TuNet: End-to-end Hierarchical Brain Tumor Segmentation using Cascaded Networks
This work addresses automatic and accurate segmentation of gliomas to improve patient survival rates and reduce treatment costs, representing an incremental advance in medical imaging.
The paper tackled brain tumor segmentation by proposing an end-to-end cascaded network that leverages hierarchical tumor sub-region structures, achieving dice scores of 88.06, 80.84, and 80.29 for whole tumor, tumor core, and enhancing tumor, respectively, on an online test set.
Glioma is one of the most common types of brain tumors; it arises in the glial cells in the human brain and in the spinal cord. In addition to having a high mortality rate, glioma treatment is also very expensive. Hence, automatic and accurate segmentation and measurement from the early stages are critical in order to prolong the survival rates of the patients and to reduce the costs of the treatment. In the present work, we propose a novel end-to-end cascaded network for semantic segmentation that utilizes the hierarchical structure of the tumor sub-regions with ResNet-like blocks and Squeeze-and-Excitation modules after each convolution and concatenation block. By utilizing cross-validation, an average ensemble technique, and a simple post-processing technique, we obtained dice scores of 88.06, 80.84, and 80.29, and Hausdorff Distances (95th percentile) of 6.10, 5.17, and 2.21 for the whole tumor, tumor core, and enhancing tumor, respectively, on the online test set.