Deep Neural Networks on the Accuracy of Skin Disease Diagnosis in Non-Dermatologists

Effect of Using Deep Neural Networks on the Accuracy of Skin Disease Diagnosis in Non-Dermatologist Physician

Background: Deep neural networks (DNN) has been applied to many kinds of skin diseases in experimental settings.

Objective: The objective of this study is to confirm the augmentation of deep neural networks for the diagnosis of skin diseases in non-dermatologist physicians in a real-world setting.

Methods: A total of 40 non-dermatologist physicians in a single tertiary care hospital will be enrolled. They will be randomized to a DNN group and control group. By comparing two groups, the investigators will estimate the effect of using deep neural networks on the diagnosis of skin disease in terms of accuracy.

Study Overview

• Status: Terminated
• Conditions: Skin Diseases
• Intervention / Treatment: Diagnostic test: Model Dermatology (deep neural networks; Build 2020)

Detailed Description

In the DNN group and control group, these steps are the same process.

1. Routine exam and capture photographs of skin lesions for all eligible consecutive series patient.
2. Make a clinical diagnosis (BEFORE-DX)
3. Make a clinical diagnosis (AFTER-DX)
4. consult to dermatologist

In the DNN group, after making the BEFORE-DX, physicians use deep neural networks and make an AFTER-DX considering the results of the deep neural networks (Model Dermatology, build 2020).

In the control group, after making the BEFORE-DX, physicians make an AFTER-DX after reviewing the pictures of skin lesions once more.

Ground truth will be based on the biopsy if available, or the consensus diagnosis of the dermatologists.

The investigators will compare the accuracy between the DNN group and control group after 6 consecutive months study.

• Study Type: Interventional
• Enrollment (Actual): 55
• Phase: Not Applicable

Contacts and Locations

Study Locations

• Korea, Republic of
  • Seoul, Korea, Republic of, 03080
    • Seoul National University Hospital

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: ADULT; OLDER_ADULT; CHILD
• Accepts Healthy Volunteers: No
• Genders Eligible for Study: All

Description

Inclusion Criteria:

• non-dermatologist physician (residents) who agree to participate in this study

Exclusion Criteria:

• dermatology residents
• non-dermatology residents who use other deep neural networks for skin lesion diagnosis

Study Plan

How is the study designed?

Design Details

• Primary Purpose: DIAGNOSTIC
• Allocation: RANDOMIZED
• Interventional Model: PARALLEL
• Masking: NONE

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
EXPERIMENTAL: DNN group; using deep neural networks for skin lesion diagnosis	Diagnostic test: Model Dermatology (deep neural networks; Build 2020); Physicians in the DNN group take pictures of the skin lesion and use the algorithm by uploading pictures.
NO_INTERVENTION: Control group; conventional diagnosis

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Top-1 diagnostic accuracy	frequency of correct Top-1 prediction	6 consecutive months

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Top-2 and 3 diagnostic accuracy	frequency of correct Top-2 and 3 prediction	6 consecutive months
Infection sensitivity	positive rate of infection diagnosis	6 consecutive months
Malignancy sensitivity	Positive rate of malignancy diagnosis	6 consecutive months

