The algorithm automatically localizes brain tumors on magnetic resonance imaging with an accuracy of about 80%. Brain tumor segmentation plays an essential role in medical imaging analysis. In recent studies, deep convolutional neural networks (DCNNs) are widely applied to solve tumor segmentation tasks. For this study, VinBigData proposes a new training method to improve segmentation results by adding an additional classifier to the network. The entire network is trained from start to finish based on the training dataset of the Multimodal Brain Tumor Segmentation Challenge (BraTS) 2020. On the BraTS test suite, the model achieves an average Dice Score of over 80%.
The algorithm automatically localizes brain tumors on magnetic resonance imaging
Tác giả: 
VBD
VBD 
