Adaptive Lightweight Framework for Semi-Supervised Segmentation of Brain Tumors

Medical image segmentation has always been one of the most challenging tasks in the field of computer vision. The requirement for segmentation accuracy makes it difficult to train models as the number of parameters increases. Among these, semi-supervised semantic segmentation has attracted increasing research interest in recent years due to its ability to effectively utilize large amounts of unlabeled data, thereby reducing the need for large-scale fully labeled training data. However, semi-supervised segmentation tasks are often limited by issues such as parameters and the number of datasets, making it difficult to achieve precise segmentation. This paper proposes a method based on an adaptive attention mechanism and lightweight feature fusion, which achieves good segmentation performance with fewer model parameters. To investigate the segmentation capability of this method, we also experimented with different feature fusion methods in the brain tumor segmentation task. The experimental results show that the proposed model not only significantly reduces the number of parameters in the baseline model but also performs better in the brain tumor semi-supervised segmentation task.