The Effect of Ultrasound Image Pre-Processing on Radiomics Feature Quality: A Study on Shoulder Ultrasound

Area of Science:

• Medical Imaging
• Radiomics
• Ultrasound Technology

Background:

• Radiomics shows promise in CT/MRI but is underexplored in musculoskeletal ultrasound.
• Ultrasound's inherent variability challenges radiomic feature reproducibility.
• Standardized pre-processing is crucial for reliable radiomics in ultrasound.

Purpose of the Study:

• To evaluate if image pre-processing enhances radiomic feature reproducibility in ultrasound.
• To assess the impact of different pre-processing techniques on feature reliability.
• To improve quantitative analysis quality in musculoskeletal ultrasound.

Main Methods:

• Retrospective analysis of 84 shoulder calcific tendinopathy ultrasound images.
• Application of Histogram Equalization, Standard CLAHE, and Advanced CLAHE pre-processing.
• Extraction of 849 radiomic features and assessment using intraclass correlation coefficient (ICC).

Main Results:

• Advanced CLAHE consistently yielded the highest ICC values, indicating superior reproducibility.
• Wavelet-transformed features (GLCM, GLRLM) showed robust reproducibility across methods.
• Shape features demonstrated lower reproducibility compared to texture-based features.

Conclusions:

• Advanced CLAHE pre-processing significantly enhances radiomic feature reproducibility in ultrasound.
• Pre-processing is vital for reliable radiomic analysis in operator-dependent ultrasound.
• This study provides a foundation for advanced radiomics in musculoskeletal ultrasound applications.