opdoc

Our Skin Cancer Algorithm is trained on the MNIST: HAM 10000 Dataset, a large collection of ten thousand multi-source dermatoscopic images classifying 7 different types of common pigmented skin lesions.

 

The data is originally sourced from Harvard Dataverse. More than 50% of lesions in our dataset are confirmed through histopathology and the rest have been validated through follow-up examination, expert consensus, or in-vivo confocal microscopy. 

 

We have utilized the PyTorch library as well as a ResNet 50 pretrained CNN to build our model.