Brain tumor detection using CNN: a systematic review paper

One of the most dangerous neurological conditions is brain tumours, and improving patient survival rates depends heavily on early identification. Convolutional Neural Networks (CNNs) have demonstrated exceptional promise in automating brain tumour identification with high accuracy and efficiency as deep learning in medical imaging has grown in popularity. This study provides a thorough examination of current developments in CNN-based methods for classifying and segmenting brain tumours using MRI data. We investigate several CNN architectures, including ResNet, VGG, Inception, Efficient Net, and YOLOv7, by thoroughly reviewing 15 cutting-edge research articles released between 2019 and 2024. These models are assessed according to their precision, resilience, computational effectiveness, and suitability for various tumour types. The majority of studies make use of publically accessible MRI datasets and concentrate on enhancing feature extraction, decreasing false positives, and refining training methods using methods such as attention processes, skip connections, and transfer learning. Finding the advantages, disadvantages, and gaps in the existing CNN-based techniques as well as making recommendations for future research areas are the objectives of this work. Our results support the increasing use of CNNs in clinical decision-making systems for brain tumour analysis and reaffirm their revolutionary importance in medical diagnosis.