Sciences de la Santé, Institut de Recherche Expérimentale et Clinique, Neuro Musculo Skeletal Lab (NMSK), Université catholique de Louvain, Brussels, Belgium
| | - Christine Detrembleur
- Secteur des Sciences de la Santé, Institut de Recherche Expérimentale et Clinique, Neuro Musculo Skeletal Lab (NMSK), Université catholique de Louvain, Brussels, Belgium
| | - Sofiane Achiche
- Department of Mechanical Engineering, Polytechnique Montréal, Montreal, QC, Canada
| |
Collapse
|
4
|
Vitamin D and adolescent idiopathic scoliosis, should we stop the hype? A cross-sectional observational prospective study based on a geometric morphometrics approach. EUROPEAN SPINE JOURNAL : OFFICIAL PUBLICATION OF THE EUROPEAN SPINE SOCIETY, THE EUROPEAN SPINAL DEFORMITY SOCIETY, AND THE EUROPEAN SECTION OF THE CERVICAL SPINE RESEARCH SOCIETY 2023; 32:1132-1139. [PMID: 36764946 PMCID: PMC9918399 DOI: 10.1007/s00586-023-07566-y] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Received: 08/14/2022] [Revised: 12/26/2022] [Accepted: 01/25/2023] [Indexed: 02/12/2023]
Abstract
PURPOSE There is strong evidence supporting the presence of fluctuating asymmetry (FA) in Adolescents with Idiopathic Scoliosis (AIS). Additionally, recent research investigating the relationship between vitamin D and AIS found a relation between them. We hypothesize a negative correlation between FA and vitamin D. METHODS We performed a surface scan of the torso of 53 AIS patients, a blood test to measure vitamin D and the radiographic Cobb angle. A correlation analysis between vitamin D and FA was carried out to test our hypothesis, and a regression of vitamin D on 3D shape was performed to observe shape differences between the vitamin D deficiency and insufficiency groups. RESULTS There was no correlation between vitamin D and FA. We found a strong negative correlation between vitamin D and the Cobb angle only in the premenarche group (n = 7; r = - 0.92). Differences in shape were observed between the deficiency and insufficiency groups, and that differences were related to the width of the torso, but not the rotation or lateral flexion. CONCLUSIONS Our results do not support the massive screening of vitamin D in AIS. Shape analysis revealed differences between the shape of the deficiency and insufficiency groups related to robustness. However, this finding had no relation with the scoliosis characteristics, it just reflected different body composition, and its importance should be explored in future.
Collapse
|
5
|
Rayward L, Pearcy M, Izatt M, Green D, Labrom R, Askin G, Little JP. Predicting spinal column profile from surface topography via 3D non-contact surface scanning. PLoS One 2023; 18:e0282634. [PMID: 36952526 PMCID: PMC10035928 DOI: 10.1371/journal.pone.0282634] [Citation(s) in RCA: 2] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/19/2022] [Accepted: 02/17/2023] [Indexed: 03/25/2023] Open
Abstract
INTRODUCTION 3D Non-Contact surface scanning (3DSS) is used in both biomechanical and clinical studies to capture accurate 3D images of the human torso, and to better understand the shape and posture of the spine-both healthy and pathological. This study sought to determine the efficacy and accuracy of using 3DSS of the posterior torso, to determine the curvature of the spinal column in the lateral lying position. METHODS A cohort of 50 healthy adults underwent 3DSS and Magnetic Resonance Imaging (MRI) to correlate the contours of the external spine surface with the internal spinal column. The correlation analysis was composed of two phases: (1) MRI vertebral points vs MRI external spine surface markers; and (2) MRI external spine surface markers vs 3DSS external spine surface markers. The first phase compared the profiles of fiducial markers (vitamin capsules) adhered to the skin surface over the spinous processes against the coordinates of the spinous processes-assessing the linear distance between the profiles, and similarity of curvature, in the sagittal and coronal planes. The second phase compared 3DSS external spine surface markers with the MRI external spine surface markers in both planes, with further qualitative assessment for postural changes. RESULTS The distance between the MRI vertebral points and MRI external spine surface markers showed strong statistically significant correlation with BMI in both sagittal and coronal planes. Kolmogorov-Smirnov (KS) tests showed similar no significant difference in curvature, k, in almost all participants on both planes. In the second phase, the coronal 3DSS external spine surface profiles were statistically different to the MRI external spine surface markers in 44% of participants. Qualitative assessment showed postural changes between MRI and 3DSS measurements in these participants. CONCLUSION These study findings demonstrate the utility and accuracy of using anatomical landmarks overlaid on the spinous processes, to identify the position of the spinal bones using 3DSS. Using this method, it will be possible to predict the internal spinal curvature from surface topography, provided that the thickness of the overlaying subcutaneous adipose layer is considered, thus enabling postural analysis of spinal shape and curvature to be carried out in biomechanical and clinical studies without the need for radiographic imaging.
