Prediction of successful shock wave lithotripsy with CT: a phantom study using texture analysis

Area of Science:

• Medical Imaging
• Urology
• Computational Analysis

Background:

• Shock wave lithotripsy (SWL) is a common treatment for urinary stones.
• Predicting SWL success is crucial for patient outcomes and resource management.
• Computed tomography (CT) provides detailed imaging of urinary calculi.

Purpose of the Study:

• To evaluate the utility of texture analysis (TA) on CT images of urinary stones.
• To correlate TA features with the number of shockwaves required for successful SWL.
• To establish a non-invasive method for predicting SWL efficacy.

Main Methods:

• CT scans of 34 urinary stones were analyzed in vitro.
• Texture analysis was performed on post-processed CT images.
• Regression models, including linear regression and machine learning (SMOreg), were used to correlate TA features with SWL outcomes.

Main Results:

• Linear regression identified Histogram 10th Percentile and GLCM S(3, 3) SumAverg as significant predictors (r=0.55, p=0.005).
• The SMOreg model achieved a high area under the curve (AUC=0.84) using four TA features, predicting SWL success.
• TA features demonstrated predictive ability for the number of shockwaves needed for stone disintegration.

Conclusions:

• Texture analysis of urinary stone CT images is a viable method for predicting SWL success.
• This study demonstrates the proof-of-principle for using TA to guide SWL treatment.
• TA offers a potential non-invasive tool to optimize urinary stone management.