A review of eye-tracking technology and its application in stroke diagnosis and assessment

• Published in: Measurement : journal of the International Measurement Confederation Vol. 252; p. 117325
• Main Authors: Zhang, Jun, Kong, Wei, Ma, Ming, Yang, Xi, Li, Weifeng, Song, Aiguo
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.08.2025
• Subjects: Active perception; Eye tracking; Eye-tracking devices; Human-computer interaction; Neurological disorder diagnosis; Stroke assessment; Video-oculography; PAR; XAI; ETT; ETS; LR; I-AC; DL; vHIT; HT; DT; MLP; MC; CAMShift; EMC; LSD; EMG; ML; Neurological disorder diagnosis; biLSTM; EEG; ROC; AI; IR; POG; PSO; SMI; HCI; PCA; VOG; APBA; AdaBoost; TCNs; NB; RF; I-DT; FREDA; VOR; Human-computer interaction; PPTs; VR; AVS; PSOG; ANN; BD; CNN; Stroke assessment; MRI; GBC; KNN; SVM; MNN; TLD; ANOVA; Video-oculography; ASL; GA; EOG; WT; Eye tracking; RBFN; DCNN; NIHSS; SSC; I-VT; iVR; Eye-tracking devices; HMM; BIT; LMM; FIR; RANSAC; CT; USN; HIT; Active perception; OSPDP
• ISSN: 0263-2241
• DOI: 10.1016/j.measurement.2025.117325

The eyes are the windows of the soul, providing essential information through eye movement. With the rapid development of eye-tracking technology (ETT), its application in health assessment, including diagnosing and treating neurological diseases, has expanded significantly. Stroke is a leading cause of adult death and disability worldwide, and studies have shown that eye movement information can serve as a quantitative indicator for stroke diagnosis and assessment. Despite significant research on ETT-based stroke diagnosis, a comprehensive review is still lacking. This paper reviews 238 papers from the past twenty years, focusing on recent advancements in ETT and its application in stroke diagnosis. The studies were selected through a systematic review process following PRISMA-ScR guidelines. This paper provides a systematic overview of ETT principles, methods, and systems, detailing the entire application process in stroke diagnosis and assessment, from data acquisition to symptom evaluation. It also compares various methods and discusses the latest advancements. Statistical analysis methods remain the majority in ETT-based stroke research, while machine learning methods are increasingly attracting attention as alternative approaches. Appearance-based deep learning methods show potential for future stroke diagnosis but require accuracy improvements. Future directions in stroke diagnosis and clinical applications mainly include synthetic data generation, VR integration, wearable ETT devices, multimodal data fusion, and expanding application scenarios. Finally, we propose an active perception strategy and a stroke medical system framework for ETT application. This paper aims to provide researchers with a rapid understanding of core technologies and comprehensive knowledge.