Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques

<i>Background and Objectives</i>: Thyroid nodules are lumps of solid or liquid-filled tumors that form inside the thyroid gland, which can be malignant or benign. Our aim was to test whether the described features of the Thyroid Imaging Reporting and Data System (TI-RADS) could improve r...

Full description

Bibliographic Details
Main Authors: Vijay Vyas Vadhiraj, Andrew Simpkin, James O’Connell, Naykky Singh Ospina, Spyridoula Maraka, Derek T. O’Keeffe
Format: Article
Language:English
Published: MDPI AG 2021-05-01
Series:Medicina
Subjects:
CAD
AI
Online Access:https://www.mdpi.com/1648-9144/57/6/527
id doaj-35c9f241611248e3a7ad55c779f2e147
record_format Article
spelling doaj-35c9f241611248e3a7ad55c779f2e1472021-06-01T00:59:24ZengMDPI AGMedicina1010-660X1648-91442021-05-015752752710.3390/medicina57060527Ultrasound Image Classification of Thyroid Nodules Using Machine Learning TechniquesVijay Vyas Vadhiraj0Andrew Simpkin1James O’Connell2Naykky Singh Ospina3Spyridoula Maraka4Derek T. O’Keeffe5School of Medicine, College of Medicine Nursing and Health Sciences, National University of Ireland Galway, H91 TK33 Galway, IrelandSchool of Mathematics, Statistics and Applied Maths, National University of Ireland, H91 TK33 Galway, IrelandSchool of Medicine, College of Medicine Nursing and Health Sciences, National University of Ireland Galway, H91 TK33 Galway, IrelandDivision of Endocrinology, Department of Medicine, University of Florida, Gainesville, FL 3210, USADivision of Endocrinology and Metabolism, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USASchool of Medicine, College of Medicine Nursing and Health Sciences, National University of Ireland Galway, H91 TK33 Galway, Ireland<i>Background and Objectives</i>: Thyroid nodules are lumps of solid or liquid-filled tumors that form inside the thyroid gland, which can be malignant or benign. Our aim was to test whether the described features of the Thyroid Imaging Reporting and Data System (TI-RADS) could improve radiologists’ decision making when integrated into a computer system. In this study, we developed a computer-aided diagnosis system integrated into multiple-instance learning (MIL) that would focus on benign–malignant classification. Data were available from the Universidad Nacional de Colombia. <i>Materials and Methods</i>: There were 99 cases (33 Benign and 66 malignant). In this study, the median filter and image binarization were used for image pre-processing and segmentation. The grey level co-occurrence matrix (GLCM) was used to extract seven ultrasound image features. These data were divided into 87% training and 13% validation sets. We compared the support vector machine (SVM) and artificial neural network (ANN) classification algorithms based on their accuracy score, sensitivity, and specificity. The outcome measure was whether the thyroid nodule was benign or malignant. We also developed a graphic user interface (GUI) to display the image features that would help radiologists with decision making. <i>Results</i>: ANN and SVM achieved an accuracy of 75% and 96% respectively. SVM outperformed all the other models on all performance metrics, achieving higher accuracy, sensitivity, and specificity score. <i>Conclusions</i>: Our study suggests promising results from MIL in thyroid cancer detection. Further testing with external data is required before our classification model can be employed in practice.https://www.mdpi.com/1648-9144/57/6/527computer aided diagnosticsCADartificial intelligenceAIdigital healthTI-RADS
collection DOAJ
language English
format Article
sources DOAJ
author Vijay Vyas Vadhiraj
Andrew Simpkin
James O’Connell
Naykky Singh Ospina
Spyridoula Maraka
Derek T. O’Keeffe
spellingShingle Vijay Vyas Vadhiraj
Andrew Simpkin
James O’Connell
Naykky Singh Ospina
Spyridoula Maraka
Derek T. O’Keeffe
Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques
Medicina
computer aided diagnostics
CAD
artificial intelligence
AI
digital health
TI-RADS
author_facet Vijay Vyas Vadhiraj
Andrew Simpkin
James O’Connell
Naykky Singh Ospina
Spyridoula Maraka
Derek T. O’Keeffe
author_sort Vijay Vyas Vadhiraj
title Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques
title_short Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques
title_full Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques
title_fullStr Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques
title_full_unstemmed Ultrasound Image Classification of Thyroid Nodules Using Machine Learning Techniques
title_sort ultrasound image classification of thyroid nodules using machine learning techniques
publisher MDPI AG
series Medicina
issn 1010-660X
1648-9144
publishDate 2021-05-01
description <i>Background and Objectives</i>: Thyroid nodules are lumps of solid or liquid-filled tumors that form inside the thyroid gland, which can be malignant or benign. Our aim was to test whether the described features of the Thyroid Imaging Reporting and Data System (TI-RADS) could improve radiologists’ decision making when integrated into a computer system. In this study, we developed a computer-aided diagnosis system integrated into multiple-instance learning (MIL) that would focus on benign–malignant classification. Data were available from the Universidad Nacional de Colombia. <i>Materials and Methods</i>: There were 99 cases (33 Benign and 66 malignant). In this study, the median filter and image binarization were used for image pre-processing and segmentation. The grey level co-occurrence matrix (GLCM) was used to extract seven ultrasound image features. These data were divided into 87% training and 13% validation sets. We compared the support vector machine (SVM) and artificial neural network (ANN) classification algorithms based on their accuracy score, sensitivity, and specificity. The outcome measure was whether the thyroid nodule was benign or malignant. We also developed a graphic user interface (GUI) to display the image features that would help radiologists with decision making. <i>Results</i>: ANN and SVM achieved an accuracy of 75% and 96% respectively. SVM outperformed all the other models on all performance metrics, achieving higher accuracy, sensitivity, and specificity score. <i>Conclusions</i>: Our study suggests promising results from MIL in thyroid cancer detection. Further testing with external data is required before our classification model can be employed in practice.
topic computer aided diagnostics
CAD
artificial intelligence
AI
digital health
TI-RADS
url https://www.mdpi.com/1648-9144/57/6/527
work_keys_str_mv AT vijayvyasvadhiraj ultrasoundimageclassificationofthyroidnodulesusingmachinelearningtechniques
AT andrewsimpkin ultrasoundimageclassificationofthyroidnodulesusingmachinelearningtechniques
AT jamesoconnell ultrasoundimageclassificationofthyroidnodulesusingmachinelearningtechniques
AT naykkysinghospina ultrasoundimageclassificationofthyroidnodulesusingmachinelearningtechniques
AT spyridoulamaraka ultrasoundimageclassificationofthyroidnodulesusingmachinelearningtechniques
AT derektokeeffe ultrasoundimageclassificationofthyroidnodulesusingmachinelearningtechniques
_version_ 1721413278893080576