A Proposed Framework to De-noise Medical Images Based on Convolution Neural Network
Abstract
These days analysing patient data in the form of medical images to perform diagnose while doing detection and prediction of a disease has emerged as a biggest research challenge. All these medical images can be in the form of X-RAY, CT scan, MRI, PET and SPECT. These images carry minute information about heart, brain, nerves etc within themselves. It may happen that these images get corrupted due to noise while capturing them. This makes the complete image interpretation process very difficult and inaccurate. It has been found that the accuracy rate of existing method is very less so improvement is required to make them more accurate. This paper proposes a Machine Learning Model based on Convolutional Neural Network (CNN) that will contain all the filters required to de-noise MRI or USI Images. This model will have same error rate efficiency like those of data mining techniques which radiologists were interested in. The filters used in the proposed work are namely Weiner Filter, Gaussian Filter, Median Filter that are capable of removing most common noises such as Salt and Pepper, Poisson, Speckle, Blurred, Gaussian existing in MRI images in Grey Scale and RGB Scale.
Authors
Anany Jain, Akash Khatana