Diagnostic performance on briefly presented digital pathology images

Abstract

Background:Identifying new and more robust assessments of proficiency/expertise (finding new “biomarkers of expertise”) in histopathology is desirable for many reasons. Advances in digital pathology permit new and innovative tests such as flash viewing tests and eye tracking and slide navigation analyses that would not be possible with a traditional microscope. The main purpose of this study was to examine the usefulness of time-restricted testing of expertise in histopathology using digital images.Methods:19 novices (undergraduate medical students), 18 intermediates (trainees), and 19 experts (consultants) were invited to give their opinion on 20 general histopathology cases after 1 s and 10 s viewing times. Differences in performance between groups were measured and the internal reliability of the test was calculated.Results:There were highly significant differences in performance between the groups using the Fisher's least significant difference method for multiple comparisons. Differences between groups were consistently greater in the 10-s than the 1-s test. The Kuder–Richardson 20 internal reliability coefficients were very high for both tests: 0.905 for the 1-s test and 0.926 for the 10-s test. Consultants had levels of diagnostic accuracy of 72% at 1 s and 83% at 10 s.Conclusions:Time-restricted tests using digital images have the potential to be extremely reliable tests of diagnostic proficiency in histopathology. A 10-s viewing test may be more reliable than a 1-s test. Over-reliance on “at a glance” diagnoses in histopathology is a potential source of medical error due to over-confidence bias and premature closure.

The Holistic Processing Account of Visual Expertise in Medical Image Perception: A Review

“…Moreover, another source of evidence supporting the key role of parafoveal/peripheral processing is the remarkable ability of medical experts to extract relevant information from a briefly presented image. For example, expert radiologists were able to detect 70% of abnormalities when chest films were displayed for only 200 ms ( Kundel and Nodine, 1975 ; for similar findings see Kundel and Nodine, 1975 ; Carmody et al, 1980a , 1981 ; Oestmann et al, 1988 ; Gale et al, 1990 ; Mugglestone et al, 1995 ; Evans et al, 2013 , 2016a ; Jaarsma et al, 2014 ; Houghton et al, 2015 ), and experts were capable of detecting some nodules that were 15° away from their point of fixation ( Carmody et al, 1980a ). Importantly, the brief exposure conditions in these studies do not provide sufficient time for the radiologists to move their eyes to the locations of the abnormalities, thereby ruling out alternative explanations of the results (such as the notion that medical expertise solely reflects foveal processing).…”

“…Empirical support for the importance of the initial glimpse includes the previously mentioned findings that experts can rapidly detect abnormalities in briefly presented images ( Kundel and Nodine, 1975 ; for similar findings see Carmody et al, 1980a , 1981 ; Oestmann et al, 1988 ; Gale et al, 1990 ; Mugglestone et al, 1995 ), as well as the findings that experts display efficient scanpaths with rapid times to the first fixation on the abnormality ( Krupinski, 1996a , b , 2000a , 2005 ; Nodine et al, 1996a , b ; Nodine and Kundel, 1997 ; Kundel and Nodine, 2004 ; Kundel et al, 2007 , 2008 ; Donovon and Litchfield, 2013 ; Wood et al, 2013 ). Of course, there are limits to the amount of information that can be extracted from the expert’s initial glance at an image, as shown by the finding that experts were at chance levels when asked to localize the abnormality under brief exposure conditions, even though they could detect the presence or absence of an abnormality at above chance levels ( Evans et al, 2013 ), findings that subtle or less conspicuous abnormalities require more time and/or foveal processing to be detected (e.g., Carmody et al, 1981 ; Oestmann et al, 1988 ) findings that diagnostic accuracy decreases as distance from the fovea increases ( Carmody et al, 1980a ), and findings that diagnostic accuracy is substantially higher under conditions that permit longer viewing times relative to brief exposure conditions (e.g., Kundel and Nodine, 1975 ; Oestmann et al, 1988 ; Mugglestone et al, 1995 ; Houghton et al, 2015 ).…”

Dermatopathology education in the era of modern technology

“…More than just a solid teaching modality (either as an online database or integration into a formalized atlas), VM has the unique capability to serve as an important research tool in furthering educational endeavors in DP and pathology. Notably, assessment of learning can be extended to investigate how users “interact” with digital images—evaluating slide exploration strategies, perceptual information gathering, and cognitive decision‐making of both trainees and expert pathologists . As histopathology is a visual discipline at its core, the skilled interpretation of slides is dependent on the honing of visual perception and processing, and pattern recognition skills over time.…”

“…In this way, VM can help to identify “biomarkers of expertise” and more objectively assess differences in image search patterns between novices and skilled pathologists—for example, which perceptual and cognitive mechanisms are active, in addition to how visual expertise is acquired and improved over time. By incorporating “tracking tools” with VM technologies, educators can more effectively monitor their trainees’ progress, and enhance learning through focused instruction and feedback . Simplistically, experts develop a “recognize/detect‐then‐search” approach to slide/image review—“after quick recognition of relevant patterns, the rest of the image is scanned for anything that contradicts schema.” Compared with novices, who typically use “search‐then‐detect” methods, experts are known to show different types of eye movements, spend more time at lower magnifications, fixate less on irrelevant features, show fewer panning and zooming movements, and make quicker and more accurate diagnoses .…”

The Use of Screencasts with Embedded Whole-Slide Scans and Hyperlinks to Teach Anatomic Pathology in a Supervised Digital Environment

“…Multiple studies in our laboratory and others have shown that similar diagnoses can be obtained using digital and light microscopy and that digital microscopy can be effectively used for primary surgical pathology diagnosis, telepathology, teaching, and other applications. [222324252627282930] Digital pathology systems offer a variety of logistical advantages, as they can streamline the flow and storage of cases, help reduce clerical errors, shorten turnaround time, and provide other operational efficiencies. These have stimulated an increasing interest in using them to replace the time-tested method of distributing glass slides throughout a laboratory for examination with light microscopes.…”