Brain Tumor Detection Based on Hybrid Convolutional Adaptive Neuro Fuzzy Inference System Using MRI Image

The abnormal or irregular growth of cells in regions of the human body that affects surrounding tissues is termed a tumor. Brain tumors are among the most dangerous and life-threatening types of tumors, arising from the abnormal growth of cells within the brain. However, existing detection methods often suffer from limitations, such as poor noise handling in MRI images, inaccurate segmentation, and low generalization across varying datasets. To resolve these difficulties, in this article, the convolutional-adaptive neuro fuzzy inference system (Conv-ANFIS) approach is devised for brain tumor detection from the magnetic resonance imaging (MRI) brain images. Conv-ANFIS is a combined model that merges a convolutional neural network (CNN) with an adaptive neuro-fuzzy inference system (ANFIS) and helps improve brain tumor detection accuracy. Initially, the de-noising of acquired input MRI images is done by applying a non-local means (NLM) filter. The skull portions are removed from the de-noised MRI image. Next, segmentation is performed by applying the structure correcting adversarial network (SCAN) to the image after skull removal. SCAN is a deep learning-based segmentation model designed for medical imaging. It ensures the accurate detection of tumors in MRI images. Subsequently, the features are mined from the segmented images by utilizing different feature extractors. Finally, the brain tumor is identified using the designed Conv-ANFIS model from the extracted feature vector. Furthermore, the effectiveness of Conv-ANFIS in detection is confirmed, achieving a recall of 92.31%, precision of 90.13%, and an F1-score of 91.21%, outperforming existing approaches.