Applicability Evaluation of Full-Reference Image Quality Assessment Methods for Computed Tomography Images

Area of Science:

• Medical Imaging
• Image Quality Assessment
• Radiology

Background:

• Accurate image quality assessment (IQA) is crucial for medical diagnosis and treatment.
• Full-reference IQA (FR-IQA) methods like PSNR and SSIM, developed for natural images, may not be suitable for medical imaging modalities such as computed tomography (CT).
• Evaluating the applicability of existing FR-IQA methods to CT images is necessary to ensure reliable image quality evaluation.

Purpose of the Study:

• To investigate the correlation between objective FR-IQA methods and subjective human assessments for CT images.
• To determine if standard FR-IQA methods, designed for natural images, can be reliably used for CT image quality evaluation.
• To identify the most effective FR-IQA method for assessing CT image quality compared to subjective evaluations.

Main Methods:

• Generated 210 distorted CT images from six original images using noise and blur degradations.
• Evaluated nine widely used FR-IQA methods (PSNR, SSIM, FSIM, IFC, VIF, NQM, VSNR, MSSSIM, IWSSIM) on the distorted CT images.
• Conducted subjective assessments using the double stimulus continuous quality scale (DSCQS) method with six observers and quantified performance using Pearson's linear correlation coefficient (PLCC), Spearman rank order correlation coefficient (SROCC), and root-mean-square error (RMSE).

Main Results:

• All nine tested FR-IQA methods showed strong correlations (PLCC and SROCC > 0.8) with subjective assessments, indicating their applicability to CT images.
• The Visual Information Fidelity (VIF) method demonstrated the best performance across all three metrics (PLCC, SROCC, and RMSE).
• These findings suggest that VIF is a highly accurate objective measure for CT image quality, comparable to subjective human evaluations.

Conclusions:

• Standard FR-IQA methods are applicable for evaluating the quality of CT images.
• Visual Information Fidelity (VIF) emerges as the most accurate objective IQA metric for CT images, closely aligning with subjective human perception.
• The study validates the use of VIF as a reliable alternative to subjective assessments in CT imaging quality control and algorithm development.