A Japanese research team has developed a computer algorithm that can search through images in a PACS and identify the correct patients in misfiled chest X-rays by using unique anatomical data in the scan.
According to a study published in Radiological Physics and Technology, previous research has determined that misfiled X-rays accounted for about 0.117 percent of 279,222 studies acquired at a university hospital in Japan over a two year period.
"Misfiled images in a PACS environment may create serious medical accidents in hospitals," the authors, led by Risa Toge of Kyushu University, said, according to AuntMinnie.com.
The algorithm creates what the researchers call "biological fingerprints" that are based on anatomical features such as cardiac shadow, lung apex, superior mediastinum, right lower lung and the whole lung field. It analyzes the biological footprint of the misplaced X-ray and, by matching the anatomical features with other PACS images, comes up with a correlation score as a measure of the overall similarity of the two images.
The correct patient is identified automatically by the image that has the high correlation score.
The researchers used the algorithm on 200 pairs of randomly selected chest X-rays and were able to correctly match 78 percent of the misfiled X-rays with the patient. Then, by varying the weighting factors given to each of the anatomical features used in the producing the correlation score, the researchers were able to increase accuracy to 87.5 percent. The researchers went on to say that the number of misfiled X-rays discovered were so small that they could simply be identified manually be radiology personnel.
Toge and his colleagues, according to the AuntMinnie.com article, believe the algorithm can be used as an automated warning system for X-rays that have been erroneously filed before they go into a PACS, and that it could be expanded to include other modalities.