Tumor image segmentation using artificial neural networks

Magnetic Resonance Imaging (MRI) brain segmentation scans were beneficial for diagnosing, treatment and evaluation of affected tumors or specific diseases. Until now, medical professionals accomplished manual segmentation, which is widely utilized in hospitals and diagnostic centers. Manual Segmentation is an authentic conventional method, which is accurate and consumes more time, expensive, finally might be not reliable. Several routine and semi-routine practices for segmentation of magneto resonance images are available in the previous works, nonetheless the obtained accurate values are not comparable with various manual segmentation methods. The proposed method in this work employs a Supervised Artificial Neural Network (ANN) algorithm. The specifications considered in this paper are PSNR, Mean square and Normalized absolute errors, Maximum and Average differences, Normalized Cross-Correlation and structural content. The proposed ANN method attained competitive results with several segmentation methods trained with Artificial Neural Network.