Article Preview
TopIntroduction
The MRI is a significant invention that has unbolted a new era in medical pathology where a disease or any abnormality in the human body is accurately and precisely identified. Brain MRI is an imaging technology that aids in the analysis and study of the human brain anatomy and wellbeing through non-invasive approaches. Over the years the imaging technology has explored new paths in the automation of the diagnosis and treatment, With the promotion and progress of development in human brain MRI technology, there is anonymous growth in data for investigation, which could not handle manually. As a part of automation, the volumetric approximation and analysis are few such parameters of the damaged region in the human brain that is to be investigated.
In the preliminary stages, the MRI image is pre-processed for the noise, which is put into the MR image at the time of capturing, due to the improper calibration of the equipment or at the time of rendering the MR image because of some external factors during digitization of the image. And then noise free MR image is then segmented for identification of the lesions and abnormalities in the human brain. When the MR image is segmented, a few of the crucial information would be identified, which would help in better analysis of the MR image. There are many semi-automated approaches for segmentation of the brain MR images that involve k-Means and fuzzy C-means based approach stated by (Bal et al., 2018), A Region growing based mechanism through gradients and variances as suggested by (Deng et al., 2010) in his article and integrated graph-cuts based for segmentation of the image as stste by (Song et al., 2006) in their article.
It is practically proven to have automated the segmentation approach for better results and precision the brain MR segmentation as stated by (Angulakshmi & Lakshmi Priya, 2017) in their articles on automated segmentation.
In the proposed mechanism, we are using a two-point crossover. The genetic Algorithm is used to identify the tumor regions and injuries in the brain. Segmentation is the fundamental task that aids in the analysis of the damaged region in the MR image, as stated by (UjjwalMaulik 2009). In the earlier stages of treatment, the approximate volume, which is estimated from the available 2D images, can aid in the preliminary. There are a few 3D MRI based volume estimation approaches proposed in recent years that calculates using different MRI field strengths proposed by (Heinen et al., 2018) and CAD system based approach proposed by (Roy et al. 14), region growing approach proposed by (Shanthi & Kumar, 2007; Benson, 2014; Lakshmi et al., 2014) has used morphological in their article and Fourier Transforms MRI through a multi spectral techniques proposed by (Chai et al., 2015) in their papers.
In a few unsophisticated countries due to lack of availability of 3D MRI technology, they have to go with multiple MRI scans, which are not preferably good due to radiations the patient undergoes while MRI scans. So we have come up with an approach using mathematical modeling for the assumption of the Volume of the damaged region,which is first of its kind for MR images. The volume of the damaged region is being evaluated through the vector calculus approach called Gauss Divergence Theorem for volume calculation for a closed, stated by (Stolze, 1978; Chansuparp et al., 2015) in their papers.
The paper is organized as follows; section 2 discusses the image pre-processing for noise removal using Otsu Based Adaptive Weighted Bilateral Filter. Section 3 discusses the image segmentation of the MR image using twin centric crossover based Genetic algorithm. Section 4 discusses the Gauss Divergence approach for volume estimation followed by section 5 discuss the experimental results and the accuracy of the proposed approach, and finally, Section 6 will be conclusion followed by Section 7 will be the future scope.