Skin Cancer Classification Using ConvNeXtLarge Architecture
Prithwish Raymahapatra, Avijit Kumar Chaudhuri, Sulekha Das · 2024
The objective of this research work is to classify skin cancer images into two different classes using convolutional neural networks (CNNs) with ConvNeXtLarge architecture. The two classes are benign and malignant. The skin cancer image dataset with 3,297 images used for this research is publicly available on https://www.w3.org/1999/xlink" xlink:href=" https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781003429609/10ac561d-a21e-4bf9-ae1e-93930025dfd0/content/www.kaggle.com "> kaggle.com . The methodology followed in this project includes data preprocessing, model building, and evaluation. The dataset is preprocessed by resizing the images to 224 × 224 and normalizing the pixel values. The ConvNeXtLarge architecture is used to build the CNN model and it is trained on the preprocessed data for 25 epochs with a batch size of 64 and a learning rate of 0.0009. The model is evaluated using the area under the receiver operating characteristic curve (AUC) metric. The results of this project show that the CNN model with ConvNeXtLarge architecture achieves an AUC of 0.91 for classifying skin cancer images into two different classes. In conclusion, the CNN model with ConvNeXtLarge architecture is an effective approach to classifying skin cancer images into two different classes. The model achieves an accuracy of 0.91 in classifying different types of skin cancer that could potentially help in the early diagnosis and treatment of skin cancers.