Collapse
Affiliation(s)
- Lionel Rayward
- Biomechanics and Spine Research Group, School of Mechanical, Medical and Process Engineering, Queensland University of Technology, Brisbane City, Australia
| | - Mark Pearcy
- Biomechanics and Spine Research Group, School of Mechanical, Medical and Process Engineering, Queensland University of Technology, Brisbane City, Australia
| | - Maree Izatt
- Biomechanics and Spine Research Group, School of Mechanical, Medical and Process Engineering, Queensland University of Technology, Brisbane City, Australia
| | | | - Robert Labrom
- Biomechanics and Spine Research Group, School of Mechanical, Medical and Process Engineering, Queensland University of Technology, Brisbane City, Australia
- Wesley Hospital, Auchenflower, Australia
| | - Geoffrey Askin
- Biomechanics and Spine Research Group, School of Mechanical, Medical and Process Engineering, Queensland University of Technology, Brisbane City, Australia
- Mater Health Services, South Brisbane, Australia
| | - J Paige Little
- Biomechanics and Spine Research Group, School of Mechanical, Medical and Process Engineering, Queensland University of Technology, Brisbane City, Australia
| |
Collapse
|
6
|
Abstract
Abstract
Purpose
Adolescent scoliosis is one of the common pediatric spinal diseases which has a high risk of progression due to the rapid growth of the skeleton during the growing stage therefore needs regular clinical monitoring including X-rays. Because X-rays could lead to ionizing radiation-related health problems, an ionizing radiation-free, non-invasive method is presented here to estimate the degree of scoliosis and to potentially support the medical assessment.
Methods
The radiation-free body scanner provides a 3D surface scan of the torso. A basic 3D structure of the human ribcage and vertebral column was modeled and simulated with computer-aided design software and finite element method calculation. For comparison with X-rays, courses of vertebral columns derived from 3D torso images and 3D models were analyzed with respect to their apex positions and angles.
Results
The methods show good results in the estimation of the apex positions of scoliosis. Strong correlations (R = 0.8924) were found between the apex and Cobb angle from X-rays. Similar correlations (R = 0.8087) was obtained between the apex angles extracted from X-rays and the combination of torso scan images with 3D model simulations. Promising agreement was obtained between the spinal trajectories extracted from X-ray and 3D torso images.
Conclusions
Very strong correlations suggest that the apex angle could potentially be used for scoliosis assessment in follow-up examinations in complement to the Cobb angle. However, further improvements of the methods and tests on a larger number of data set are necessary before their introduction into the clinical application.
Collapse
|
7
|
Investigating Human Torso Asymmetries: An Observational Longitudinal Study of Fluctuating and Directional Asymmetry in the Scoliotic Torso. Symmetry (Basel) 2021. [DOI: 10.3390/sym13101821] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/16/2022] Open
Abstract
The presence of directional and fluctuating asymmetry in adolescent idiopathic scoliosis has not been deeply studied. We aimed to test the presence of both in a scoliosis group and a control group. 24 patients with adolescent idiopathic scoliosis and 24 control subjects were subjected to geometric morphometrics analyses to address our main hypotheses and to make qualitative visualizations of the 3D shape changes in patients with scoliosis. Our results support the hypothesis that both asymmetric traits are present in the scoliosis and control groups, but to a greater degree in patients. A qualitative visualization tool that allows us to measure the impact that directional and fluctuating asymmetry have on the 3D shape of our patients has been developed. Adolescent idiopathic scoliosis is the result of developmental instabilities during growth and the visualization of the 3D shape changes in response to both asymmetric variables has shown different morphological behaviors. Measuring these variables is important, as they can prevent the localization and deformation that is expected to occur during the course of scoliosis in every individual patient and therefore acts as a key clinical finding that may be used in the prognosis of the condition.
