A Comprehensive Review of Deep Convolutional Neural Networks for Brain Tumor Detection and Classification

Abstract

Brain tumours are a major health burden worldwide, and the early and accurate detection of them is vital to improve treatment radiation doses and overall patient survival. The development of deep learning, in particular CNN, has drastically changed medical image analysis by automating the process of feature extraction and diagnosis. In high-quality medical images, MICs can also cause erroneous detections. In this review paper, we comprehensively investigate the deep learning technologies for brain tumour detection and classification on MR and CT images in the state of the art. This paper demonstrates that the CNN model has superiority over traditional manual diagnosis, effectively learns the hierarchical features of images, and reduces misdiagnosed cases. It addresses essential model designs, benchmarks, and issues related to integrating these models into clinical workflows. Furthermore, the survey investigates the ability of CNN-based systems to provide robust and efficient telemedicine-based and high-quality neuro-oncological diagnostics utilizing high-dimensional imaging whilst accurately segmenting tumorous regions. The inclusion of AI in medical imaging drives not only an increase in workflow efficiency but also the prospect of personalized treatment planning. This paper highlights the breakthrough applications of CNNs in neuroimaging and future research directions, which include further enhancing the model generalization on various datasets and real-world clinical applications that can help radiologists achieve better patient care.