Finding anatomical matches among DICOM images
Publication Date: 2003-May-13
The IP.com Prior Art Database
It would be very useful to a radiologist to be able to scan through two or more series of the same patient comparatively, i.e., when he views an image in one of the series, the other series should automatically show a matching image. A matching image is an image of the same body part of the same patient viewed from the same angle.
Finding anatomical matches among DICOM images.
It would be very useful to a radiologist to be able to scan through two or more series of the same patient comparatively, i.e., when he views an image in one of the series, the other series should automatically show a matching image.
A matching image is an image of the same body part of the same patient viewed from the same angle.
It is not necessary that simply flipping to the next/previous image across different series will cause matching images to be seen. This is because the spacing between images may be different at the time of the two scans; or some images may have been rejected because of their poor quality; or the radiologist may have changed the image order.
Thus, the aim is to display a matching image to the radiologist. Medical images have various DICOM attributes as part of their header information: among them are DICOM attributes that specify the location and orientation of the image that can be used to find a match.
In their header, many DICOM images have (among other information) the position of the image (Image Position (tag value: 0020,0032)) and the orientation of the image (Image Orientation (tag value: 0020,0037)). This information is used to determine the closest image (in terms of the position) at nearly the same orientation in the other series. If the closest image determined in this manner is at a close enough value, then it is displayed as the anatomically matching image.
The details of the algorithm are given in the flowchart drawn below.
There are two main advantages of this approach to find an anatomically matching image:
1. It makes use of (existing) DICOM attributes that are part of most medical images: so no extra or new information is needed for this purpose. Hence this logic can be applied to image series that were obtained from the scanners before this invention was created too.
2. It compares each and every image in the other series as a potential match: hence it will find and show the closest image at the same angle as a match. It also provides for a configurable tolerance parameter that...