Advanced U-Net++ Architecture for Precise Brain Tumor Segmentation in MRI Images: A Robust Solution for Medical Image Analysis

Tumor segmentation in MRI images is a crucial process in medical imaging, aimed at accurately identifying and delineating tumor regions within the brain or other tissues. Hence proposed a modified U-Net++ based segmentation with predicted mask. To further improve the performance of the proposed method, incorporated additional layers into the U Net++ architecture. U-Net++ consists of Encoder Decoder and the encoder captures image content, while the decoder delineates object boundaries. U-Net++ excels in capturing both high-level and intricate details, which significantly boosts segmentation performance. This helps in accurate tumor segmentation of complex MRI images. To measure the effectiveness of the proposed model for segmentation and detection, Accuracy, Intersection over Union(IOU), Dice Coefficient and Area under the ROC Curve(AUC) scores are used. The proposed U-Net++ brought accuracy to 99 percent, which could significantly help in the process of accurate segmentation of human tissues. Accuracy of 39 % for IOU of 80, Dice Coefficient of $89 \%$, and $33 \%$ for the test set. This is equivalent to $8 \%$ sensitivity and accuracy and AUC of 78. The proposed method can be used as an additional diagnosis aid to help doctors determine the suitability of the treatment and the changes occurring in the body during the development of the disease