Leveraging Deep Learning Model for Computer Vision-Based Brain Tumor Classification in 3D MRI Brain Images.
Journal:
Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference
PMID:
40039474
Abstract
This study uses computer vision techniques to combine EfficientNet-3D and 3D Residual neural network(3DResnet) deep learning architecture to detect brain tumours in magnetic resonance imaging (MRI). The dataset includes a collection of 586 sets of brain MRI images obtained from the 2021 RSNA Brain Tumor Challenge. Before analysis, preprocessing steps were performed, including contrast enhancement, resampling, interpolation and centre alignment. Implementing the combined model yielded remarkable results, achieving an accuracy of 97.32%, 96.55% sensitivity and 98.15% specificity value through 5-fold cross-validation. The combined model leverages the strengths of the two algorithms, enabling it to handle various scenarios through comprehensive analysis. Its ability to classify instances as "unknown" allows for further examination and study by human experts. These findings highlight the potential usage of deep learning models in the early detection of brain tumours, ultimately contributing to improved patient outcomes and facilitating effective treatment planning.