Data Archiving Requirements Diminished with Image Compression
Researchers from Austria and England working with image compression technology report that this approach to management of imaging studies may support the healthcare by shortening interpretation time along with decreased storage costs. The international team published their research with 3D JPEG 2000 algorithms used for abdominal CTs obtained in the emergency room in the August issue of the journal Radiology, suggesting that compression does not comprise diagnostic accuracy.
The team examined abdominal CTs from 104 patients seen in the emergency room for acute abdominal complaints over a four month period. The team created a sophisticated compression algorithm and evaluated a series of different compression ratios after application of the algorithm. The following compression ratios were considered: 10:1, 12.5:1, and 15:1.
After compression, studies were transferred to a workstation for review by three different independent radiologists. Studies compressed at different ratios were then compared for diagnostic accuracy for the abdominal CT as a whole, rather than one specific organ system.
The experienced radiologists could view the images that were compressed in the three different iterations or the uncompressed image. They were restricted to a five minute window for interpretation of each study. In addition to a presumptive diagnosis, radiologists also scored any artifact observed and confidence in their diagnosis.
The correct primary diagnosis was most frequent with the uncompressed image, 91.3 percent. But it only decreased marginally with compression; 90.5 percent for 1:10, 89 percent for 1: 12.5, and 90.1 percent for 1:15.
Diagnostic confidence, similarly, was in the same general range for each four groups and was actually highest in the 1:10 compression at 4.87 on scale to five. The lowest mean confidence rate was for the 1:12.5 group, at 4.77 out of five. Compression artifact was most often noted in the areas which are most homogeneous, such as in the mesenteric and retroperiteneal fat, liver, and spleen.
August 14, 2008 Related topics: Facilities, Imaging, IT & software