An Efficient Approach for the Detection of Brain Tumor Using Fuzzy Logic and U-NET CNN Classification

• Published in: Computational Science and Its Applications - ICCSA 2021 Vol. 12953; pp. 105 - 118
• Main Authors: Maqsood, Sarmad, Damasevicius, Robertas, Shah, Faisal Mehmood
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2021; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Brain MRI; Deep learning; Fuzzy logic; Image classification; Medical image processing; Segmentation
• ISBN: 303086975X; 9783030869755
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-86976-2_8

Clinical diagnosis has increased marvelous significance in current day healthcare systems. This article proposes a brain tumor detection method using edge detection based fuzzy logic and U-NET Convolutional Neural Network (CNN) classification method. The proposed tumor segmentation system is based on image enhancement, fuzzy logic based edge detection, and classification. The input images are pre-processed using the contrast enhancement and fuzzy logic-based edge detection method is applied to identify the edge in the source images and dual tree-complex wavelet transform (DTCWT) is used at different scale levels. The features are calculated from the decayed sub-band images and these features are then categorized using U-NET CNN classification which recognizes the meningioma and non-meningioma brain images. The proposed method is evaluated using accuracy, sensitivity, specificity, and dice coefficient index. Simulation study demonstrates that the proposed technique achieves better performance, both visually and quantitatively in comparison with other approaches.