Trimean Filtering for Accelerated Convergence in Ultrasound Tomography Using Diffraction-Based Imaging Technique

Ultrasound computed tomography (USCT) is a valuable tool in medical diagnostics, used widely to assess various diseases. However, USCT images often suffer from speckle noise, which diminishes image quality and complicates diagnosis. The trimean filter outperforms other filters by combining mean, median, and quartile information, providing greater robustness against outliers while preserving essential details. Therefore, in this study, we explore the use of the trimean-filtered Distorted Born Iterative Method (trimean-filtered DBIM) to improve USCT image quality. Compared to standard DBIM and medianfiltered DBIM, the trimean-filtered DBIM demonstrates superior performance in both noise reduction and convergence speed. Across cases with different pixel counts (400, 625, and 900), the trimean-filtered method consistently shows faster convergence and lower normalized error values, indicating a more accurate reconstruction in fewer iterations. By the last iteration, the trimean-filtered DBIM achieves clearer images with less noise, highlighting essential details effectively. This filtering approach offers a promising enhancement in USCT imaging, addressing the challenges of noise and resolution. These improvements can support more reliable diagnoses, making the trimean-filtered DBIM a valuable method in biomedical imaging applications.