Towards Automated Melanoma Screening: Proper Computer Vision & Reliable Results
This work addresses the need for reliable and reproducible automated screening to assist professionals in diagnosing melanoma, an increasingly common and dangerous cancer, but it is incremental as it builds on existing computer vision approaches.
The paper tackles the problem of automated melanoma screening by analyzing current methods and proposing two novel computer vision techniques, achieving AUC scores of 84.6% and 89.3% compared to a baseline of 81.2%.
In this paper we survey, analyze and criticize current art on automated melanoma screening, reimplementing a baseline technique, and proposing two novel ones. Melanoma, although highly curable when detected early, ends as one of the most dangerous types of cancer, due to delayed diagnosis and treatment. Its incidence is soaring, much faster than the number of trained professionals able to diagnose it. Automated screening appears as an alternative to make the most of those professionals, focusing their time on the patients at risk while safely discharging the other patients. However, the potential of automated melanoma diagnosis is currently unfulfilled, due to the emphasis of current literature on outdated computer vision models. Even more problematic is the irreproducibility of current art. We show how streamlined pipelines based upon current Computer Vision outperform conventional models - a model based on an advanced bags of words reaches an AUC of 84.6%, and a model based on deep neural networks reaches 89.3%, while the baseline (a classical bag of words) stays at 81.2%. We also initiate a dialog to improve reproducibility in our community