Radiology image perception and observer performance: How does expertise and clinical information alter interpretation? Stroke detection explored through eye-tracking

Abstract

Additional Information:• ABSTRACTHistorically, radiology research has been dominated by chest and breast screening. Few studies have examined complex interpretative tasks such as the reading of multidimensional brain CT or MRI scans. Additionally, no studies at the time of writing have explored the interpretation of stroke images; from novices through to experienced practitioners using eye movement analysis. Finally, there appears a lack of evidence on the clinical effects of radiology reports and their influence on image appraisal and clinical diagnosis. A computer-based, eye-tracking study was designed to assess diagnostic accuracy and interpretation in stroke CT and MR imagery. Eight predetermined clinical cases, five images per case, were presented to participants (novices, trainee, and radiologists; n=8). The presence or absence of abnormalities was rated on a five-point Likert scale and their locations reported. Half cases of the cases were accompanied by clinical information; half were not, to assess the impact of information on observer performance. Results highlight differences in visual search patterns amongst novice, trainee and expert observers; the most marked differences occurred between novice readers and experts. Experts spent more time in challenging areas of interest (AOI) than novices and trainee, and were more confident unless a lesion was large and obvious. The time to first AOI fixation differed by size, shape and clarity of lesion. 'Time to lesion' dropped significantly when recognition appeared to occur between slices. The influence of clinical information was minimal.

The assessment of stroke multidimensional CT and MR imaging using eye movement analysis: does modality preference enhance observer performance?

“…Experts were quicker to reach an AOI and spent less time in and around the AOI than novices and trainees in the most challenging cases, appearing to confirm the qualitative image analysis results demonstrating quicker, more accurate decisions and also confirming findings from our previous pilot study (Cooper et al, 2009). False positive decisions were characterised by more inconsistencies within and between groups in terms of eye-movements; either taking an unusually long time to first fixate within an AOI (e.g.…”

“…In agreement with a prior pilot study, experts appeared to be more consistent both within and between cases than novices and operated within a smaller time range than novice participants in every case (Cooper et al, 2009). Although there was slightly more variability between trainees and experts in terms of average time spent in and around AOIs and total slice time, trainees followed a similar visual search pattern to the experts between stroke types i.e.…”

“…Visual search studies in medical imaging have been published since the 1960's (Thomas & Lansdown,1963;Kundel, 1978;Nodine & Kundel, 1987), yet only recently have experimental studies been designed to track eye movements across a number of related medical images in radiology (e.g. Phillips et al, 2005Phillips et al, & 2008Cooper et al, 2009). …”

Review of prospects and challenges of eye tracking in volumetric imaging

“…By contrast, only a limited number of studies have aimed at shedding light on the relationship between visual search, perception, cognition, and performance in the context of volumetric images: five experiments have explored radiologists' reading strategies of stimulus material that was acquired by a volumetric imaging technique, [45][46][47][48][49] while five other studies looked at the influence of expertise on gaze behavior, [50][51][52][53][54] four explored the implications of different workstation configurations on gaze, [55][56][57][58][59] and one study examined the presence of inattentional blindness in the interpretation of chest CT cases. 60 Finally, one study looked at the visual search characteristics of readers when using computer-aided detection in CT colonography presented in the fly through mode.…”

Incidental Findings on Brain MRI: Insight from HUNT MRI - A General Population Study

“…The reasons are manifold, but include: 1) The HUNT MRI scan protocol: same comprehensive clinical protocol used for all participants; 2) All images were read by two senior neuroradiologists [17][18][19]; and 3) Inclusion of middle aged subjects, as prevalence of cerebrovascular disease and intracranial tumors increases with age [4,10].…”