tees3r / Automated-Melanoma-Classification

Classifying skin-lesions to be either malign or benign with convolutional neural networks.

Automated Melanoma Classification

This implementation was built as part of my bachelor thesis at the University of Jena, Germany. The focus of the work was to build a classifier which is able to distinguish skin lesions between being malign or benign with the help of a (deep) convolutional neural network.

Datasets

The major part of my used data can be obtained from the ISIC Archive with the help of the ISIC-Data-Downloader.

Acknowledgement and Resources

Special thanks to my advisor Prof. Dr. Joachim Denzler, University of Jena.

• Dermatologist-level classification of skin cancer with deep neural networks, Nature volume 542, pages 115–118 (02 February 2017)
• Vision-Based Classification of Skin Cancer using Deep Learning
• Skin Lesion Analysis towards Melanoma Detection Using Deep Learning Network
• Deep Learning with Python by François Chollet
• AMBOSS Medical Dictionary
• Keras and TensorFlow
• A lot more included in the thesis itself..