Texture Analysis of T2-Weighted Images as Reliable Biomarker of Chronic Kidney Disease Microstructural State

Objectives: The diagnostics of chronic kidney disease (CKD) consist of three basic groups of examinations: laboratory tests, radiological imaging and histopathological examinations. However, in the most severe clinical cases, where a fast, undisputed decision is required, histopathological tests are the only suitable option. Unfortunately, such tests require an invasive kidney biopsy, which is not possible in many patients. The aim of this study is to create an algorithm that can categorize CKD patients into active and non-active phases on the basis of MRI texture analysis and compare the results with histopathological examinations. Methods: MRI examinations were performed on healthy volunteers (group 1, N = 14) and CKD patients who also received kidney biopsy. The histopathological examination was used to divide the patients into active phase CKD (group 2, N = 58) and non-active phase CKD (group 3, N = 22). The T2-weighted MRI images were analyzed using a Support Vector Machine (SVM) model created with qMazDa software, which was trained to classify images into the appropriate group of CKD activity. Results: The following evaluation metrics were calculated for the final SVM models corresponding to confusion matrices: for texture analysis—balanced accuracy 81.6%, sensitivity 68.2–92.0%, specificity 82.5–97.5% and precision 62.5–95.8%; for texture and shape analysis—balanced accuracy 87.3%, sensitivity 77.3–100.0%, specificity 87.5–100.0% and precision 65.4–100.0%. Conclusions: Texture analysis of T2-weighted images associated with kidney shape features seems to be reliable method of assessing the state of ongoing CKD.