Automatic Brain Tumor Detection Using Pattern Recognition Technique

Abstract

Distinct imaging modalities are being employed for identifying and analyzing, and diagnosing tumor lesions. In the light of noninvasive approaches such as MRI, CT filter, Ultrasound, SPECT, PET, and X beam, many types of therapeutic imaging advances. Magnetic Resonance Imaging outperforms Computed Tomography in the field of medicinal analysis frameworks since Magnetic Resonance Imaging provides greater complexity between various delicate tissues of the human body. A capable, attractive fields component in an MRI filter determines radiofrequency beats and provides spiffy gritty pictures of organs, delicate tissues, bone, and other inner parts of the body. The MRI technique is the most effective for detecting brain tumors. newlineIn the first algorithm, the Bilateral filter technique filters the noise from the input MRI brain image. The Canny edge detection with Haar Wavelet decomposition algorithm is used to determine the edges of the brain and ridges of the images tumor area. Finally, the hierarchical clustering algorithm is used to highlight the area of a brain tumor in the image. The results of this algorithm have a good compression ratio with high precision and Recall values compared to recent existing works. The second algorithm proposes an image handling system used to upgrade the information of MRI picture, and division of brain image, with the help of versatile kimplies bunching calculation for the examination of colorectal growth. The performance analysis of the algorithm shows reasonable accuracy, specificity and sensitivity rate compared to various recent works. In the third algorithm, the firefly algorithm is improved by the density-based spatial clustering algorithm, and a fitness function used based on soft computing criteria. Although the standard SVM is proven better yet it is not suitable for large data sets and also, its performance goes low if the noise level is high or target classes overlap. Therefore, the performance of different optimization algorithms is dependent on applications. newline