An Optimized Approach for Tumor Detection using Co-Occurrence Matrix and Back Propagation Neural Network

Brain tumours are a significant contributor to mortality rates in the population. The early detection and accurate classification of a tumour can significantly enhance the likelihood of survival. Magnetic resonance imaging (MRI) is presently an essential diagnostic modality in the investigation of the human brain. A brain tumour is characterized as an anomalous proliferation of cells within the brain or the central spinal canal. The present paper introduces the Backpropagation network as an artificial neural network methodology. The classification of tumour types in CT or MRI images of patients with pituitary, glioma, and meningioma tumours is facilitated through its utilization. The given paper involves implementation of image augmentation and normalization techniques, followed by the extraction of image features through the utilization of Gray Level Co-occurrence Matrix (GLCM). The findings involve two distinct modes, namely Training/Learning and Testing/Recognition, which ultimately yield a classified output. The results present linear relationship between epochs and number of iterations thereby validating the proposed approach.