An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation
Summary: Background: Assessment of spine alignment is crucial in the management of scoliosis, but current auto-analysis of spine alignment suffers from low accuracy. We aim to develop and validate a hybrid model named SpineHRNet+, which integrates artificial intelligence (AI) and rule-based methods...
Main Authors: | , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2022-01-01
|
Series: | EClinicalMedicine |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2589537021005332 |
_version_ | 1798034129956110336 |
---|---|
author | Nan Meng Jason P.Y. Cheung Kwan-Yee K. Wong Socrates Dokos Sofia Li Richard W. Choy Samuel To Ricardo J. Li Teng Zhang |
author_facet | Nan Meng Jason P.Y. Cheung Kwan-Yee K. Wong Socrates Dokos Sofia Li Richard W. Choy Samuel To Ricardo J. Li Teng Zhang |
author_sort | Nan Meng |
collection | DOAJ |
description | Summary: Background: Assessment of spine alignment is crucial in the management of scoliosis, but current auto-analysis of spine alignment suffers from low accuracy. We aim to develop and validate a hybrid model named SpineHRNet+, which integrates artificial intelligence (AI) and rule-based methods to improve auto-alignment reliability and interpretability. Methods: From December 2019 to November 2020, 1,542 consecutive patients with scoliosis attending two local scoliosis clinics (The Duchess of Kent Children's Hospital at Sandy Bay in Hong Kong; Queen Mary Hospital in Pok Fu Lam on Hong Kong Island) were recruited. The biplanar radiographs of each patient were collected with our medical machine EOS™. The collected radiographs were recaptured using smartphones or screenshots, with deidentified images securely stored. Manually labelled landmarks and alignment parameters by a spine surgeon were considered as ground truth (GT). The data were split 8:2 to train and internally test SpineHRNet+, respectively. This was followed by a prospective validation on another 337 patients. Quantitative analyses of landmark predictions were conducted, and reliabilities of auto-alignment were assessed using linear regression and Bland-Altman plots. Deformity severity and sagittal abnormality classifications were evaluated by confusion matrices. Findings: SpineHRNet+ achieved accurate landmark detection with mean Euclidean distance errors of 2·78 and 5·52 pixels on posteroanterior and lateral radiographs, respectively. The mean angle errors between predictions and GT were 3·18° and 6·32° coronally and sagittally. All predicted alignments were strongly correlated with GT (p < 0·001, R2 > 0·97), with minimal overall difference visualised via Bland-Altman plots. For curve detections, 95·7% sensitivity and 88·1% specificity was achieved, and for severity classification, 88·6–90·8% sensitivity was obtained. For sagittal abnormalities, greater than 85·2–88·9% specificity and sensitivity were achieved. Interpretation: The auto-analysis provided by SpineHRNet+ was reliable and continuous and it might offer the potential to assist clinical work and facilitate large-scale clinical studies. Funding: RGC Research Impact Fund (R5017–18F), Innovation and Technology Fund (ITS/404/18), and the AOSpine East Asia Fund (AOSEA(R)2019–06). |
first_indexed | 2024-04-11T20:39:59Z |
format | Article |
id | doaj.art-91d2f52454494951a6898619f9dfc4ed |
institution | Directory Open Access Journal |
issn | 2589-5370 |
language | English |
last_indexed | 2024-04-11T20:39:59Z |
publishDate | 2022-01-01 |
publisher | Elsevier |
record_format | Article |
series | EClinicalMedicine |
spelling | doaj.art-91d2f52454494951a6898619f9dfc4ed2022-12-22T04:04:14ZengElsevierEClinicalMedicine2589-53702022-01-0143101252An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validationNan Meng0Jason P.Y. Cheung1Kwan-Yee K. Wong2Socrates Dokos3Sofia Li4Richard W. Choy5Samuel To6Ricardo J. Li7Teng Zhang8Digital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, ChinaDigital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, China; Corresponding authors.Department of Computer Science, The University of Hong Kong, Pokfulam, Hong Kong, ChinaGraduate School of Biomedical Engineering, University of New South Wales, Sydney, AustraliaDigital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, ChinaDigital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, ChinaDigital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, ChinaDigital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, ChinaDigital Health Laboratory, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, The University of Hong Kong, 5/F, Professorial Block, Pokfulam, Hong Kong, China; Corresponding authors.Summary: Background: Assessment of spine alignment is crucial in the management of scoliosis, but