Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach

Purpose: To assess radiomics and deep learning (DL) methods in identifying symptomatic Carotid Artery Disease (CAD) from carotid CT angiography (CTA) images. We further compare the performance of these novel methods to the conventional calcium score. Methods: Carotid CT angiography (CTA) images from...

Full description

Saved in:
Bibliographic Details
Main Authors: Elizabeth P.V. Le, Mark Y.Z. Wong, Leonardo Rundo, Jason M. Tarkin, Nicholas R. Evans, Jonathan R. Weir-McCall, Mohammed M. Chowdhury, Patrick A. Coughlin, Holly Pavey, Fulvio Zaccagna, Chris Wall, Rouchelle Sriranjan, Andrej Corovic, Yuan Huang, Elizabeth A. Warburton, Evis Sala, Michael Roberts, Carola-Bibiane Schönlieb, James H.F. Rudd
Format: Article
Language:English
Published: Elsevier 2024-12-01
Series:European Journal of Radiology Open
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2352047724000492
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850257658874953728
author Elizabeth P.V. Le
Mark Y.Z. Wong
Leonardo Rundo
Jason M. Tarkin
Nicholas R. Evans
Jonathan R. Weir-McCall
Mohammed M. Chowdhury
Patrick A. Coughlin
Holly Pavey
Fulvio Zaccagna
Chris Wall
Rouchelle Sriranjan
Andrej Corovic
Yuan Huang
Elizabeth A. Warburton
Evis Sala
Michael Roberts
Carola-Bibiane Schönlieb
James H.F. Rudd
author_facet Elizabeth P.V. Le
Mark Y.Z. Wong
Leonardo Rundo
Jason M. Tarkin
Nicholas R. Evans
Jonathan R. Weir-McCall
Mohammed M. Chowdhury
Patrick A. Coughlin
Holly Pavey
Fulvio Zaccagna
Chris Wall
Rouchelle Sriranjan
Andrej Corovic
Yuan Huang
Elizabeth A. Warburton
Evis Sala
Michael Roberts
Carola-Bibiane Schönlieb
James H.F. Rudd
author_sort Elizabeth P.V. Le
collection DOAJ
description Purpose: To assess radiomics and deep learning (DL) methods in identifying symptomatic Carotid Artery Disease (CAD) from carotid CT angiography (CTA) images. We further compare the performance of these novel methods to the conventional calcium score. Methods: Carotid CT angiography (CTA) images from symptomatic patients (ischaemic stroke/transient ischaemic attack within the last 3 months) and asymptomatic patients were analysed. Carotid arteries were classified into culprit, non-culprit and asymptomatic. The calcium score was assessed using the Agatston method. 93 radiomic features were extracted from regions-of-interest drawn on 14 consecutive CTA slices. For DL, convolutional neural networks (CNNs) with and without transfer learning were trained directly on CTA slices. Predictive performance was assessed over 5-fold cross validated AUC scores. SHAP and GRAD-CAM algorithms were used for explainability. Results: 132 carotid arteries were analysed (41 culprit, 41 non-culprit, and 50 asymptomatic). For asymptomatic vs symptomatic arteries, radiomics attained a mean AUC of 0.96(± 0.02), followed by DL 0.86(± 0.06) and then calcium 0.79(± 0.08). For culprit vs non-culprit arteries, radiomics achieved a mean AUC of 0.75(± 0.09), followed by DL 0.67(± 0.10) and then calcium 0.60(± 0.02). For multi-class classification, the mean AUCs were 0.95(± 0.07), 0.79(± 0.05), and 0.71(± 0.07) for radiomics, DL and calcium, respectively. Explainability revealed consistent patterns in the most important radiomic features. Conclusions: Our study highlights the potential of novel image analysis techniques in extracting quantitative information beyond calcification in the identification of CAD. Though further work is required, the transition of these novel techniques into clinical practice may eventually facilitate better stroke risk stratification.
format Article
id doaj-art-5f3bf2f517c440329ea8eac866760768
institution OA Journals
issn 2352-0477
language English
publishDate 2024-12-01
publisher Elsevier
record_format Article
series European Journal of Radiology Open
spelling doaj-art-5f3bf2f517c440329ea8eac8667607682025-08-20T01:56:20ZengElsevierEuropean Journal of Radiology Open2352-04772024-12-011310059410.1016/j.ejro.2024.100594Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approachElizabeth P.V. Le0Mark Y.Z. Wong1Leonardo Rundo2Jason M. Tarkin3Nicholas R. Evans4Jonathan R. Weir-McCall5Mohammed M. Chowdhury6Patrick A. Coughlin7Holly Pavey8Fulvio Zaccagna9Chris Wall10Rouchelle Sriranjan11Andrej Corovic12Yuan Huang13Elizabeth A. Warburton14Evis Sala15Michael Roberts16Carola-Bibiane Schönlieb17James H.F. Rudd18Department of Medicine, University of Cambridge, United KingdomDepartment of Medicine, University of Cambridge, United Kingdom; Corresponding author.Department of Radiology, University of Cambridge, United Kingdom; Cancer Research UK Cambridge Centre, University of Cambridge, United Kingdom; Department of Information and Electrical Engineering and Applied Mathematics (DIEM), University of Salerno, ItalyDepartment of Medicine, University of Cambridge, United KingdomDepartment of Clinical Neurosciences, University of Cambridge, United KingdomDepartment of Radiology, University of Cambridge, United Kingdom; Department of Radiology, Royal Papworth Hospital, Cambridge, UKDivision of Vascular Surgery, Department of Surgery, University of Cambridge, United KingdomDepartment of Vascular Surgery, University of Leeds, United KingdomDivision of Experimental Medicine and Immunotherapeutics, University of Cambridge, United KingdomDepartment of Radiology, University of Cambridge, United Kingdom; Department of Imaging, Cambridge University Hospitals NHS Foundation Trust, Cambridge Biomedical Campus, Cambridge, United Kingdom; Investigative Medicine Division, Radcliffe Department of Medicine, University of Oxford, Oxford, United KingdomDepartment of Medicine, University of Cambridge, United KingdomDepartment