Photon-Counting CT Scan Phantom Study: Stability of Radiomics Features

Area of Science:

• Medical Imaging
• Radiomics
• Quantitative Imaging

Background:

• Radiomics, a quantitative imaging biomarker, shows correlation with survival in oncology patients.
• Clinical implementation of radiomics is hindered by feature instability.
• Photon-counting CT (PCCT) offers higher resolution and reduced artifacts, potentially improving radiomics stability.

Purpose of the Study:

• To evaluate and optimize PCCT reconstruction parameters for stable radiomics signal.
• To identify optimal parameter settings for reproducible radiomics feature extraction.

Main Methods:

• Used two organic phantoms (sponge, sausage) with a PCCT scanner.
• Acquired ultra-high-resolution images with varying reconstruction parameters (kernel, QIR, keV, matrix size).
• Extracted 93 radiomic features and assessed stability using coefficient of variation (CV < 15%).

Main Results:

• Five optimal reconstruction parameter sets were selected based on image quality.
• Feature stability varied with reconstruction parameters and phantom type.
• The most stable configuration (Br36, QIR 4, 60 keV, 1024x1024 matrix) yielded 90 stable features for sausage and 65 for sponge.

Conclusions:

• PCCT imaging demonstrates potential for stable radiomics feature extraction.
• Optimal parameter configurations can enhance radiomics reproducibility for clinical application.
• This study provides a foundation for integrating PCCT-based radiomics into routine oncology practice.