FF-UNET: An Enhanced Framework for Improved Segmentation of Brain Tumors in MRI Images

With the development of medical imaging technology, accurate and effective tumor segmentation is crucial. In this study, we investigated the potential of deep learning based brain tumor MRI image segmentation technology. We propose an enhancement method called FF-UNET, which utilizes feature fusion and combines the capabilities of UNet and CNN models to improve segmentation accuracy. Using preprocessing techniques to improve tumor visibility, and then using a customized layered UNet model for segmentation. To reduce overfitting, a discard layer is introduced after each convolutional block stack. In addition, the CNN process utilizes the context of brain tumor MRI images to further enhance the segmentation performance of the model. In two datasets, our method achieved an accuracy of over 98%, with accuracy and Jaccard coefficients exceeding 90%. The experimental results indicate that our proposed framework outperforms state-of-the-art models in distinguishing brain tissue. The FF-UNET model has great potential as a feasible diagnostic tool, enabling radiologists to accurately segment brain tumor images and improve patient care.