Entropy-based straight kernel filter for echocardiography image denoising

Area of Science:

• Medical Imaging
• Image Processing
• Biomedical Engineering

Background:

• Speckle noise in 2D echocardiography images can obscure anatomical details, leading to misinterpretations.
• Accurate visualization of cardiac structures is crucial for diagnosis and treatment planning.

Purpose of the Study:

• To develop and evaluate a new filter for speckle noise reduction in 2D echocardiography.
• To improve the clarity of anatomical structures in echocardiographic images.

Main Methods:

• A novel filter utilizing an entropy parameter to quantify pixel disorganization.
• Application of straight 3-pixel kernels slid across the image for noise elimination.
• Quantitative evaluation using Peak Signal-to-Noise Ratio (PSNR) and Root Mean Square Error (RMSE).

Main Results:

• Achieved a high PSNR in the range of 147 dB.
• Demonstrated a very low RMSE of approximately 0.15.
• Successfully implemented on 36 echocardiography images, preserving anatomical edges.

Conclusions:

• The proposed filter effectively eliminates speckle noise in 2D echocardiography.
• The filter enhances image visibility and accurately reveals anatomical facts without edge degradation.
• This method offers a promising approach for improving echocardiographic image quality.