A Comparison of the Concordance of Digital Images, Live Examinations, and Skin Biopsies for the Diagnosis of Hospitalized Dermatology Consultation Patients

Noah Scheinfeld, MD, JD, Jeremy Kurz, MD, Ellen Teplitz, MD

Disclosures

Skinmed. 2003;2(1) 

In This Article

Discussion

Previous studies have evaluated the ability of dermatologists to provide reliable diagnoses and recommend appropriate treatment by examining images rather than patients in outpatient settings. Interobserver and intraobserver agreement have been assessed by Lowitt et al.[3] and Whited et al.[4] Generally, in these studies, concordance of live and digital differential diagnoses averages 80% but ranges to almost 95%. Our study's finding of a 76% concordance between a resident's differential diagnosis (up to three diagnoses) was based on a review of digital images with definitive diagnostic procedures such as skin biopsies and potassium hydroxide preparations. This rate of concordance, when subjective clinical examinations and objective data are compared, is consonant with previous findings. It should of course be noted that only more difficult cases were biopsied so this lower rate was not expected. While the 81% rate of concordance of the resident's live and digital diagnoses is of less certain import because the resident's review of digital images was followed by live examination, its rate is in accordance with previous studies.

Patients accepted the use of digital imaging to record their skin examinations. Only two patients refused imaging; the other patients were not imaged because the student imager was not available. This level of acceptance mirrors other studies. One group has found that 85% of patients felt comfortable using a video link for consultation.[5]

Our study highlights the ease of setting up and using digital images for diagnosing patients. In this study, we used the Nikon Coolpix 950 camera, which could be obtained at the time of this study (March, 2000) at a retail cost of approximately $600 and provided resolution of up to 2.11 million pixels. As of August, 2002, three-megapixel cameras were available for $500. The images were stored and viewed as a standard JPEG algorithm. JPEGs can be viewed with image viewers that are components of Windows ME, NT, or 2000. They may also be viewed with Internet browsers such as Microsoft Internet Explorer (Microsoft Corp., Redmond, WA).

Our level of image compression (1:4) and limited resolution, (an 800 x 600 pixels of laptop video monitor or a digital camera monitor with an effective resolution of 640 x 480 pixels), did not seem to affect our digital diagnostic ability. These findings are consonant with the findings of others. Resolution of 640 x 480 pixels is sufficient for the purposes of making digital dermatologic diagnoses.[6] Previous studies have shown that compression of dermatologic images by JPEG or fractal image format (FIF, up to 1:40) compression algorithms did not result in decreased diagnostic performance.[7]

Our study is notable in that the data and image collection occurred in the hospital, where lighting and other conditions were variable rather than in one clinic-based location. Initial patient contact occurred in hospital wards, the emergency department, or diagnostic testing areas. Our results indicate that it is feasible to obtain quality images in a variety of lighting and environmental situations without expert or extensive knowledge of photographic techniques. Wherever possible we tried to acquire multiple image sets of patients, a technique that has been used by Kvedar et al.[8]

The skin conditions encountered in hospitalized patients differ in many respects from those found in outpatient clinics. Hospitalized patients have many generalized vascular reaction pattern rashes (e.g., drug rashes, urticaria). In our study, digitally diagnosing such rashes resulted in the greatest variances between digital and definitive diagnosis, such as in a drug rash that was misdiagnosed as xerosis or an eczematous process (Figure 3). Digital images do not always provide a full sense of a generalized rash's scope, extent, and texture. As our study progressed, the ability of the student image-taker to capture clinically useful images improved, especially as multiple images were captured and such generalized rashes were diagnosed digitally with greater accuracy. Rashes that were discrete and three-dimensional rarely resulted in variations between digital, live, and definitive diagnoses. Rashes where digital, live, and definitive diagnosis almost always concurred included trichotillomania (Figure 4), skin tags, keloids and psoriasis, dermatofibromas, and warts. This finding is at variance with some studies in outpatient teledermatology, where discrete lesions such as basal cell cancers can present challenges for digital diagnosis.[9]

Image taken early in the study which was digitally diagnosed as xerosis or eczema and diagnosed by live examination and skin biopsy as a drug rash

Trichotillomania with diagnostic concordance of digital and live examinations

The digital examiner's eye needs to be retrained to see subtle palpable objects in two dimensions perhaps partially by using other visual cues. For example, urticaria can be accurately interpreted if the eye is trained to always look for a zone of pallor associated with erythema. Even trained dermatologists may have difficulty in assessing generalized rashes and complex dermatologic reaction patterns and their live clinical diagnoses are sometimes not borne out until after histologic examination.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....