Research on the RIS-PACS environment operation and big data testing of artificial intelligence diagnostic model of bone age imaging
10.3969/j.issn.1002-1671.2024.03.033
- VernacularTitle:骨龄影像人工智能诊断模型的RIS-PACS环境运行及大数据测试研究
- Author:
Lihong LI
1
;
Xiujun YANG
;
Shuang LAI
Author Information
1. 上海交通大学医学院附属儿童医院影像科,上海 200062
- Keywords:
bone age;
artificial intelligence model;
clinical application
- From:
Journal of Practical Radiology
2024;40(3):487-490
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the feasibility and effectiveness of the artificial intelligence(AI)diagnostic model for bone age imaging in the radiology information system-picture archiving and communication system(RIS-PACS)environment operation.Methods The optimized bone age AI model was integrated into the RIS-PACS platform.The bone age imaging data of 88 038 patients aged 0-18 years old were automatically evaluated.The reference bone age was determined by the consensus of two experienced radi-ologists based on GP map,with an error of±1.0 year old.The success rate,accuracy,system compatibility,stability,and influen-cing factors of results were further analyzed.Results The time for bone age AI evaluation of each case did not exceed 3 seconds,and the success rate of automatic evaluation reached 100%.AI model of bone age and RIS-PACS in hospital could be well integrated.Accord-ing to the readings evaluated by pediatric radiologists based on GP maps,the accuracy rate was 93.05%for girls and 89.53%for boys,with a mean absolute error(MAE)of(0.42±0.54)years old for girls and(0.45±0.60)years old for boys,respectively.The AI model could run efficiency in the RIS-PACS,which significantly reduced the burden of radiologists.The factors that affect the accuracy of the model were image position,exposures,multiple images in a single sequence and hand deformity,etc.Conclusion The bone age imaging AI diagnostic model can be seamlessly embedded into RIS-PACS in hospital,achieving one click bone age imaging diagnosis.