Collaborators and Investigators

• Sponsor: Pyoeng Gyun Choe

Publications and helpful links

General Publications

• Wang P, Liu X, Berzin TM, Glissen Brown JR, Liu P, Zhou C, Lei L, Li L, Guo Z, Lei S, Xiong F, Wang H, Song Y, Pan Y, Zhou G. Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised study. Lancet Gastroenterol Hepatol. 2020 Apr;5(4):343-351. doi: 10.1016/S2468-1253(19)30411-X. Epub 2020 Jan 22. Erratum In: Lancet Gastroenterol Hepatol. 2020 Apr;5(4):e3.
• Lin H, Li R, Liu Z, Chen J, Yang Y, Chen H, Lin Z, Lai W, Long E, Wu X, Lin D, Zhu Y, Chen C, Wu D, Yu T, Cao Q, Li X, Li J, Li W, Wang J, Yang M, Hu H, Zhang L, Yu Y, Chen X, Hu J, Zhu K, Jiang S, Huang Y, Tan G, Huang J, Lin X, Zhang X, Luo L, Liu Y, Liu X, Cheng B, Zheng D, Wu M, Chen W, Liu Y. Diagnostic Efficacy and Therapeutic Decision-making Capacity of an Artificial Intelligence Platform for Childhood Cataracts in Eye Clinics: A Multicentre Randomized Controlled Trial. EClinicalMedicine. 2019 Mar 17;9:52-59. doi: 10.1016/j.eclinm.2019.03.001. eCollection 2019 Mar.
• Liu Y, Jain A, Eng C, Way DH, Lee K, Bui P, Kanada K, de Oliveira Marinho G, Gallegos J, Gabriele S, Gupta V, Singh N, Natarajan V, Hofmann-Wellenhof R, Corrado GS, Peng LH, Webster DR, Ai D, Huang SJ, Liu Y, Dunn RC, Coz D. A deep learning system for differential diagnosis of skin diseases. Nat Med. 2020 Jun;26(6):900-908. doi: 10.1038/s41591-020-0842-3. Epub 2020 May 18.
• Han SS, Park I, Eun Chang S, Lim W, Kim MS, Park GH, Chae JB, Huh CH, Na JI. Augmented Intelligence Dermatology: Deep Neural Networks Empower Medical Professionals in Diagnosing Skin Cancer and Predicting Treatment Options for 134 Skin Disorders. J Invest Dermatol. 2020 Sep;140(9):1753-1761. doi: 10.1016/j.jid.2020.01.019. Epub 2020 Mar 31.
• Sellheyer K, Bergfeld WF. A retrospective biopsy study of the clinical diagnostic accuracy of common skin diseases by different specialties compared with dermatology. J Am Acad Dermatol. 2005 May;52(5):823-30. doi: 10.1016/j.jaad.2004.11.072.
• Cui X, Wei R, Gong L, Qi R, Zhao Z, Chen H, Song K, Abdulrahman AAA, Wang Y, Chen JZS, Chen S, Zhao Y, Gao X. Assessing the effectiveness of artificial intelligence methods for melanoma: A retrospective review. J Am Acad Dermatol. 2019 Nov;81(5):1176-1180. doi: 10.1016/j.jaad.2019.06.042. Epub 2019 Jun 27.
• Tschandl P, Codella N, Akay BN, Argenziano G, Braun RP, Cabo H, Gutman D, Halpern A, Helba B, Hofmann-Wellenhof R, Lallas A, Lapins J, Longo C, Malvehy J, Marchetti MA, Marghoob A, Menzies S, Oakley A, Paoli J, Puig S, Rinner C, Rosendahl C, Scope A, Sinz C, Soyer HP, Thomas L, Zalaudek I, Kittler H. Comparison of the accuracy of human readers versus machine-learning algorithms for pigmented skin lesion classification: an open, web-based, international, diagnostic study. Lancet Oncol. 2019 Jul;20(7):938-947. doi: 10.1016/S1470-2045(19)30333-X. Epub 2019 Jun 12.

Study record dates

Study Major Dates

• Study Start (ACTUAL): November 27, 2020
• Primary Completion (ACTUAL): November 27, 2021
• Study Completion (ACTUAL): December 27, 2021

Study Registration Dates

• First Submitted: November 13, 2020
• First Submitted That Met QC Criteria: November 13, 2020
• First Posted (ACTUAL): November 19, 2020

Study Record Updates

• Last Update Posted (ACTUAL): October 27, 2022
• Last Update Submitted That Met QC Criteria: October 25, 2022
• Last Verified: October 1, 2022

More Information

Terms related to this study

• Keywords: skin diseases; deep neural networks
• Additional Relevant MeSH Terms: Skin Diseases
• Other Study ID Numbers: 2020-3233

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No