{"cells":[{"metadata":{},"cell_type":"markdown","source":"
\n

SIIM-ISIC Melanoma - Analysis, EDA and model

\n \"Melanoma\n
\n

Hello everyone,\n In this notebook, we will carry out the EDA of the SIIM-ISIC Melanoma Classification challenge. We shall first study why this is an important topic of research and what's at stake in this competition. We will then carry out an exploratory data analysis. Finally, I will show you how to setup a running pipeline with PyTorch so that you can quickly make experiments.

\n
\n

I'll continuously update this notebook over the next few weeks.

\n
\n

I'd like to give credit to:\n

\n

\n
\n

Sources:\n

\n

","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"
\n

Table of Contents

\n 1. What is a melanoma?1\n 2. Why is this competition important? 2\n 3. Evaluation metric3\n 4. EDA - Overview4 \n 5. EDA - Meta-features5 \n 6. EDA - Target distribution6 \n 7. EDA - Correlation7 \n 8. EDA - Image visualization8 \n 9. Starter model - EfficientNet B19 \n
","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"

1. What is a melanoma?

\n\n

1.1. A quick recap': what cancer actually is...

\n\n

Cancer is the name given to a collection of related diseases. In all types of cancer, some of the body’s cells begin to divide without stopping and spread into surrounding tissues.
\n
\nCancer can start almost anywhere in the human body, which is made up of trillions of cells. Normally, human cells grow and divide to form new cells as the body needs them. When cells grow old or become damaged, they die, and new cells take their place.
\n
\n\"Cancer\n \n
\nWhen cancer develops, however, this orderly process breaks down. As cells become more and more abnormal, old or damaged cells survive when they should die, and new cells form when they are not needed. These extra cells can divide without stopping and may form growths called tumors.

","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"

1.2. Skin cancer and melanoma

\n\n

As stated above, cancer can develop in any part of the body. Skin cancer is one of the most wide-spread cancer found in global population.

\n\nSkin cancer is the out-of-control growth of abnormal cells in the epidermis, the outermost skin layer, caused by unrepaired DNA damage that triggers mutations. These mutations lead the skin cells to multiply rapidly and form malignant tumors. The main types of skin cancer are basal cell carcinoma (BCC), squamous cell carcinoma (SCC), melanoma and Merkel cell carcinoma (MCC).

\n\nAs you have guessed, for this competition, we will be interested in detecting melanoma.

\n\nThe two main causes of skin cancer are the sun’s harmful ultraviolet (UV) rays and the use of UV tanning machines. Often, the doctor may even detect the growth at a precancerous stage, before it has become a full-blown skin cancer or penetrated below the surface of the skin.\n\n

","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"

1.3. Melanoma, in 4 questions

\n\n\"4

\n\n

\n \nWhat is it? Melanoma is a cancer that develops from melanocytes, the skin cells that produce melanin pigment, which gives skin its color.

\n\nWhere is it usually found? Melanomas often resemble moles and sometimes may arise from them. They can be found on any area of the body, even in areas that are not typically exposed to the sun.

\n\nHow many people get it? In 2019, more than 192,000 new cases of melanoma are expected to occur in the U.S., about 96,000 of which will be invasive.

\n\nHow serious is it? Melanoma is the most dangerous of the three most common forms of skin cancer. Melanomas can be curable when caught and treated early. In 2019, melanoma is projected to cause about 7,200 deaths.

\n \n

","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"

2. Why is this competition important?

\n \n> Skin cancer is the most prevalent type of cancer. Melanoma, specifically, is responsible for 75% of skin cancer deaths, despite being the least common skin cancer. The American Cancer Society estimates over 100,000 new melanoma cases will be diagnosed in 2020. It's also expected that almost 7,000 people will die from the disease. As with other cancers, early and accurate detection—potentially aided by data science—can make treatment more effective.\n\n

As stated above, this disease can be deadly and an early and accurate diagnosis is critical to enhance the probability of survival of the patient.

\n\n> In this competition, you’ll identify melanoma in images of skin lesions. In particular, you’ll use images within the same patient and determine which are likely to represent a melanoma. Using patient-level contextual information may help the development of image analysis tools, which could better support clinical dermatologists.\n\n

This competition aims to classify skin moles into melanomas or simple moles. To do so, you'll have to develop a model that takes into account contextual information about the patient.

\n ","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"

3. Evaluation metric

\n \n

The evaluation metric is the ROC-AUC score as expected in medical imaging competition, and in healthcare competition broadly speaking. Since the dataset is likely to be highly imbalanced (more non-cancerous moles than cancerous ones) we want a metric that takes into account this imbalance.

\n \nFor more information on Recall, Precision, False Positive, True Negatives and ROC-AUC score, please check the links hereafter:\n