Initial clinical evaluation of an automated patient recognition system based on a template-matching technique of radiographic images

摘要

For maintaining reliability of picture archiving and communication system (PACS), it is important that all of images are stored in correct locations such as in the proper patients' folders. However, if the patient information, such as the identification number or the patient name, has been entered incorrectly, it is difficult to find this error. In this case, the image will be assigned to a different patient's folder and may not be stored in the proper patient's folder. If radiologists interpret a wrong patient image for a given patient, this may create a serious problem. It is, therefore, desirable to discover wrong patient's images immediately after an acquired image is transferred to PACS server so that corrective action can be taken before the image is stored in the wrong patient folder. We are developing an automated patient recognition method based on a template-matching technique in which we determine the correlation value for a current chest radiograph with a previous one. We have installed an automated patient recognition system in a hospital to evaluate the potential usefulness of the method developed.