This study presents a novel method for enhancing brain tumor MR images by combining Unsharp Masking and Contrast Limited Adaptive Histogram Equalization (CLAHE). The proposed enhancement technique aims to improve the visual quality and diagnostic effectiveness of MR images. We evaluated the performance of this proposed method against the standard CLAHE approach using various image quality metrics, including Peak Signal-to-Noise Ratio (PSNR), Absolute Mean Brightness Error (AMBE), Mean Squared Error (MSE), Entropy, and Contrast. Experimental results demonstrate that our combined Unsharp Masking and CLAHE method significantly outperforms CLAHE alone across all evaluated metrics, indicating its potential for better visual enhancement and more accurate medical diagnosis in clinical settings. Additionally, the enhanced images were used for segmentation using a modified U-Net model with a combined loss function incorporating binary cross-entropy and dice loss. The segmentation results yielded impressive performance metrics: Accuracy of 0.980, Precision of 0.993, Recall of 0.980, Specificity of 0.979, Dice Score of 0.986, Intersection over Union (IoU) of 0.973, Jaccard Index of 0.973, and F1-Score of 0.986.