A Hybrid CNN and Fuzzy Logic Framework for Enhanced Brain Tumor Detection and Textual Contextualization

Abstract

One of the most dangerous diseases is brain tumors (BT), which must be identified early and accurately in order to be effectively treated. The manual tumor tagging and image processing used in current diagnosis techniques are labor-intensive and error-prone. In order to overcome these limitations and increase the accuracy of brain tumor identification, this research proposes a hybrid approach that combines deep learning techniques such as convolutional neural networks (CNNs) and VGG versions, fuzzy logic systems, and MRI imaging. This work investigates tumor classification using deep learning and fuzzy logic. Features like tumor size and global threshold values were retrieved from the pre-processed pictures. Tumor size was measured using the watershed and region-growing approaches, and the results, along with the threshold values, were inputs to the fuzzy system. According to experimental data, the fuzzy system achieves good classification accuracy, with CNN + VGG16 models reaching up to 99%, especially when using the region-growing technique for tumor size detection and deep learning feature extraction. This hybrid fuzzy-CNN method shows promise as an effective and precise automated classification system by offering improved region-of-interest (ROI) segmentation and reliable feature identification. This approach may facilitate prompt and accurate clinical decision-making, advancing medical imaging analysis and improving patient outcomes by lowering the need for human involvement.