The utility of texture analysis of kidney MRI for evaluating renal dysfunction with multiclass classification model

Abstract

We evaluated a multiclass classification model to predict estimated glomerular filtration rate (eGFR) groups in chronic kidney disease (CKD) patients using magnetic resonance imaging (MRI) texture analysis (TA). We identified 166 CKD patients who underwent MRI comprising Dixon-based T1-weighted in-phase (IP)/opposed-phase (OP)/water-only (WO) images, apparent diffusion coefficient (ADC) maps, and T2* maps. The patients were divided into severe, moderate, and control groups based on eGFR borderlines of 30 and 60 mL/min/1.73 m². After extracting 93 texture features (TFs), dimension reduction was performed using inter-observer reproducibility analysis and sequential feature selection (SFS) algorithm. Models were created using linear discriminant analysis (LDA); support vector machine (SVM) with linear, rbf, and sigmoid kernels; decision tree (DT); and random forest (RF) classifiers, with synthetic minority oversampling technique (SMOTE). Models underwent 100-time repeat nested cross-validation. Overall performances of our classification models were modest, and TA based on T1-weighted IP/OP/WO images provided better performance than those based on ADC and T2* maps. The most favorable result was observed in the T1-weighted WO image using RF classifier and the combination model was derived from all T1-weighted images using SVM classifier with rbf kernel. Among the selected TFs, total energy and energy had weak correlations with eGFR.

Figure 1

The receiver operating characteristic (ROC) curves and area under the curve (AUC) values of representative classification models using T1-weighted water-only images with a random forest classifier (A) and all T1-weighted images using a support vector machine with rbf kernel classifier (B) in classifying the three groups of chronic kidney disease. Severe renal dysfunction group (se-RD, estimated glomerular filtration rate [eGFR] < 30 mL/min/1.73 m²), moderate renal dysfunction group (mo-RD, 30 ≤ eGFR < 60 mL/min/1.73 m²), and control group (CG, eGFR ≥ 60 mL/min/1.73 m²). The AUC values are expressed as means.

Figure 2

Confusion matrices show the status of representative classification models using T1-weighted water-only images with a random forest classifier (A) and all T1-weighted images using a support vector machine with rbf kernel classifier (B) in classifying the three groups of chronic kidney disease. Severe renal dysfunction group (se-RD, estimated glomerular filtration rate [eGFR] < 30 mL/min/1.73 m²), moderate renal dysfunction group (mo-RD, 30 ≤ eGFR < 60 mL/min/1.73 m²), and control group (CG, eGFR ≥ 60 mL/min/1.73 m²). The data are expressed as means ± standard deviations.

Figure 3

Flow chart of the inclusion and exclusion criteria for the study. ADC apparent diffusion coefficient, DWI diffusion-weighted imaging, MRI magnetic resonance imaging, T1WI T1-weighted imaging.

Figure 4

Flow chart showing the technical study pipeline. After segmentation, image processing, texture feature extraction, and reproducibility analysis were conducted for each imaging method (T1-weighted in-phase/opposed-phase/water-only images, ADC maps, and T2* maps), followed by texture feature selection and ML-based model construction in separate classification attempts. The combinations of texture features were also examined: those derived from all T1-weighted images and those derived from all imaging methods.

Figure 5

A method to set the region of interest (ROI) for each group and each image. ROIs were manually drawn on the contour lines of both kidneys, as shown by the red curves (avoiding the cystic area). ADC apparent diffusion coefficient, IP in-phase, OP opposed-phase, WO water-only. Severe renal dysfunction group (se-RD, estimated glomerular filtration rate [eGFR] < 30 mL/min/1.73 m²), moderate renal dysfunction group (mo-RD, 30 ≤ eGFR < 60 mL/min/1.73 m²), and control group (CG, eGFR ≥ 60 mL/min/1.73 m²).