TY - JOUR
T1 - The CASH (color, architecture, symmetry, and homogeneity) algorithm for dermoscopy
AU - Henning, J. Scott
AU - Dusza, Stephen W.
AU - Wang, Steven Q.
AU - Marghoob, Ashfaq A.
AU - Rabinovitz, Harold S.
AU - Polsky, David
AU - Kopf, Alfred W.
N1 - Funding Information:
Supported by the Stavros S. Niarchos Foundation, The Rahr Family Foundation.
PY - 2007/1
Y1 - 2007/1
N2 - Background: The color, architecture, symmetry, and homogeneity (CASH) algorithm for dermoscopy includes a feature not used in prior algorithms, namely, architecture. Architectural order/disorder is derived from current concepts regarding the biology of benign versus malignant melanocytic neoplasms. Objective: We sought to evaluate the accuracy of the CASH algorithm. Methods: A total CASH score (TCS) was calculated for dermoscopic images of 325 melanocytic neoplasms. Sensitivity, specificity, diagnostic accuracy, and receiver operating characteristic curve analyses were performed by comparing the TCS with the histopathologic diagnoses for all lesions. Results: The mean TCS was 12.28 for melanoma, 7.62 for dysplastic nevi, and 5.24 for nondysplastic nevi. These differences were statistically significant (P < .001). A TCS of 8 or more yielded a sensitivity of 98% and specificity of 68% for the diagnosis of melanoma. Limitations: This is a single-evaluator pilot study. Additional studies are needed to verify the CASH algorithm. Conclusions: The CASH algorithm can distinguish melanoma from melanocytic nevi with sensitivity and specificity comparable with other algorithms. Further study is warranted to determine its intraobserver and interobserver correlations.
AB - Background: The color, architecture, symmetry, and homogeneity (CASH) algorithm for dermoscopy includes a feature not used in prior algorithms, namely, architecture. Architectural order/disorder is derived from current concepts regarding the biology of benign versus malignant melanocytic neoplasms. Objective: We sought to evaluate the accuracy of the CASH algorithm. Methods: A total CASH score (TCS) was calculated for dermoscopic images of 325 melanocytic neoplasms. Sensitivity, specificity, diagnostic accuracy, and receiver operating characteristic curve analyses were performed by comparing the TCS with the histopathologic diagnoses for all lesions. Results: The mean TCS was 12.28 for melanoma, 7.62 for dysplastic nevi, and 5.24 for nondysplastic nevi. These differences were statistically significant (P < .001). A TCS of 8 or more yielded a sensitivity of 98% and specificity of 68% for the diagnosis of melanoma. Limitations: This is a single-evaluator pilot study. Additional studies are needed to verify the CASH algorithm. Conclusions: The CASH algorithm can distinguish melanoma from melanocytic nevi with sensitivity and specificity comparable with other algorithms. Further study is warranted to determine its intraobserver and interobserver correlations.
UR - http://www.scopus.com/inward/record.url?scp=33845694252&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=33845694252&partnerID=8YFLogxK
U2 - 10.1016/j.jaad.2006.09.003
DO - 10.1016/j.jaad.2006.09.003
M3 - Article
C2 - 17190620
AN - SCOPUS:33845694252
SN - 0190-9622
VL - 56
SP - 45
EP - 52
JO - Journal of the American Academy of Dermatology
JF - Journal of the American Academy of Dermatology
IS - 1
ER -