Brain tumors can have detrimental effects on brain function and pose a serious threat to life. Detecting and treating brain tumors early is vital for saving lives. However, identifying tumor-affected brain cells is a difficult and time-consuming process. Common imaging techniques like Computer Tomography scans and Magnetic Resonance Images (MRIs), while helpful, can present challenges for radiologists in manual assessments. The field of image processing faces significant obstacles in achieving accurate and efficient brain tumor detection. This research work proposes an improved deep learning-based model for efficient brain tumors detection. Preprocessing, segmentation, feature extraction, feature selection, and classification are some of the processes that make up the proposed model. To improve the quality of brain images, preprocessing steps are employed using the compound filter made up of Gaussian, mean, and median filters. In addition, morphological and threshold-based segmentation are used to separate the tumor from healthy brain tissue. By using the grey-level co-occurrence matrix (GLCM)-based technique is employed to extract the texture and intensity patterns for identifying tumor areas. The optimal feature selection is performed by using the Whale Social Spider-based Optimization Algorithm (WSSOA)-based metaheuristic. Finally, Deep Convolutional Neural Network (DCNN) is used for accurate tumors detection. The proposed technique is evaluated using a publicly well-known Figshare dataset. Performance is compared with seven latest state-of-art models using metrics such as accuracy, precision, recall, and F1-score. The results demonstrate that the proposed technique achieves exceptional brain tumor classification accuracy of 99.29%. These promising findings highlight the potential of the proposed model to enhance accurate and efficient brain tumor detection, ultimately leading to improve diagnosis and potentially saving more lives.

脑肿瘤会对大脑功能产生有害影响，并对生命构成严重威胁。及早发现和治疗脑肿瘤对于挽救生命至关重要。然而，识别受肿瘤影响的脑细胞是一个困难且耗时的过程。计算机断层扫描和磁共振图像 (MRI) 等常见成像技术虽然有用，但也给放射科医生的手动评估带来了挑战。图像处理领域在实现准确有效的脑肿瘤检测方面面临着重大障碍。这项研究工作提出了一种改进的基于深度学习的模型，用于有效检测脑肿瘤。预处理、分割、特征提取、特征选择和分类是构成该模型的一些过程。为了提高大脑图像的质量，使用由高斯滤波器、均值滤波器和中值滤波器组成的复合滤波器进行预处理步骤。此外，形态学和基于阈值的分割用于将肿瘤与健康脑组织分开。通过使用基于灰度共生矩阵（GLCM）的技术来提取纹理和强度模式以识别肿瘤区域。最佳特征选择是通过使用基于鲸鱼社交蜘蛛的优化算法（WSSOA）的元启发式来执行的。最后，深度卷积神经网络（DCNN）用于准确的肿瘤检测。使用众所周知的 Figshare 数据集对所提出的技术进行评估。使用准确度、精确度、召回率和 F1 分数等指标将性能与七个最新的最先进模型进行比较。结果表明，所提出的技术实现了 99.29% 的出色脑肿瘤分类准确率。这些有希望的发现凸显了所提出的模型在增强脑肿瘤检测的准确性和效率方面的潜力，最终改善诊断并有可能挽救更多生命。