Melanoma Detection with Uncertainty Quantification
This work addresses early melanoma detection for medical applications, but it is incremental as it builds on existing methods with dataset integration and calibration.
The paper tackled melanoma detection by combining multiple datasets and applying uncertainty quantification to classifiers, achieving 93.2% accuracy before and 97.8% after uncertainty-based rejection, with a 40.5% reduction in misdiagnoses.
Early detection of melanoma is crucial for improving survival rates. Current detection tools often utilize data-driven machine learning methods but often overlook the full integration of multiple datasets. We combine publicly available datasets to enhance data diversity, allowing numerous experiments to train and evaluate various classifiers. We then calibrate them to minimize misdiagnoses by incorporating uncertainty quantification. Our experiments on benchmark datasets show accuracies of up to 93.2% before and 97.8% after applying uncertainty-based rejection, leading to a reduction in misdiagnoses by over 40.5%. Our code and data are publicly available, and a web-based interface for quick melanoma detection of user-supplied images is also provided.