eversvd,+3 (12)

Ultrasound technologies have grown popular in the medical field because they are more accurate; however, the image quality of hand-held ultrasound devices is comparably low. This study aims to improve image quality in portable ultrasound devices using Convolutional Neural Networks (CNN). The methodology employs histogram equalization and median filtering to reduce noise while preserving details, thereby enhancing the dynamic range of the images. Additionally, graph balancing techniques are applied to stretch the intensity values of the images, thereby improving their contrast. The technique of unsharp masking is employed to sharpen the images further, emphasizing details crucial for effective diagnosis. CNN, designed primarily for handling pixel data, enables more accurate training and produces high-quality reconstructed images with intricate details and spatial structures. The results indicate that through the application of CNN and other image processing techniques, portable ultrasound devices can achieve enhanced image quality, significantly benefitting the medical imaging field. This advancement has the potential to lead to better diagnostic outcomes and improved patient care.