Brain Tumor Segmentation Using the Hybrid CNN & LSTM Model

The present study offers a convolutional neural network (CNN) architecture with tumor segmentation using a hybrid CNN & LSTM model. The study makes use of a split dataset, where the training data is utilized for the training test information is utilized for model assessment of methods after preliminary processing. Utilizing the training dataset, together with additional models based on Gaussian Naive Bayes, Random Forest, and Logistic Regression, the proposed hybrids CNN & LSTM model was trained. These additional models—Gaussian Naive Bayes, Random Forest, and Logistic Regression serve as benchmarks for comparing and assessing the efficacy of the proposed hybrids CNN & LSTM model applying a variety of performance metrics. The F1score of $\mathbf{9 7. 6 6 \%}$, accuracy of $\mathbf{9 8. 4} \%$, precision of $\mathbf{9 6. 3 4 \%}$, recall of 97.9% & recall of 97.9% are impressive findings that show the efficiency of the suggested hybrid CNN & LSTM approach. These results surpass the performance of previous models, highlighting the effectiveness of the hybrid CNN & LSTM architecture considering tumor segmentation. Additionally, a VGG UNet architecture is employed for segmentation purposes, enhancing the accuracy of the segmentation process.