Lightweight Segmentation Algorithm of Liver Image Based on PPLC-UNet
ZHANG Hongxin
HU Jingmeng
ZHU Feng
LIU Qinghua
YANG Xinyi
ZHAO Mengdi
Abstract：In response to the common problems of massive parameters and low operating efficiency in current deep learning al-gorithms applied to liver segmentation,a lightweight liver segmentation algorithm of liver image based on PPLC-UNet is proposed.Based on the U-Net with codec structure,to decrease the quantity of parameters in the algorithm model,the backbone network of PP-LCNet is utilized to encode liver image features.In addition,the decoding structure is added with the attention module and re-sidual idea to improve the utilization of useful features.Experiments conducted on the Liver Tumour Segmentation Challenge LiTS2017 dataset show that the PPLC-UNet algorithm model had a number of parameters of 3.88 MB,a Dice coefficient of 90.86％.The prediction time for per liver CT slice image of approximately 7.36 ms,it is a lightweight algorithm model with superior perfor-mance that can be used to assist physicians in the diagnosis of liver diseases.
Keywords：liver segmentationPPLC-UNetcodec structureattention modulelightweight
Publication Date：2025-08-20
Online Publishing Date：2025-12-12(First online date of this platform, not the publication date of the document)
Pages：6( 2095-2100 )