Enhancing quality of Low Dose Images using frequency band aware and self-guided network

This research concentrates on overcoming the difficulty in getting paired (Normal-dose computed tomography) NDCT and (Low-dose computed tomography) LDCT by generating LDCT image by reconstructing with sparse angle acquisition using Radon Transform using NDCT images and enhancing the quality of low-dose medical images, crucial for precise diagnosis while minimizing radiation exposure. A novel method using the frequency band aware, self-governed network presents with sharpening images in the frequency space. The approach emphasizes important features in the model and progressively amplifies image properties, thereby causing a drastic diminution of noise and artifacts and, hence yielding better quality than current methods. The paper focuses on issues that involve balancing noise suppression against detail preservation as well as the computational load from the model itself. In a nutshell, the paper shows the viability of the proposed model in enhancing low-dose images and unlocking future development in medical imaging.