Padmini Gopalan MBBS, Presenter: Nothing to Disclose

Amaka Offiah MBBS, PhD, Abstract Co-Author: Nothing to Disclose

Visual tracking, by demonstrating 1) how visual knowledge is acquired over time and 2) which visual information is useful for decision-making will allow a dramatic reduction in the time taken to train novice analysts and increase the efficiency of practicing experts. Track development of expertise during training and identify critical interventions. Modify the training programme as indicated.

Radiologists of different years of experience, speciality radiologists will be assessed. Phase 1 study consisted of a computerised experimental task to confirm that abnormalities can be identified from images in a manner consistent with clinical practice. 150 paediatric musculoskeletal radiographs (133 abnormal, 17 normal) were selected and classified into easy, medium and difficult. Participants were given six options as to how confident they felt about the presence/absence of an abnormality. If there were abnormalities they were asked to click the position(s) on the image. Distance from the centre of the identified abnormality was used to score performance

So far, accuracy was higher and location error lower amongst consultants than trainees. On average the consultants were faster than the novices to make their initial decision about the presence of an abnormality (14.1 sec Vs 18.9 Sec), made fewer incorrect abnormality identifications (0.2 Vs 0.5) and were more accurate at locating an abnormality (average distance to centre of reference location 23.6 pixels Vs 55.9).

The consultants showed a quicker decision time but longer location time, suggesting that having identified an abnormality, they double check the review areas. This is something that Phase 2 of the study will reveal.

Radiologists must detect subtle fractures that indicate the abuse has taken place. A previous ROC study states that in suspected abuse, diagnostic accuracy of fracture detection was generally low and appeared to be affected more by observer related factors. 2005 UK consultant radiologists’ survey showed significant dissatisfaction with the training and service provided at that time. There is potential to increase numbers of radiology child abuse experts by 27% if given improved training and support. How expert levels of performance manifest in different eye movement patterns and decision times can be studied in view of implementing in teaching.

Gopalan, P, Offiah, A, Development of Radiological Expertise—A Visual Tracking Experiment do the Radiologists’ Eyes See what their Brains Ought to See?.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045564.html