Collapse
|
8
|
Rothstock S, Weiss HR, Krueger D, Paul L. Clinical classification of scoliosis patients using machine learning and markerless 3D surface trunk data. Med Biol Eng Comput 2020; 58:2953-2962. [PMID: 33001363 DOI: 10.1007/s11517-020-02258-x] [Citation(s) in RCA: 6] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/07/2020] [Accepted: 08/29/2020] [Indexed: 11/30/2022]
Abstract
Markerless 3D surface topography for scoliosis diagnosis and brace treatment can avoid repeated radiation known from standard X-ray analysis and possible side effects. Combined with the method of torso asymmetry analysis, curve severity and progression can be evaluated with high reliability. In the current study, a machine learning approach was utilised to classify scoliosis patients based on their trunk surface asymmetry pattern. Frontal X-ray and 3D scanning analysis with a clinical classification based on Cobb angle and spinal curve pattern were performed with 50 patients. Similar as in a previous study, each patient's trunk 3D reconstruction was used for an elastic registration of a reference surface mesh with fixed number of vertices. Subsequently, an asymmetry distance map between original and reflected torso was calculated. A fully connected neural network was then utilised to classify patients regarding their Cobb angle (mild, moderate, severe) and an Augmented Lehnert-Schroth (ALS) classification based on their full torso asymmetry distance map. The results reveal a classification success rate of 90% (SE: 80%, SP: 100%) regarding the curve severity (mild vs moderate-severe) and 50-72% regarding the ALS group. Identifying patient curve severity and treatment group was reasonably possible allowing for a decision support during diagnosis and treatment planning. Graphical abstract.
Collapse
Affiliation(s)
- Stephan Rothstock
- Society for the Advancement of Applied Computer Science Berlin, GFaI Gesellschaft zur Förderung angewandter Informatik e. V., Volmerstraße 3, D-12489, Berlin, Germany.
| | - Hans-Rudolf Weiss
- KOOB ScoliTechGmbH & Co KG, Haarbergweg 2, D-55546, Neu Bamberg, Germany
| | - Daniel Krueger
- Society for the Advancement of Applied Computer Science Berlin, GFaI Gesellschaft zur Förderung angewandter Informatik e. V., Volmerstraße 3, D-12489, Berlin, Germany
| | - Lothar Paul
- Society for the Advancement of Applied Computer Science Berlin, GFaI Gesellschaft zur Förderung angewandter Informatik e. V., Volmerstraße 3, D-12489, Berlin, Germany
| |
Collapse
|
9
|
Gardner A, Berryman F, Pynsent P. A cluster analysis describing spine and torso shape in Lenke type 1 adolescent idiopathic scoliosis. EUROPEAN SPINE JOURNAL : OFFICIAL PUBLICATION OF THE EUROPEAN SPINE SOCIETY, THE EUROPEAN SPINAL DEFORMITY SOCIETY, AND THE EUROPEAN SECTION OF THE CERVICAL SPINE RESEARCH SOCIETY 2020; 30:620-627. [DOI: 10.1007/s00586-020-06620-3] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Received: 05/20/2020] [Accepted: 09/27/2020] [Indexed: 10/23/2022]
Abstract
Abstract
Purpose
The purpose of this work is to identify the variability and subtypes of the combined shape of the spine and torso in Lenke type 1 adolescent idiopathic scoliosis (AIS).
Methods
Using ISIS2 surface topography, measures of coronal deformity, kyphosis and skin angulation (as a measure of torso asymmetry) in a series of children with Lenke 1 convex to the right AIS were analyzed using k-means clustering techniques to describe the combined variability of shape in the spine and torso. Following this, a k-nearest neighbor algorithm was used to measure the ability to automatically identify the correct cluster for any particular datum.
Results
There were 1399 ISIS2 images from 691 individuals available for analysis. There were 5 clusters identified in the data representing the variability of the 3 measured parameters which included mild, moderate and marked coronal deformity, mild, moderate and marked asymmetry alongside normal and hypokyphosis. The k-nearest neighbor identification of the correct cluster had an accuracy of 93%.
