Noise and noise texture in CT images before and after post-processing

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Computed tomography (CT) imaging is susceptible to noise, which can degrade image quality and hinder diagnostic accuracy.
• Convolution filters are commonly used in CT image processing to reduce noise and enhance image features.

Purpose of the Study:

• To evaluate the noise-reducing effect of various convolution filters on CT images.
• To assess the impact of filtering on noise texture and spatial resolution.
• To determine the potential for improved lesion detection using filtered CT images.

Main Methods:

• Acquisition of CT data from homogeneous and inhomogeneous water phantoms under varied scanner settings.
• Application of different convolution filters available in the Ohio-Nuclear Delta 50 FS tomograph software.
• Analysis of noise levels using Root Mean Square Deviation (RMSD) and noise texture via autocovariance matrix.

Main Results:

• Convolution filters reduced noise by a factor of approximately 8 (RMSD).
• A minor sacrifice in spatial resolution was observed with noise reduction.
• Preliminary experiments with an inhomogeneous phantom showed improved detection of smaller lesions after filtering.

Conclusions:

• CT image noise can be effectively reduced using available convolution filters, enhancing diagnostic capabilities.
• The trade-off between noise reduction and spatial resolution needs careful consideration for optimal imaging.
• Filtered CT images show promise for detecting subtle pathologies, such as small lesions.