Automatic diagnosis of extraocular muscle palsy based on machine learning and diplopia images - Intelligent Ophthalmology

Aug 4, 2026

Upload by Riley

Aim focuses on building and comparing multiple machine learning models using diplopia images and data generated by a computerized diplopia test. Retrospective collection includes 3244 cases, training with 2757 images and testing with 487 images, while diagnostic models are trained using logistic regression, decision tree, support vector machine, extreme gradient boosting, and deep learning. Evaluation uses test accuracy, confusion matrices, and precision-recall curves to identify the best-performing model.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Clinical background and need for automated diagnosis

Participants and Methods

Ethical approval and data collection

Abstract

Aims, methods, results, and conclusion

·Intelligent Ophthalmology · Automatic diagnosis of extraocular muscle palsy based on machine learning and diplopia images Xiao-Lu Jin, Xue-Mei Li, Tie-Juan Liu, Ling-Yun Zhou Ocular Motility Disorder Treatment Centre, the First Affiliated Hospital of Harbin Medical University, Harbin 150001, Heilongjiang Province, China Correspondence to: Ling-Yun Zhou. Ocular Motility Disorder Treatment Centre, the First Affiliated Hospital of Harbin Medical University, 199 Dongdazhi Street, Nangang District, Harbin 150001, Heilongjiang Province, [China. no1zhly@163.com](China. no1zhly@163.com)[ ](China. no1zhly@163.com)[Received: 2024-04-22 Accepted: 2025-01-25](Received: 2024-04-22 Accepted: 2025-01-25) Abstract ● AIM: To develop different machine learning models to train and test diplopia images and data generated by the computerized diplopia test. ● METHODS: Diplopia images and data generated by computerized diplopia tests, along with patient medical records, were retrospectively collected from 3244 cases. Diagnostic models were constructed using logistic regression (LR), decision tree (DT), support vector machine (SVM), extreme gradient boosting (XGBoost), and deep learning (DL) algorithms. A total of 2757 diplopia images were randomly selected as training data, while the test dataset contained 487 diplopia images. The optimal diagnostic model was evaluated using test set accuracy, confusion matrix, and precision-recall curve (P-R curve) .● RESULTS: The test set accuracy of the LR, SVM, DT, XGBoost, DL (64 categories), and DL (6 binary classifications) algorithms was 0.762, 0.811, 0.818, 0. 812, 0.858 and 0. 858, respectively. The accuracy in the training set was 0.785, 0.815, 0.998, 0.965, 0.968, and 0.967, respectively. The weighted precision of LR, SVM, DT, XGBoost, DL (64 categories), and DL (6 binary classifications) algorithms was 0.74, 0.77, 0.83, 0.80, 0.85, and 0.85, respectively; weighted recall was 0.76, 0.81, 0.82, 0.81, 0.86, and 0.86, respectively; weighted F1 score was 0. 74, 0. 79, 0.82, 0.80, 0.85, and 0.85, respectively.● CONCLUSION: In this study, the 7 machine learning algorithms all achieve automatic diagnosis of extraocular muscle palsy. The DL (64 categories) and DL (6 binary classifications) algorithms have a significant advantage over other machine learning algorithms regarding diagnostic accuracy on the test set, with a high level of consistency with clinical diagnoses made by physicians. Therefore, it can be used as a reference for diagnosis. ● KEYWORDS: machine learning; extraocular muscle paralysis; automatic diagnosis; diplopia images DOI:10.18240/ijo.2025.05.01 Citation: Jin XL, Li XM, Liu TJ, Zhou LY. Automatic diagnosis of extraocular muscle palsy based on machine learning and diplopia images. Int JOphthalmol 2025;18(5):757-764 INTRODUCTION E xtraocular muscle palsy is caused by vascular diseases, cranial nerve injuries, infections, and other factors affecting the ocular motor nerve system or extraocular muscles themselves, resulting in complete or partial dysfunction of ≥1 extraocular muscle. This can lead to diplopia, restricted eye movement, and strabismus [1-3] . Hess screen test is a common clinical examination method used to distinguish and diagnose extraocular muscle palsy by breaking binocular visual fusion[4-5] . The computerized diplopia test, a computerautomated detection device based on the traditional Hess screen principle, showed accurate and reliable results in the clinic[6-7] . The diplopia image is the plot generated by the computerized diplopia test, reflecting the image perceived by the patient during the gaze test[6](Figure 1) . However, the shapes of diplopia images can be complex and vary depending on which extraocular muscles are paralyzed. Therefore, a professionally trained doctor need to manually interpret diplopia images to diagnose the paralyzed extraocular muscles. However, this requirement for specialized expertise and manual interpretation is not

FAQs

What was the study’s main goal?

To develop and compare several machine learning models for automatic diagnosis of extraocular muscle palsy using diplopia images and computerized diplopia test data.

How were the data and models prepared?

Diplopia images and associated data were retrospectively collected from 3244 cases, with 2757 images used for training and 487 for testing. Models were built using LR, DT, SVM, XGBoost, and deep learning approaches.

Which models performed best and how was performance evaluated?

Deep learning models (64 categories and 6 binary classifications) achieved the highest test-set accuracy, evaluated using test accuracy, confusion matrices, and precision-recall curves. Weighted precision, recall, and F1 scores were also reported to support consistency with clinical diagnoses.

Aug 4, 2026

Upload by Riley