Anouk Van Der Gijp MD, PhD, Presenter: Nothing to Disclose

Cecile Ravesloot MD, Abstract Co-Author: Nothing to Disclose

Josephine Huige, Abstract Co-Author: Nothing to Disclose

Irene van der Schaaf, Abstract Co-Author: Nothing to Disclose

Koen L. Vincken PhD, Abstract Co-Author: Nothing to Disclose

Jan P.J. van Schaik MD, PhD, Abstract Co-Author: Nothing to Disclose

Marieke Van Der Schaaf, Abstract Co-Author: Nothing to Disclose

Olle Ten Cate, Abstract Co-Author: Nothing to Disclose

In current practice radiologists interpret digital images, including a substantial amount of volume datasets. We hypothesize that interpretation of volume datasets demands different cognitive skills than the interpretation of two-dimensional (2D) cross-sectional images. This study aims to investigate and compare the cognitive processes occurring during interpretation of volume datasets versus 2D images.

Twenty radiology clerks of a Dutch university medical center were asked to think aloud while reading four to five volume CT datasets and 2D CT images (a selection of cross sectional slices). Cases were presented using a digital program, that allows for volume dataset viewing in different planes and contrast settings. Participants were asked to formulate a (differential) diagnosis. Thoughts verbalised by the subjects were registered and coded by a previously constructed framework of sixteen knowledge and skill elements, arranged in three main components: perception, analysis and synthesis (the latter includes generating a differential diagnosis and giving advice). A within-subject analysis with Friedman and Wilcoxon Signed Rank tests was performed to compare knowledge and skills used during volume dataset readings versus 2D readings.

In general, most of the utterances concerned perceptual knowledge and skills (46%). A smaller part involved synthesis (31%) and analysis (23%). During the interpretation of volume datasets, the largest part of utterances was perceptual (50%), which was significantly larger than in 2D image interpretation (37%), χ2= 16.2, p<.001, T=1, p<.001. In contrast, during 2D image interpretation, synthesis represented the largest part of utterances (41%), significantly larger than in volume dataset interpretation (26%), χ2= 16.2, p<0.001, T=1, p<0.001. No significant differences were found in the proportion of analysis during volume dataset and 2D image interpretation (22% and 23% respectively).

Volume dataset interpretation draws predominantly on perceptual processes while 2D image interpretation is mainly characterised by synthesis.

The results encourage the use of volume datasets for teaching and testing perceptual skills, while 2D images of cross sectional studies could be sufficient for educational purposes concerning the ability to generate a differential diagnosis or give advice.

Van Der Gijp, A, Ravesloot, C, Huige, J, van der Schaaf, I, Vincken, K, van Schaik, J, Van Der Schaaf, M, Ten Cate, O, Differences in Knowledge and Skills used for Interpretation of Radiologic Volume Datasets Compared to 2D Images.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045695.html