Brain Tumor Detection from MRI Images Based on ResNet18
Michael Tang, Soo Siang Teoh
Abstract
The detection of brain tumors using image processing has recently emerged as a significant area of research. Despite numerous efforts, the development of an accurate brain tumor detection method remains a challenge. Machine learning techniques have been widely explored for this purpose, with deep learning models being a subset of particular interest. While various deep learning models such as GoogLeNet and CapsNet have been proposed, the use of pre- trained models such as ResNet18 has yet to be fully investigated. This study presents a deep learning approach utilizing ResNet18 for the detection of brain tumors in MRI images. The model was evaluated using a publicly available dataset obtained from Kaggle. Results of the experiments indicate that the model achieved an accuracy of 0.8833, sensitivity of 0.8667, specificity of 0.9000, and precision of 0.8966. Our findings demonstrate that this model outperforms previous deep learning models in terms of accuracy, specificity, and precision.