Artificial intelligence algorithm for the histopathological diagnosis of skin cancer

Victor Antonio Kuiava, Eliseu Luiz Kuiava, Eduardo Ottobelli Chielle, Francisco Madalosso De Bittencourt


Introduction: Cutaneous neoplasms are the most common cancers in the world, and have high morbidity rates. A definitive diagnosis can only be obtained after histopathological evaluation of the lesions. Objective: To develop an artificial intelligence program to establish the histopathological diagnosis of cutaneous lesions. Methodology: A deep learning program was built using three neural network architectures: MobileNet, Inception and convolutional networks. A database was constructed using 2732 images of melanomas, basal and squamous cell carcinomas, and normal skin. The validation set consisted of 284 images from all 4 categories, allowing for the calculation of sensitivity and specificity. All images were provided by the Path Presenter website. Results: The sensitivity and specificity of the MobileNet model were 92% (95%CI, 83-100%) and 97% (95%CI, 90-100%), respectively; corresponding figures for the Inception model were 98.3% (95%CI, 86-100%) and 98.8% (95%CI, 98.2-100%); lastly, the sensitivity and specificity of the convolutional network model were 91.6% (95%CI, 73.8-100%) and 95.7% (95%CI, 94.4-97.2%). The maximum sensitivity for the differentiation of malignant conditions was 91%, and specificity was 95.4%. Conclusion: The program developed in the present study can efficiently distinguish between the main types of skin cancer with high sensitivity and specificity.


Skin lesions, Artificial Intelligence, Diagnosis, Neoplasms, Melanoma.

Full Text:


Copyright (c) 2021 Clinical & Biomedical Research

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.

ISSN: 2357-9730



Apoio Financeiro:


Licença Creative Commons
The Clinical & Biomedical Research is licenced under Creative Commons Atribuição 4.0 Internacional.