Conclusion
These clusters represent a new description of Lenke 1 AIS that comprises both coronal and sagittal measures of the spine combined with a measure of torso asymmetry. Automated identification of the clusters is accurate. The ability to identify subtypes of deformity, based on parameters that affect both the spine and the torso in AIS, leads to as better understanding of the totality of the deformity seen.
Collapse
|
10
|
González-Ruiz JM, Pérez-Núñez MI, García-Alfaro MD, Bastir M. Geometric morphometrics of adolescent idiopathic scoliosis: a prospective observational study. EUROPEAN SPINE JOURNAL : OFFICIAL PUBLICATION OF THE EUROPEAN SPINE SOCIETY, THE EUROPEAN SPINAL DEFORMITY SOCIETY, AND THE EUROPEAN SECTION OF THE CERVICAL SPINE RESEARCH SOCIETY 2020; 30:612-619. [DOI: 10.1007/s00586-020-06583-5] [Citation(s) in RCA: 4] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Track Full Text] [Subscribe] [Scholar Register] [Received: 04/23/2020] [Revised: 08/04/2020] [Accepted: 08/25/2020] [Indexed: 01/13/2023]
|
11
|
Rothstock S, Weiss HR, Krueger D, Kleban V, Paul L. Innovative decision support for scoliosis brace therapy based on statistical modelling of markerless 3D trunk surface data. Comput Methods Biomech Biomed Engin 2020; 23:923-933. [PMID: 32543233 DOI: 10.1080/10255842.2020.1773449] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/24/2022]
Abstract
Recently markerless 3D scanning methods receive an increased interest for therapy planning and brace treatment of patients with scoliosis. This avoids repeated radiation known from standard X-Ray analysis. Several authors introduced the method of asymmetry distance maps in order to classify curve severity and progression. The current work extends this approach by statistical mean shape 3D models of the human trunk in order to classify patients. 50 patients were included in this study performing frontal X-ray and 3D scanning analysis. All patients were classified by a clinician according to their Cobb angle and spinal curve pattern (Augmented-Lehnert-Schroth ALS). 3D reconstructions of each patient trunk were processed in a way to elastically register a reference surface mesh with fixed number of data points. Mean 3D shape models were generated for each curve pattern. An asymmetry distance map was then calculated for each patient and mean shape model. Single patient 3D reconstructions were classified according to severity and ALS treatment group. Optimal sensitivity and specificity was 97%/39% thoracic and 87%/42% lumbar respectively for detecting mild and moderate-severe patients. Identifying a treatment group was possible for three combined groups allowing to support decisions during diagnosis and therapy planning.
Collapse
Affiliation(s)
- Stephan Rothstock
- GFaI Gesellschaft zur Förderung angewandter Informatik e. V, Society for the Advancement of Applied Computer Science Berlin, Berlin, Germany
| | | | - Daniel Krueger
- GFaI Gesellschaft zur Förderung angewandter Informatik e. V, Society for the Advancement of Applied Computer Science Berlin, Berlin, Germany
| | - Victoria Kleban
- GFaI Gesellschaft zur Förderung angewandter Informatik e. V, Society for the Advancement of Applied Computer Science Berlin, Berlin, Germany
| | - Lothar Paul
- GFaI Gesellschaft zur Förderung angewandter Informatik e. V, Society for the Advancement of Applied Computer Science Berlin, Berlin, Germany
| |
Collapse
|
12
|
Murtagh P, Greene G, O'Brien C. Current applications of machine learning in the screening and diagnosis of glaucoma: a systematic review and Meta-analysis. Int J Ophthalmol 2020; 13:149-162. [PMID: 31956584 DOI: 10.18240/ijo.2020.01.22] [Citation(s) in RCA: 18] [Impact Index Per Article: 4.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/05/2019] [Accepted: 09/23/2019] [Indexed: 12/22/2022] Open
Abstract
AIM To compare the effectiveness of two well described machine learning modalities, ocular coherence tomography (OCT) and fundal photography, in terms of diagnostic accuracy in the screening and diagnosis of glaucoma. METHODS A systematic search of Embase and PubMed databases was undertaken up to 1st of February 2019. Articles were identified alongside their reference lists and relevant studies were aggregated. A Meta-analysis of diagnostic accuracy in terms of area under the receiver operating curve (AUROC) was performed. For the studies which did not report an AUROC, reported sensitivity and specificity values were combined to create a summary ROC curve which was included in the Meta-analysis. RESULTS A total of 23 studies were deemed suitable for inclusion in the Meta-analysis. This included 10 papers from the OCT cohort and 13 from the fundal photos cohort. Random effects Meta-analysis gave a pooled AUROC of 0.957 (95%CI=0.917 to 0.997) for fundal photos and 0.923 (95%CI=0.889 to 0.957) for the OCT cohort. The slightly higher accuracy of fundal photos methods is likely attributable to the much larger database of images used to train the models (59 788 vs 1743). CONCLUSION No demonstrable difference is shown between the diagnostic accuracy of the two modalities. The ease of access and lower cost associated with fundal photo acquisition make that the more appealing option in terms of screening on a global scale, however further studies need to be undertaken, owing largely to the poor study quality associated with the fundal photography cohort.
