Prediction of Benign and Malignant Solid Renal Masses: Machine Learning-Based CT Texture Analysis

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Distinguishing benign from malignant renal masses is crucial for appropriate patient management.
• Computed tomography (CT) texture analysis offers quantitative imaging biomarkers.
• Machine learning (ML) algorithms can process complex image data for classification tasks.

Purpose of the Study:

• To evaluate the efficacy of ML-based CT texture analysis in differentiating benign and malignant solid renal masses.
• To identify optimal texture features for classification of renal lesions.

Main Methods:

• Retrospective analysis of 84 solid renal masses (21 benign, 63 malignant) using CT.
• Extraction of 271 texture features from unenhanced and contrast-enhanced CT images.
• Application of eight ML algorithms, including random forest, for classification after feature selection and dimension reduction.

Main Results:

• High reproducibility of texture features was observed (198 unenhanced, 244 contrast-enhanced).
• The random forest algorithm achieved the highest performance.
• Accuracy reached 91.7% and AUC 0.916 using selected contrast-enhanced CT texture features.

Conclusions:

• ML-based contrast-enhanced CT texture analysis demonstrates satisfactory performance for distinguishing benign and malignant solid renal masses.
• This technique holds potential as a non-invasive tool for renal mass characterization.