Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.
Who will be affected?
Readers: No access to all 28 journals. We recommend accessing our articles via PubMed Central
Authors: No access to the submission form or your user account.
Reviewers: No access to your user account. Please download manuscripts you are reviewing for offline reading before Wednesday, July 01, 2020 at 7:00 PM.
Editors: No access to your user account to assign reviewers or make decisions.
Copyeditors: No access to user account. Please download manuscripts you are copyediting before Wednesday, July 01, 2020 at 7:00 PM.
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Image Generation of Common Dermatological Diagnoses by Artificial Intelligence; Evaluation of the Potential for Education and Training Purposes
Sarah Kooper-Johnson;
Subin Lim;
Jumana Aldhalaan;
Gabriela Cobos;
Vijay Kodumudi;
Daniel S Loo;
Joshua Mervis;
Madalyn Petrillo;
Sarah N Robinson;
Bichchau Michelle Nguyen
ABSTRACT
The integration of artificial intelligence (AI) into dermatology holds promise for education and diagnostic purposes, particularly through image generation, which has not been well studied. Assess whether AI image generation software can generate accurate images of classic dermatological conditions and whether they are recognizable as computer-generated. Images of ten dermatologic conditions were generated from DALLE-2 and DALLE-3 programs. These images were randomized amongst clinical photographs and distributed to dermatology residents and attending physicians. Participants were instructed to: (1) identify AI-generated images and (2) provide their diagnosis. AI-generated images were detected as computer-generated in 70.8% of cases. Correct diagnoses were made based on all AI images 40.83% of the time. This was significantly lower than the 72.0% recognition rate for clinical photographs (p<0.001). DALLE-2 images were diagnosed correctly less frequently (25.0%) than DALLE-3 images (56.6%) (p<0.001). AI generated images of common dermatological conditions are becoming more accurate. This holds great implications for education but should be used with caution as further research is needed with more advanced, specific, and inclusive training data. Limitations include the use of AI image generators created by a single parent company as well as the use of a limited set of diagnoses.
Citation
Please cite as:
Kooper-Johnson S, Lim S, Aldhalaan J, Cobos G, Kodumudi V, Loo DS, Mervis J, Petrillo M, Robinson SN, Nguyen BM
Image Generation of Common Dermatological Diagnoses by Artificial Intelligence: Evaluation Study of the Potential for Education and Training Purposes