A hybrid deep learning based brain tumor classification and segmentation by stationary wavelet packet transform and adaptive kernel fuzzy c means clustering.

• One of the deadly and dangerous types of cancer seen in children and adults named Brain tumor. • Brain tumor's accurate and early diagnosis is significant for the treatment process. • The tumor segmentation's accuracy is very crucial to diagnosis accuracy. • The primary goal of this suggested approach is to accurately and efficiently assess the tumor. One of the deadly and dangerous types of cancer seen in children and adults named Brain tumor. Brain tumor's accurate and early diagnosis is significant for the treatment process. The tumor segmentation's accuracy is very crucial to diagnosis accuracy. As a consequence, employing brain MR image segmentation, this research provides a hybrid deep learning-based brain tumor diagnosis and classification. The primary goal of this suggested approach is to accurately and efficiently assess the tumor. The MRI images were first preprocessed for the segmentation phase. The preprocessed photos are then used throughout the feature extraction procedure. Stationary wavelet packet transform (SWPT) is applied for the feature extraction process. Then, a hybrid Adaptive Black widow optimization with Moth Flame optimization (HABWMFO) is utilized for the optimal feature selection. Following that, the feature values are being sent to clustering for segmentation. For segmentation, the Adaptive Kernel Fuzzy C Means clustering technique (AKFCM) was developed. Finally, a Hybrid Convolution Neural Network- Long Short-Term Memory (CNN-LSTM) deep learning is used to improve tumor classification accuracy. The suggested technique is implemented on the MATLAB platform, and its effectiveness is assessed using performance measures such as F1 score, accuracy, precision, loss, and recall. These analyses show that the suggested strategy is far more successful than the existing ones. [ABSTRACT FROM AUTHOR]