An Effective Recurrent Neural Network For Image Denoising With Long Short-Term Memory Founded Batch Normalization

Abstract

These days, a high-quality medical image is essential for doctors and radiologists to provide a quick and accurate diagnosis. Noise is a common issue with medical images, both during their procurement and program, reaction, storing, and reclamation. The integrity of the medical imaging is compromised by noise. Edges, structural features, boundary sharpness, etc., are all lost in the degrading process. Researchers still face difficulties in solving the challenge of removing noise from the original image. The vast majority of researchers have developed efficient noise-reduction techniques. Denoising and restoring images is one of the most studied and fundamental problems in computer science. This is the first effort of its kind to introduce a RNN (Recurrent Neural Network) with LSTM (long short-term memory) -based batch normalization for medical image denoising. We start with noisy CT lung pictures as our input. The RNN is used to remove noise from the input image. It has become increasingly common to use batch normalization to rapidly train deep feed-forward neural networks. Batch normalization is used in the LSTM, which proves that it can hasten optimization and enhance generalization. The PSO (Particle Swarm Optimization) algorithm is used to determine the best batch size for batch normalization. MATLAB was used to create the proposed system. Experiment results are compared to those of the current system and shown to be superior.