Brain Cancer Classification based on T1-Weighted Images Using Deep Learning

Brain Cancer, a devastating and potentially life-threatening disease affecting millions globally, necessitates early and accurate detection for optimal treatment planning and improved patient outcomes. This paper introduces an enhanced VGG-16 Deep Learning Model, refined through fine-tuning, the objective is to accelerate early detection of brain cancer types, encompassing gliomas, meningiomas, and pituitary tumors for better optimal treatment planning, based on T1-weighted CE-MRI dataset that was acquired from Nanfang Hospital and general hospital, Tianjin Medical University from 2005 to 2019. The fine-tuned VGG-16 model performed very well by achieving 95% accuracy with a significant and stable validation loss decrease from 0.9 to 0.2 for the last epoch, indicating no overfitting, suggesting that it has the potential to be used in real-world applications. The model's long training process, which might take hours given the restricted computing resources available, is the challenge in this study. By assessing the accuracy and recall of the model, the model performance is assessed. The methodology's results are presented, demonstrating how well the suggested model classifies brain cancer. The current model can be used as a support for medical expertise to classify brain cancer, however further research and development are necessary for enhancing this method in the future.