current auto-analysis of spine alignment suffers from low accuracy. We aim to develop and validate a hybrid model named SpineHRNet+, which integrates artificial intelligence (AI) and rule-based methods to improve auto-alignment reliability and interpretability. Methods: From December 2019 to November 2020, 1,542 consecutive patients with scoliosis attending two local scoliosis clinics (The Duchess of Kent Children's Hospital at Sandy Bay in Hong Kong; Queen Mary Hospital in Pok Fu Lam on Hong Kong Island) were recruited. The biplanar radiographs of each patient were collected with our medical machine EOS™. The collected radiographs were recaptured using smartphones or screenshots, with deidentified images securely stored. Manually labelled landmarks and alignment parameters by a spine surgeon were considered as ground truth (GT). The data were split 8:2 to train and internally test SpineHRNet+, respectively. This was followed by a prospective validation on another 337 patients. Quantitative analyses of landmark predictions were conducted, and reliabilities of auto-alignment were assessed using linear regression and Bland-Altman plots. Deformity severity and sagittal abnormality classifications were evaluated by confusion matrices. Findings: SpineHRNet+ achieved accurate landmark detection with mean Euclidean distance errors of 2·78 and 5·52 pixels on posteroanterior and lateral radiographs, respectively. The mean angle errors between predictions and GT were 3·18° and 6·32° coronally and sagittally. All predicted alignments were strongly correlated with GT (p < 0·001, R2 > 0·97), with minimal overall difference visualised via Bland-Altman plots. For curve detections, 95·7% sensitivity and 88·1% specificity was achieved, and for severity classification, 88·6–90·8% sensitivity was obtained. For sagittal abnormalities, greater than 85·2–88·9% specificity and sensitivity were achieved. Interpretation: The auto-analysis provided by SpineHRNet+ was reliable and continuous and it might offer the potential to assist clinical work and facilitate large-scale clinical studies. Funding: RGC Research Impact Fund (R5017–18F), Innovation and Technology Fund (ITS/404/18), and the AOSpine East Asia Fund (AOSEA(R)2019–06).http://www.sciencedirect.com/science/article/pii/S2589537021005332Adolescent idiopathic scoliosisSpine malalignmentArtificial intelligenceDeep learningOpen platform |
spellingShingle | Nan Meng Jason P.Y. Cheung Kwan-Yee K. Wong Socrates Dokos Sofia Li Richard W. Choy Samuel To Ricardo J. Li Teng Zhang An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation EClinicalMedicine Adolescent idiopathic scoliosis Spine malalignment Artificial intelligence Deep learning Open platform |
title | An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation |
title_full | An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation |
title_fullStr | An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation |
title_full_unstemmed | An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation |
title_short | An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation |
title_sort | artificial intelligence powered platform for auto analyses of spine alignment irrespective of image quality with prospective validation |
topic | Adolescent idiopathic scoliosis Spine malalignment Artificial intelligence Deep learning Open platform |
url | http://www.sciencedirect.com/science/article/pii/S2589537021005332 |
work_keys_str_mv | AT nanmeng anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT jasonpycheung anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT kwanyeekwong anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT socratesdokos anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT sofiali anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT richardwchoy anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT samuelto anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT ricardojli anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT tengzhang anartificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT nanmeng artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT jasonpycheung artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT kwanyeekwong artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT socratesdokos artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT sofiali artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT richardwchoy artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT samuelto artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT ricardojli artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation AT tengzhang artificialintelligencepoweredplatformforautoanalysesofspinealignmentirrespectiveofimagequalitywithprospectivevalidation |