complex wavelet transform with progressive network for medical imaging super resolution

This research provides a new method of improving the resolution of medical images by employing a correlation filter in conjunction with a progressive dilated convolution network (PDCNCF). In an effort to enhance automatic identification and segmentation, the technique targets problems with blur, noise, and low texture quality in medical images. The model uses a combination of correlation filtering and dilated convolution to increase the receptive field while preserving pixel information, drawing inspiration from deep learning's capacity to preserve perceived quality. DRIVE, CHASEB1, MRI, ultrasound, and histopathological datasets at 2x, 4x, and 8 × upscaling are among the medical datasets and upscaling factors that the model has been tested on. On the histopathological dataset, the model obtains strong PSNR (49.74) and SSIM (0.9860) at 2 × upscaling. The suggested method outperforms current networks and demonstrates greater performance in medical image reconstruction by combining the complicated wavelet transform, correlation filter, and progressive dilated convolution network. This model has promise for improving medical diagnosis and therapy because of its ability to preserve perceptual quality and enhance resolution.