Amaka Offiah MBBS, Presenter: Nothing to Disclose

Lorraine Moon, Abstract Co-Author: Nothing to Disclose

Vera Mann, Abstract Co-Author: Nothing to Disclose

Christine Hall, Abstract Co-Author: Nothing to Disclose

Andrew E. Todd-Pokropek PhD, Abstract Co-Author: Nothing to Disclose

To assess the effects of edge enhancement and soft and hard copy display on digital image quality and diagnostic accuracy in suspected non-accidental injury.

50 radiographs exposed as part of post mortem skeletal surveys in 12 infants with suspected non-accidental injury formed the material for this study. The radiographs covered all skeletal sites. Images were obtained using a Fuji 5000R computed radiography system. Hard copies of each radiograph were printed with edge enhancement factors 0, 0.5 and 1.2. Images (edge enhancement 0.5) were also displayed on a 1K x 1K monitor. Six observers independently evaluated all 200 images (over four sessions) for the presence of fractures. Observers also scored each image (scale 1 = poor to 5 = excellent) for visualisation of soft tissues, visualisation of trabecular markings and overall image quality. Statistical analysis was performed on individual and pooled results of all observers. Linear regression (ANOVA) and location ROC analyses were used to compare quality scores and diagnostic accuracy of each display method. For the purposes of ROC analysis, histology results were taken as the gold standard.

Average image quality scores for edge enhancement 0, 0.5, 1.2 and PACS were 3.70, 3.75, 3.67 and 4.00 respectively. Differences between groups was significant between PACS and film (p = 0.003), but not between films at different levels of edge enhancement. Similar results were obtained for visualisation of soft tissues and visualisation of trabecular markings. The area under the ROC curve for edge enhancement 0 and PACS was 0.85 and 0.87 respectively. This difference was not significant (two tailed p level = 0.35, one tailed p level = 0.17).

Computed radiography images in infants with suspected non-accidental injury are best viewed in soft copy format with a small degree of edge enhancement. This knowledge is of importance both to those seeking, and to those giving expert opinion in such cases.

A.O.: Part funded by a grant from the Royal College of Radiology Pump Priming Fund

Offiah, A, Moon, L, Mann, V, Hall, C, Todd-Pokropek, A, Digital Imaging in Non-Accidental Injury: A ROC Study.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4410564.html