Textural Analysis and Effect of ROI Size on Infrared Thermography in Athletes With Patellar Tendinopathy

Patellar tendinopathy (PT) is a common knee disorder, particularly among elite athletes, with a reported prevalence of approximately 14.2%. Athletes affected by PT may experience persistent pain, functional impairment, reduced quality of life, decreased physical performance, and even premature career termination. Diagnosing PT remains challenging due to the absence of a gold standard diagnostic method. Although imaging techniques such as ultrasonography (US) and magnetic resonance imaging (MRI) can aid in confirming the diagnosis and assessing severity, MRI is costly and less accessible, and US shows poor correlation with clinical symptoms. Consequently, diagnosis largely relies on clinical examination and medical history. Infrared thermography (IT) has emerged as a potential alternative imaging technique, offering a low-cost, reliable, and non-invasive method to detect thermal asymmetries indicative of underlying pathologies. Technological advancements have enhanced the precision of IT, reducing the thermal asymmetry threshold from 2-3 ºC in the 1970s to 0.5 ºC in current knee assessments. First-order statistics, such as mean gray intensity, and second-order features based on the gray-level co-occurrence matrix (GLCM), have been extensively used in medical image analysis, including IT, to quantify structural and textural characteristics. The size of the region of interest (ROI) is also a critical factor in thermal and texture analyses, as it can influence sensitivity and diagnostic accuracy. Given these considerations, the objectives of this study were: (1) to evaluate differences in thermal and GLCM-based textural features between athletes with PT and healthy controls; (2) to compare the diagnostic performance of IT and GLCM features applied to thermographic images; and (3) to identify the most appropriate ROI size for optimal characterization of PT using both thermal and textural analysis.