Automated measurement of detectability index in CT imaging: Development and validation

Area of Science:

• Medical Imaging Physics
• Computational Imaging
• Radiology

Background:

• Accurate assessment of image quality is crucial in computed tomography (CT) for diagnostic performance.
• The detectability index (d') is a key metric for quantifying image quality and lesion detectability.
• Automating the measurement of d' can streamline quality control and research workflows.

Purpose of the Study:

• To develop and validate software for the automated measurement of the detectability index (d') using ACR 464 CT phantom images.
• To integrate this software into the IndoQCT platform for enhanced CT image analysis.

Main Methods:

• Developed automated software using Python 3.9.13 and PyQt5 GUI for d' measurement.
• Measured task-transfer function (TTF) and noise power spectrum (NPS) to derive spatial resolution and noise texture.
• Evaluated software performance by comparing its d' measurements against ImQuest results across various CT parameters (tube current, kernel type, object diameter, contrast).

Main Results:

• The developed software demonstrated strong agreement (r > 0.98) with ImQuest for d' measurements across all tested variations.
• Increased tube current, larger object diameters, and higher object contrasts consistently improved the d' values.
• The Standard kernel type yielded the highest detectability, while the Lung kernel yielded the lowest.

Conclusions:

• Successfully developed user-friendly software for automated d' measurement in CT images.
• The software provides a fast, accurate, and intuitive workflow for CT image quality assessment.
• The strong correlation with ImQuest validates the software's reliability for clinical and research applications.