Quantitative evaluation of the infrapatellar fat pad in knee osteoarthritis: MRI-based radiomic signature

Abstract

BACKGROUND: The infrapatellar fat pad (IFP) may have bilateral influence on knee osteoarthritis (KOA). IFP evaluation may be a key contributor to the diagnostic and clinical management of KOA. Few studies have evaluated KOA-related IFP alteration with radiomics. We investigated radiomic signature for the assessment of IFP for KOA progression in older adults. METHODS: A total of 164 knees were enrolled and grouped based on Kellgren-Lawrence (KL) scoring. MRI-based radiomic features were calculated from IFP segmentation. The radiomic signature was developed using the most predictive subset of features and the machine-learning algorithm with minimum relative standard deviation. KOA severity and structure abnormality were assessed using a modified whole-organ magnetic resonance imaging score (WORMS). The performance of the radiomic signature was evaluated and the correlation with WORMS assessments was analyzed. RESULTS: The area under the curve of the radiomic signature for diagnosing KOA was 0.83 and 0.78 in the training and test datasets, respectively. Rad-scores were 0.41 and 2.01 for the training dataset in the groups with and without KOA (P < 0.001) and 0.63 and 2.31 for the test dataset (P = 0.005), respectively. WORMS significantly and positively correlated with rad-scores. CONCLUSIONS: The radiomic signature may be a reliable biomarker to detect IFP abnormality of KOA. Radiomic alterations in IFP were associated with severity and knee structural abnormalities of KOA in older adults.