| Abstrakt: |
Our novel methodology for tumor identification improves accuracy and efficiency significantly by utilizing advanced techniques. Using the BRATS dataset, we combine Maximum A Posteriori (MAP) optimization for pixel extraction, Wiener deconvolution, and the EI-Fusion-Net deep neural network. Additionally, preprocessing techniques such as resizing, grayscale conversion, and Gaussian filtering are used to improve image quality. For better results, our novel image fusion EI-Fusion-Net approach uses specific wavelet transform techniques and a fusion network architecture capable of capturing both spatial and temporal information. Indeed, our findings show remarkable performance metrics, with a peak signal-to-noise ratio (PSNR) of 48.42 dB and a structural similarity index (SSIM) of 0.992, which outperform those of existing methods on the BRATS dataset. This demonstrates the EI-Fusion-Net model's ability to effectively combine diverse data sources, resulting in promising advances in brain tumor detection via refined medical image processing techniques. [ABSTRACT FROM AUTHOR] |
