Tumor Image Identification based on DWT and PNN Classifier

Medical specialists have a hard time finding tumors in the ground truth image of the brain because the intensity, magnitude, position, distinction, and shape of the tumors vary between images. Other reasons include different acquisition methods, modalities, and finally the age of the affected patient. The primary diagnosis of Magnetic resonance imaging (MRI) brain image is regularly termed as a brain injury. The cells can be tumorous or non-tumorous. The presence of abnormal tumors, sources the stress in the inner part of the infected part to intensify. The aforementioned situation results in damage to the sensitive brain image cells, resulting in life-threatening situations. It is challenging to identify abnormalities in the human brain using conventional imaging tools. MRI can differentiate and clarify the network architectural section of the brain image. This proposed work deliberates on noise reduction approaches, grey-level cooccurrence matrix feature extraction, and Discrete Wavelet Transform based tumor imaged area growth segmentation to simplify and improve values. Morphological filtering was used to remove any potential noise after segmentation. MRI data were used to train and assess a Probabilistic Neural Network classifier for identifying infected parts. The proposed method was tested and trained brain MRI images, were more accurate in discriminating between usual and infected brain MR images in the studies, demonstrating its utility.