of Medicine, University of Cambridge, United KingdomDepartment of Medicine, University of Cambridge, United KingdomDepartment of Medicine, University of Cambridge, United Kingdom; Department of Radiology, University of Cambridge, United Kingdom; EPSRC Centre for Mathematical Imaging in Healthcare, University of Cambridge, United KingdomDepartment of Clinical Neurosciences, University of Cambridge, United KingdomDipartimento di Scienze Radiologiche ed Ematologiche, Università Cattolica del Sacro Cuore, Rome, Italy; Dipartimento Diagnostica per Immagini, Radioterapia Oncologica ed Ematologia, Policlinico Universitario A. Gemelli IRCCS, Rome, ItalyDepartment of Medicine, University of Cambridge, United Kingdom; EPSRC Centre for Mathematical Imaging in Healthcare, University of Cambridge, United Kingdom; Department of Applied Mathematics and Theoretical Physics, University of Cambridge, United KingdomEPSRC Centre for Mathematical Imaging in Healthcare, University of Cambridge, United KingdomDepartment of Medicine, University of Cambridge, United Kingdom; EPSRC Centre for Mathematical Imaging in Healthcare, University of Cambridge, United KingdomPurpose: To assess radiomics and deep learning (DL) methods in identifying symptomatic Carotid Artery Disease (CAD) from carotid CT angiography (CTA) images. We further compare the performance of these novel methods to the conventional calcium score. Methods: Carotid CT angiography (CTA) images from symptomatic patients (ischaemic stroke/transient ischaemic attack within the last 3 months) and asymptomatic patients were analysed. Carotid arteries were classified into culprit, non-culprit and asymptomatic. The calcium score was assessed using the Agatston method. 93 radiomic features were extracted from regions-of-interest drawn on 14 consecutive CTA slices. For DL, convolutional neural networks (CNNs) with and without transfer learning were trained directly on CTA slices. Predictive performance was assessed over 5-fold cross validated AUC scores. SHAP and GRAD-CAM algorithms were used for explainability. Results: 132 carotid arteries were analysed (41 culprit, 41 non-culprit, and 50 asymptomatic). For asymptomatic vs symptomatic arteries, radiomics attained a mean AUC of 0.96(± 0.02), followed by DL 0.86(± 0.06) and then calcium 0.79(± 0.08). For culprit vs non-culprit arteries, radiomics achieved a mean AUC of 0.75(± 0.09), followed by DL 0.67(± 0.10) and then calcium 0.60(± 0.02). For multi-class classification, the mean AUCs were 0.95(± 0.07), 0.79(± 0.05), and 0.71(± 0.07) for radiomics, DL and calcium, respectively. Explainability revealed consistent patterns in the most important radiomic features. Conclusions: Our study highlights the potential of novel image analysis techniques in extracting quantitative information beyond calcification in the identification of CAD. Though further work is required, the transition of these novel techniques into clinical practice may eventually facilitate better stroke risk stratification.http://www.sciencedirect.com/science/article/pii/S2352047724000492AIStrokeRadiomicsMachine learningCarotid arteryCoronary calcium
spellingShingle Elizabeth P.V. Le
Mark Y.Z. Wong
Leonardo Rundo
Jason M. Tarkin
Nicholas R. Evans
Jonathan R. Weir-McCall
Mohammed M. Chowdhury
Patrick A. Coughlin
Holly Pavey
Fulvio Zaccagna
Chris Wall
Rouchelle Sriranjan
Andrej Corovic
Yuan Huang
Elizabeth A. Warburton
Evis Sala
Michael Roberts
Carola-Bibiane Schönlieb
James H.F. Rudd
Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach
European Journal of Radiology Open
AI
Stroke
Radiomics
Machine learning
Carotid artery
Coronary calcium
title Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach
title_full Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach
title_fullStr Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach
title_full_unstemmed Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach
title_short Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach
title_sort using machine learning to predict carotid artery symptoms from ct angiography a radiomics and deep learning approach
topic AI
Stroke
Radiomics
Machine learning
Carotid artery
Coronary calcium
url http://www.sciencedirect.com/science/article/pii/S2352047724000492
work_keys_str_mv AT elizabethpvle usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT markyzwong usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT leonardorundo usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT jasonmtarkin usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT nicholasrevans usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT jonathanrweirmccall usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT mohammedmchowdhury usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT patrickacoughlin usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT hollypavey usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT fulviozaccagna usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT chriswall usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT rouchellesriranjan usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT andrejcorovic usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT yuanhuang usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT elizabethawarburton usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT evissala usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT michaelroberts usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT carolabibianeschonlieb usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach
AT jameshfrudd usingmachinelearningtopredictcarotidarterysymptomsfromctangiographyaradiomicsanddeeplearningapproach