Abstract:Accurate segmentation of the acquired 3D stack fluorescence images of mice gland is a prerequisite for qualitative and quantitative analysis. However, the traditional methods cannot fully satisfy the need of accurate segmentation for 3D reconstruction result of the generally low signal-to-noise ratio of the 3D fluorescence images and the unique structure of mice gland. In this article, the task of mice gland segmentation is regarded as a multi-label classification problem and design to bind the widely-used method U-Net up with an ensemble learning method for 3D segmentation of mice gland. The proposed method is tested on our 3D confocal images of mice gastric gland. The experiment results show that our method can implement effective and fast segmentation of mice gastric gland. We would expect that this method is applicable for similar structural segmentation in other glands and organs.

3D Segmentation of Mice Gland Based on Ensemble Learning