CT-derived textural analysis parameters discriminate high-attenuation renal cysts from solid renal neoplasms.

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Indeterminate renal lesions on CT require further characterization.
• Differentiating benign cysts from solid renal neoplasms is crucial for patient management.

Purpose of the Study:

• To evaluate the effectiveness of CT textural features in distinguishing benign cysts from solid renal lesions.
• To assess the utility of a random forest model incorporating textural and clinical parameters.

Main Methods:

• A random forest model was developed using 95 textural and 4 clinical parameters.
• 234 patients with 278 indeterminate renal lesions were analyzed.
• Lesions were categorized as benign cysts or indeterminate/solid renal lesions.

Main Results:

• The model achieved an area under the curve of 0.71 (95% CI: 0.65, 0.78).
• Sensitivity was 81.2% and specificity was 38.9% for discriminating lesion types.
• 69% of lesions were benign cysts, while 31% were indeterminate or solid.

Conclusions:

• A multivariate model using CT textural and clinical parameters demonstrated moderate performance.
• This study is a proof of concept for characterizing indeterminate renal lesions on CT.
• The findings may potentially reduce the need for additional follow-up imaging.