Kalman Bucy Filtered Neuro Fuzzy Image Denoising for Medical Image Processing

Abstract

 Neutrosophic sets (NS) have referred to as interval fuzzy sets applied in minimizing the 
uncertainty and fuzziness in computer-vision and machine-learning communities and hence 
employed for several applications. As far as medical image processing applications are concerned 
NSs are obtained as an important technique for de-noising. Also, fuzzy segmentation with machine 
and deep learning is determined as a familiar procedure that splits input image into distinct regions 
for precise learning. Several research works conducted in different image-processing domains. 
However, less works was focused on denoising and segmentation of medical image processing with 
minimal time complexity and accuracy. In this work we plan to develop a Kalman–Bucy Filtered 
Neutrosophic Neuro Fuzzy Image Denoising (KBF-NNFID) method with the objective of reducing 
the noisy artifacts with higher peak signal-to-noise ratio in a computationally efficient manner. First, 
medical images obtained from Brain MRI LGG segmentation dataset are subjected to filtering 
employing Kalman Bucy Filtering algorithm with series of measurements examined. Second with 
the filtered medical images provided as input, uncertainty is handled by utilizing Neutrosophic 
Neuro Fuzzy set (NNFS) with help of the membership grade. With the aid of three membership 
grades, i.e., truth, indeterminacy and falsity, uncertainty involved in noisy image are said to be 
handled in a time efficient manner. By this way, an efficient image denoising process is performed 
with better PSNR. Experimental evaluation is carried out using medical images with different 
performance metrics such as enhanced PSNR and true positive rate up to 13%, 14% as well 
minimum execution time by 38% using medical images. 

DOI: 10.5281/zenodo.13175808