Research on Urinary Calculus Recognition and Area Measurement Based on U-Net
LIAN Wenyong
JIAO Min
LIU Na
TIAN Min
Abstract：The traditional manual identification method for urinary calculi is inefficient and prone to subjective influence by doctors,leading to certain errors and affecting the accuracy of diagnosis and the timeliness of treatment.This study proposes a deep learning method based on the U-Net network for automatic identification and size measurement of urinary calculi.By using the U-Net model to segment the computed tomography(CT)images of calculi,the automatic recognition of the calculus region is achieved,and the size of the calculi is measured through the least squares method model.The research results show that the U-Net model demonstrates high accuracy in segmentation,with an average intersection over union(IoU)of 84.39%and an average Dice coefficient of 90.8%,accurately identifying and locating the calculi.The size measurement model based on the least squares method also proves its effectiveness in the measurement of calculus area.This study provides a reliable technical means for the diagnosis of urinary calculi through precise segmentation and automatic measurement of calculus location,assisting clinicians in making better decisions.
Keywords：Urinary calculiU-NetDeep learningImage segmentationLeast squares method
Publication Date：2025-06-15
Online Publishing Date：2025-09-26(First online date of this platform, not the publication date of the document)
Pages：4( 20-22,后插5-后插6 )