刑事技术2023，Vol.48Issue(6)：570-576,7.DOI:10.16467/j.1008-3650.2022.0070
椎体骨折影像AI快速诊断方法研究
Research on AI Rapid Diagnosis of Imaging of Vertebral Fracture
摘要
Abstract
Vertebral fracture evaluation is a common task in forensic practice.Vertebral fracture diagnosis and discrimination based on imaging have been bottleneck problems that have plagued judicial appraisal for a long time.The massive group-petitions-and-stalking lawsuits caused by the large diagnosis deviation,low efficiency,repeat and multiple appraisals have seriously affected social harmony.This study aims to construct an artificial intelligence(AI)method for the rapid diagnosis of human vertebral fracture images to realize the automated assessment of vertebral fractures.A total of 1 151 cases of lumbar vertebral fractures were collected as research samples,divided into a training set of 800 cases,a validation set of 151 cases,and a testing set of 200 cases.Using the training set and validation set samples,the vertebral fracture evaluation model was constructed through six steps,including image preprocessing,spinal positioning model construction,vertebral body positioning and recognition model construction,vertebral body segmentation model construction,fracture diagnosis model construction and model evaluation.We applied the test set to test the model.The results show that the established AI evaluation model achieved an accuracy rate of 76％in identifying vertebral fractures.This model provides technical support for developing automated assessment tool for vertebral fractures.关键词
法医影像学/椎体骨折/人工智能/图像识别/深度学习Key words
forensic imaging/vertebral fracture/artificial intelligence/image recognition/deep learning分类
政治法律引用本文复制引用
马文静,刘凡,刘华,裴京哲,张睿,施维..椎体骨折影像AI快速诊断方法研究[J].刑事技术,2023,48(6):570-576,7.基金项目
公安部"双十计划"重点攻关项目(2020SSGG0703) （2020SSGG0703）
