TY - JOUR
T1 - Performance of a computer-aided digital dermoscopic image analyzer for melanoma detection in 1,076 pigmented skin lesion biopsies
AU - del Rosario, Francis
AU - Farahi, Jessica M.
AU - Drendel, Jesse
AU - Buntinx-Krieg, Talayesa
AU - Caravaglio, Joseph
AU - Domozych, Renee
AU - Chapman, Stephanie
AU - Braunberger, Taylor
AU - Dellavalle, Robert P.
AU - Norris, David A.
AU - Fathi, Ramin
AU - Alkousakis, Theodore
N1 - Publisher Copyright:
© 2017 American Academy of Dermatology, Inc.
PY - 2018/5
Y1 - 2018/5
N2 - Background: Digital dermoscopic image analysis of pigmented skin lesions (PSLs) has become increasingly popular, despite its unclear clinical utility. Unbiased, high-powered studies investigating the efficacy of commercially available systems are limited. Objective: To investigate the diagnostic performance of the FotoFinder Mole-Analyzer in assessing PSLs for cutaneous melanoma. Methods: In this 15-year retrospective study, the histopathologies of 1076 biopsied PSLs among a total of 2500 imaged PSLs were collected. The biopsied PSLs were categorized as benign or malignant (cutaneous melanoma) based on histopathology. Analyzer scores (0-1.00) for these PSLs were obtained and grouped according to histopathology. Results: At an optimized cutoff score of 0.50, a sensitivity of 56% and a specificity of 74% were achieved. The area under the receiver operating characteristics curve was 0.698, indicating poor accuracy as a diagnostic tool. Limitations: This study had a retrospective design and involved only a single institution. Conclusion: Our study reveals a low sensitivity of the scoring function of this digital dermoscopic image analyzer for detecting cutaneous melanomas. Physicians must apply keen clinical judgment when using such devices in the screening of suspicious PSLs.
AB - Background: Digital dermoscopic image analysis of pigmented skin lesions (PSLs) has become increasingly popular, despite its unclear clinical utility. Unbiased, high-powered studies investigating the efficacy of commercially available systems are limited. Objective: To investigate the diagnostic performance of the FotoFinder Mole-Analyzer in assessing PSLs for cutaneous melanoma. Methods: In this 15-year retrospective study, the histopathologies of 1076 biopsied PSLs among a total of 2500 imaged PSLs were collected. The biopsied PSLs were categorized as benign or malignant (cutaneous melanoma) based on histopathology. Analyzer scores (0-1.00) for these PSLs were obtained and grouped according to histopathology. Results: At an optimized cutoff score of 0.50, a sensitivity of 56% and a specificity of 74% were achieved. The area under the receiver operating characteristics curve was 0.698, indicating poor accuracy as a diagnostic tool. Limitations: This study had a retrospective design and involved only a single institution. Conclusion: Our study reveals a low sensitivity of the scoring function of this digital dermoscopic image analyzer for detecting cutaneous melanomas. Physicians must apply keen clinical judgment when using such devices in the screening of suspicious PSLs.
KW - computer-aided diagnosis
KW - diagnostic accuracy
KW - digital dermoscopy
KW - image analysis
KW - melanoma
KW - pigmented skin lesions
KW - screening
KW - skin cancer
KW - test sensitivity
UR - http://www.scopus.com/inward/record.url?scp=85056315196&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85056315196&partnerID=8YFLogxK
U2 - 10.1016/j.jaad.2017.01.049
DO - 10.1016/j.jaad.2017.01.049
M3 - Article
C2 - 29678380
AN - SCOPUS:85056315196
SN - 0190-9622
VL - 78
SP - 927-934.e6
JO - Journal of the American Academy of Dermatology
JF - Journal of the American Academy of Dermatology
IS - 5
ER -
