A Lightweight Fully Convolutional Network for Cardiac MRI Segmentation
Wenjie Yang, Shufang Li · 2020 International Conference on Computer Information and Big Data Applications (CIBDA) · 2020
Cardiac magnetic resonance imaging is a commonly used method to assess the function and structure of the cardiovascular system. Separating the heart structure from Cardiac MRI is an important step in calculating the cardiac function index. The automatic segmentation algorithm can significantly reduce the workload of the doctor. Most of the current automatic segmentation algorithm has a large calculation amount and size, has high requirements on the hardware. Based on the U-Net model, this paper proposes a lightweight full convolution model with separable convolution and residual connection. Experiments show that the model reduces the number of parameters while ensuring learning ability and balances speed and accuracy.