Title | ||
---|---|---|
Determining scanned body part from DICOM study description for relevant prior study matching. |
Abstract | ||
---|---|---|
The typical radiology reporting workflow involves the radiologist first looking at one or more relevant prior studies before interpreting the current study. To improve workflow efficiency, PACS systems can display relevant prior imaging studies, typically based on a study's anatomy as indicated in the Body Part Examined field of the DICOM header. The content of the Body Part Examined field can be very generic. For instance, an imaging study to exclude pancreatitis and another one to exclude renal stones will both have "abdomen" in their body part field, making it difficult to differentiate them. To improve prior study matching and support better study filtering, in this paper, we present a rule-based approach to determine specific body parts contained in the free-text DICOM Study Description field. Algorithms were trained using a production dataset of 1200 randomly selected unique study descriptions and validated against a test dataset of 404 study descriptions. Our validation resulted in 99.94% accuracy. The proposed technique suggests that a rule-based approach can be used for domain specific body part extraction from DICOM headers. |
Year | DOI | Venue |
---|---|---|
2013 | 10.3233/978-1-61499-289-9-67 | Studies in Health Technology and Informatics |
Keywords | Field | DocType |
PACS,radiology,SNOMED,medical informatics applications,radiology informatics,relevant prior study matching | Information retrieval,Computer science,DICOM Study,Multimedia | Conference |
Volume | ISSN | Citations |
192 | 0926-9630 | 1 |
PageRank | References | Authors |
0.39 | 0 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Thusitha Mabotuwana | 1 | 10 | 5.54 |
Yuechen Qian | 2 | 80 | 10.74 |