An Improved Approach Using Interactive Segmentation for Skin Wound Measurement in Forensic Medicine

摘要：

　　Introduction: Wound segmentation is a pivotal process of photography,which is an important and prevalent method for skin wound analysis and documentation in forensic science.However,due to the complexity and diversity of forensic wound images,no automatic segmentation algorithm is capable of judging the boundaries of the wound accurately.Meanwhile,the currently approach using lasso tool,though accurate,is time-consuming and labor-intensive for practical application.Materials and Methods: In this paper,interactive segmentation algorithms are applied to skin wound extraction.This application improves the performance of traditional segmentation methods and reduces the time of manual method.In order to verify the segmentation,a software platform hosting image preprocessing,image segmentation and wound area calculation is developed.This platform integrates the manual method and three interactive segmentation methods,and is completely available for forensic wound measurement based on photography technique.Results: The segmentation accuracy is evaluated by precision,recall and F-measure,which are higher than 0.96,0.95,and 0.96 respectively.The error of the area calculation is less than 5%,with respect to the golden standard.Discussion: These results demonstrate that interactive segmentation is practicable and precise enough to meet the demands of skin wounds measurement based on photography in forensic medicine.

关键词： Interactive segmentation Wound measurement Forensic medicine Photography

作者: Q. Chen Y. Xu F. Pu Y.B. Fan

作者单位: Key Laboratory for Biomechanics and Mechanobiology of Ministry of Education, School of Biological Science and Medical Engineering,Beihang University, Beijing 100191, China

会议类型: 国际会议

会议名称: World Congress on Medical Physics and Biomedical Engineering (2012年医学物理及生物医学工程国际会议(IFMBE))

会议地点: 北京

会议语种:英文

页码: 912-915

在线出版日期: 2012-05-26（万方平台首次上网日期，不代表论文的发表时间）