A Deep Learning based Brain Tumour Detection using Multimodal MRI Images
R. Jansi, S. Kowsalya, S. Seetha, A. Yogadharshini
Abstract
Unchecked and fast cell proliferation leads to brain tumors. Death could result if the condition is not treated in the beginning stages. The variety of tumor location, shape, and size poses a significant obstacle to the detection of brain tumors. Through image preprocessing, data augmentation, and convolutional neural networks (CNNs), this research gives a thorough method for detecting brain tumors. This work utilizes multimodal MRI images from brain tumor Brats dataset, which is improved through preprocessing, dilation and erosion. Dilation and erosion produces a better view of the tumor regions for accurate detection. A CNN model is then trained with a focus on shuffling the data for better performance. TensorFlow and Keras libraries were utilized for the implementation of the proposed system. The proposed framework achieved a good accuracy of 98.2% in detection of brain tumours using the brain tumor Brats dataset. This system can assist the health care professionals to accurately predict the presence of brain tumors.