NUMERO: #20230978

Título: CONVNEXT FOR THE CLASSIFICATION OF ORAL POTENTIALLY MALIGNANT DISORDERS AND SQUAMOUS CELL CARCINOMA

Nome do Apresentador: Anna Luiza Damaceno ARAUJO

Categoria do Trabalho: Painel de pesquisa científica (PPC)

Área Temática: Estomatologia

Resumo: Objective: To implement a Deep Learning model for automatic classification of clinical photographs into potentially malignant and malignant lesions.
Study Design: A dataset of 527 clinical images from three institutions was used to train (n=519) and validate (n=58) a ConvNext. The CNN was pre-trained on the ImageNet dataset and fine-tuned with the training subset, with the Adam optimizer at a learning rate of 0.0001.
Results: The model reached a mean accuracy of 0.98, precision of 0.98, recall of 0.98 and F1score of 0.98 during training. The internal validation reached a mean accuracy of 0.81, precision of 0.80, recall of 0.82 and F1score of 0.80.
Conclusion: These preliminary results demonstrate a great performance of a Deep Learning algorithm for classifying potentially malignant and malignant lesions, which can be a premise of an innovative non-invasive screening tool for cancer detection. Further steps will seek to extend the dataset and perform external test to better evaluate the model´s generalization capabilities.

Autor 1:  Anna Luiza Damaceno ARAUJO

E-mail 1:  [email protected]

Autor 2:   Eduardo Santos Carlos DE SOUZA

E-mail 2:  [email protected]

Autor 3 :  Cristina SALDIVIA-SIRACUSA

E-mail 3:  [email protected]

Autor 4:  Marcio Ajudarte LOPES

E-mail 4:  [email protected]

Autor 5:  André Carlos Ponce de Leon Ferreira DE CARVALHO

E-mail 5:  [email protected]

Autor 6:  Alan Roger SANTOS-SILVA

E-mail 6:  [email protected]

Autor 7:  Luiz Paulo KOWALSKI

E-mail 7:  [email protected]