ML Anthology

Synthetic Generation of Dermatoscopic Images with GAN and Closed-Form Factorization

Rohan Reddy Mekala, Frederik Pahde, Simon Baur, Sneha Chandrashekar, Madeline Diep, Markus Wenzel, Eric L. Wisotzky, Galip Ümit Yolcu, Sebastian Lapuschkin, Jackie Ma, Peter Eisert, Mikael Lindvall, Adam A. Porter, Wojciech Samek
ECCVW 2024 pp. 368-384
doi:10.1007/978-3-031-91907-7_22 /eccvw/2024/mekala2024eccvw-synthetic/

Abstract

In the realm of dermatological diagnoses, where the analysis of dermatoscopic and microscopic skin lesion images is pivotal for the accurate and early detection of various medical conditions, the costs associated with creating diverse and high-quality annotated datasets have hampered the accuracy and generalizability of machine learning models. We propose an innovative semi-supervised augmentation solution that harnesses Generative Adversarial Network (GAN) based models and associated techniques over their latent space to generate controlled “semi-automatically-discovered” semantic variations in dermatoscopic images. We created synthetic images to incorporate the semantic variations and augmented the training data with these images. With this approach, we were able to increase the performance of machine learning models and set a new benchmark amongst non-ensemble based models in skin lesion classification on the HAM10000 dataset; and used the observed analytics and generated models for detailed studies on model explainability, affirming the effectiveness of our solution.

PDF ECCVW Semantic Scholar

Cite

Text

Mekala et al. "Synthetic Generation of Dermatoscopic Images with GAN and Closed-Form Factorization." European Conference on Computer Vision Workshops, 2024. doi:10.1007/978-3-031-91907-7_22

Markdown

[Mekala et al. "Synthetic Generation of Dermatoscopic Images with GAN and Closed-Form Factorization." European Conference on Computer Vision Workshops, 2024.](https://mlanthology.org/eccvw/2024/mekala2024eccvw-synthetic/) doi:10.1007/978-3-031-91907-7_22

BibTeX

@inproceedings{mekala2024eccvw-synthetic,
  title     = {{Synthetic Generation of Dermatoscopic Images with GAN and Closed-Form Factorization}},
  author    = {Mekala, Rohan Reddy and Pahde, Frederik and Baur, Simon and Chandrashekar, Sneha and Diep, Madeline and Wenzel, Markus and Wisotzky, Eric L. and Yolcu, Galip Ümit and Lapuschkin, Sebastian and Ma, Jackie and Eisert, Peter and Lindvall, Mikael and Porter, Adam A. and Samek, Wojciech},
  booktitle = {European Conference on Computer Vision Workshops},
  year      = {2024},
  pages     = {368-384},
  doi       = {10.1007/978-3-031-91907-7_22},
  url       = {https://mlanthology.org/eccvw/2024/mekala2024eccvw-synthetic/}
}