Collapse
Affiliation(s)
- Patrick Murtagh
- Department of Ophthalmology, Mater Misericordiae University Hospital, Eccles Street, Dublin D07 R2WY, Ireland
| | - Garrett Greene
- RCSI Education and Research Centre, Beaumont Hospital, Dublin D05 AT88, Ireland
| | - Colm O'Brien
- Department of Ophthalmology, Mater Misericordiae University Hospital, Eccles Street, Dublin D07 R2WY, Ireland
| |
Collapse
|
13
|
Little JP, Rayward L, Pearcy MJ, Izatt MT, Green D, Labrom RD, Askin GN. Predicting spinal profile using 3D non-contact surface scanning: Changes in surface topography as a predictor of internal spinal alignment. PLoS One 2019; 14:e0222453. [PMID: 31557174 PMCID: PMC6762190 DOI: 10.1371/journal.pone.0222453] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/12/2019] [Accepted: 08/29/2019] [Indexed: 01/20/2023] Open
Abstract
INTRODUCTION 3D non-contact surface scanners capture highly accurate, calibrated images of surface topography for 3D structures. This study sought to establish the efficacy and accuracy of using 3D surface scanning to characterise spinal curvature and sagittal plane contour. METHODS 10 healthy female adults with a mean age of 25 years, (standard deviation: 3.6 years) underwent both MRI and 3D surface scanning (3DSS) (Artec Eva, Artec Group Inc., Luxembourg) while lying in the lateral decubitus position on a rigid substrate. Prior to 3DSS, anatomical landmarks on the spinous processes of each participant were demarcated using stickers attached to the skin surface. Following 3DSS, oil capsules (fiducial markers) were overlaid on the stickers and the subject underwent MRI. MRI stacks were processed to measure the thoracolumbar spinous process locations, providing an anatomical reference. 3D coordinates for the markers (surface stickers and MRI oil capsules) and for the spinous processes mapped the spinal column profiles and were compared to assess the quality of fit between the 3DSS and MRI marker positions. RESULTS The RMSE for the polynomials fit to the spinous process, fiducial and surface marker profiles ranged from 0.17-1.15mm for all subjects. The MRI fiducial marker location was well aligned with the spinous process profile in the thoracic and upper lumbar spine for nine of the subjects. Over the 10 subjects, the mean RMSE between the MRI and 3D scan sagittal profiles for all surface markers was 9.8mm (SD 4.2mm). Curvature was well matched for seven of the subjects, with two showing differing curvatures across the lumbar spine due to inconsistent subject positioning. CONCLUSION Comparison of the observed trends for vertebral position measured from MRI and 3DSS, suggested the surface markers may provide a useful method for measuring internal changes in sagittal curvature or skeletal changes.
Collapse
Affiliation(s)
- J. Paige Little
- Biomechanics and Spine Research Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia
| | - Lionel Rayward
- Biomechanics and Spine Research Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia
| | - Mark J. Pearcy
- Biomechanics and Spine Research Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia
| | - Maree T. Izatt
- Biomechanics and Spine Research Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia
| | | | - Robert D. Labrom
- Biomechanics and Spine Research Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia
- Wesley Hospital, Brisbane, Australia
| | - Geoffrey N. Askin
- Biomechanics and Spine Research Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia
- Mater Health Services, Brisbane, Australia
| |
Collapse
|