Image denoising methods for tumor discrimination in high-resolution computed tomography.

Area of Science:

• Medical Imaging
• Image Processing
• Quantitative Analysis

Background:

• Pixel accuracy in high-resolution computed tomography (HRCT) is limited by reconstruction error and noise.
• This uncertainty impacts computer-aided quantitative analysis, including densitometric and shape studies.
• Accurate quantitative analysis requires addressing pixel uncertainty in HRCT imaging.

Purpose of the Study:

• To evaluate and compare the performance of different image denoising methods for HRCT.
• To assess the effectiveness of geometric mean filter, Wiener filtering, and wavelet denoising.
• To determine the impact of denoising on pixel accuracy for quantitative analysis.

Main Methods:

• Investigated geometric mean filter, Wiener filtering, and wavelet denoising techniques.
• Assessed method performance using visual inspection and profile region intensity analysis.
• Utilized global figures of merit with brain and thoracic phantoms, and real thoracic HRCT images.

Main Results:

• Comparative analysis of denoising methods was performed on HRCT datasets.
• Quantitative metrics and visual inspection guided the assessment of each method's efficacy.
• Performance varied across methods, with specific findings detailed for phantom and real-world data.

Conclusions:

• Denoising methods significantly impact pixel accuracy in HRCT images.
• Wavelet denoising demonstrated potential for enhancing quantitative analysis in HRCT.
• Careful selection of denoising techniques is essential for reliable computer-aided quantitative studies in medical imaging.