Xiuquan Du, Zheng Pei, Ying Liu, Xinzhi Cao, Lei Li, Shuo Li
{"title":"MultiJSQ: Direct joint segmentation and quantification of left ventricle with deep multitask-derived regression network","authors":"Xiuquan Du, Zheng Pei, Ying Liu, Xinzhi Cao, Lei Li, Shuo Li","doi":"10.1049/cit2.12382","DOIUrl":null,"url":null,"abstract":"<p>Quantitative analysis of clinical function parameters from MRI images is crucial for diagnosing and assessing cardiovascular disease. However, the manual calculation of these parameters is challenging due to the high variability among patients and the time-consuming nature of the process. In this study, the authors introduce a framework named MultiJSQ, comprising the feature presentation network (FRN) and the indicator prediction network (IEN), which is designed for simultaneous joint segmentation and quantification. The FRN is tailored for representing global image features, facilitating the direct acquisition of left ventricle (LV) contour images through pixel classification. Additionally, the IEN incorporates specifically designed modules to extract relevant clinical indices. The authors’ method considers the interdependence of different tasks, demonstrating the validity of these relationships and yielding favourable results. Through extensive experiments on cardiac MR images from 145 patients, MultiJSQ achieves impressive outcomes, with low mean absolute errors of 124 mm<sup>2</sup>, 1.72 mm, and 1.21 mm for areas, dimensions, and regional wall thicknesses, respectively, along with a Dice metric score of 0.908. The experimental findings underscore the excellent performance of our framework in LV segmentation and quantification, highlighting its promising clinical application prospects.</p>","PeriodicalId":46211,"journal":{"name":"CAAI Transactions on Intelligence Technology","volume":"10 1","pages":"175-192"},"PeriodicalIF":8.4000,"publicationDate":"2024-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1049/cit2.12382","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"CAAI Transactions on Intelligence Technology","FirstCategoryId":"94","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1049/cit2.12382","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
引用次数: 0
Abstract
Quantitative analysis of clinical function parameters from MRI images is crucial for diagnosing and assessing cardiovascular disease. However, the manual calculation of these parameters is challenging due to the high variability among patients and the time-consuming nature of the process. In this study, the authors introduce a framework named MultiJSQ, comprising the feature presentation network (FRN) and the indicator prediction network (IEN), which is designed for simultaneous joint segmentation and quantification. The FRN is tailored for representing global image features, facilitating the direct acquisition of left ventricle (LV) contour images through pixel classification. Additionally, the IEN incorporates specifically designed modules to extract relevant clinical indices. The authors’ method considers the interdependence of different tasks, demonstrating the validity of these relationships and yielding favourable results. Through extensive experiments on cardiac MR images from 145 patients, MultiJSQ achieves impressive outcomes, with low mean absolute errors of 124 mm2, 1.72 mm, and 1.21 mm for areas, dimensions, and regional wall thicknesses, respectively, along with a Dice metric score of 0.908. The experimental findings underscore the excellent performance of our framework in LV segmentation and quantification, highlighting its promising clinical application prospects.
期刊介绍:
CAAI Transactions on Intelligence Technology is a leading venue for original research on the theoretical and experimental aspects of artificial intelligence technology. We are a fully open access journal co-published by the Institution of Engineering and Technology (IET) and the Chinese Association for Artificial Intelligence (CAAI) providing research which is openly accessible to read and share worldwide.