JMIR Publications

ABSTRACT

Background:

Skin Cancer is the most common cancer and is often ignored by people at an early stage. ​There are 5.4 million new cases of skin cancer worldwide every year. One in every three cancers diagnosed is skin cancer. Although melanomas represent fewer than 5% of all skin cancers, they account for approximately 75% of all skin-cancer-related deaths and are responsible for over 10,000 deaths annually. These can be prevented by early detection of the mole. Skin Cancer is significantly less in India due to the presence of eumelanin in India’s dark-skinned population which acts as a protection against the development of skin cancer. Still, Skin cancer constituted 3.18% of all patients registered with cancer. Of this BCC was 54.76% while 36.91% and malignant melanoma was only 8.33%. 88% of patients from the rural sector of which 92% were in the profession of agriculture.

Objective:

We propose a skin lesion classification system that has the ability to detect such moles at an early stage and is able to easily differentiate between a cancerous mole and non-cancerous mole. Using this system, we would be able to save time and resources for both patients and practitioners.

Methods:

We searched for research papers which may help us from Google Scholar, PubMed, Research Gate, and ISIC archive. We included the result in the literature survey.

Results:

We have created a deep convolutional neural network using InceptionV3 and DenseNet201 pre-trained model.

Conclusions:

Through extensive research conducted and through results obtained we draw upon the conclusion that deep learning algorithms are highly suitable for classifying skin cancer images. Also by using concepts of finetuning and ensemble learning model yielded better results. We found that fine-tuning the whole model helps model converge faster than compared to fine-tuning the top layers only hence giving overall better accuracy.