PAD-UFES-20: a skin lesion dataset composed of patient data and clinical images collected from smartphones

Pacheco, André G. C.; Lima, Gustavo R.; Salomão, Amanda S.; Krohling, Breno A.; Biral, Igor Peixoto; Angelo, Gabriel G. de; Alves, Fábio C. R.; Esgario, José G. M.; Simora, Alana C.; Castro, Pedro B. C.; Rodrigues, Felipe B.; Frasson, P; Krohling, Renato A.; Knidel, Helder; Santos, Maria Carolina Salustino dos; Santo, Rachel Bertolani do Espírito; Macedo, Telma L. S. G.; Canuto, Tania R. P.; Barros, Luíz F. S. de

doi:10.48550/arxiv.2007.00478

Cited by 1 publication

(2 citation statements)

References 17 publications

(22 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…) reconfirms the recent findings that FedAvg significantly outperforms the local models with relative improvement (RI) up to19.74% (see Local/FixMatch vs. FedAvg). Interestingly, both lower FedAvg and FedAvg ‡ (SSFL) outperform the local SSL, and the local upper bound, respectively.…”

supporting

confidence: 88%

“…Dataset. We validated FedPerl on 38,000 skin lesion images coming from four publicly available datasets, namely ISIC19 [4] which consists of 25K images with 8 classes of melanoma (MEL), melanocytic nevus (NV), basal cell carcinoma (BCC), actinic keratosis (AK), benign keratosis (BKL), dermatofibroma (DF), the vascular lesion (VASC), and squamous cell carcinoma (SCC); HAM [30] with 10K images (7 classes); Derm7pt [13] with 1K images (6 classes), and PAD-UFES [19] with 2K images (6 classes). The databases are distributed to ten clients besides the global model, cf.…”

Section: Experiments and Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Semi-Supervised Federated Peer Learning for Skin Lesion Classification

Bdair¹,

Navab²,

Albarqouni³

2021

Preprint

View full text Add to dashboard Cite

Skin cancer is one of the most deadly cancers worldwide. Yet, it can be reduced by early detection. Recent deep-learning methods have shown a dermatologist-level performance in skin cancer classification. Yet, this success demands a large amount of centralized data, which is oftentimes not available. Federated learning has been recently introduced to train machine learning models in a privacy-preserved distributed fashion demanding annotated data at the clients, which is usually expensive and not available, especially in the medical field. To this end, we propose FedPerl, a semi-supervised federated learning method that utilizes peer learning from social sciences and ensemble averaging from committee machines to build communities and encourage its members to learn from each other such that they produce more accurate pseudo labels. We also propose the peer anonymization (PA) technique as a core component of FedPerl. PA preserves privacy and reduces the communication cost while maintaining the performance without additional complexity. We validated our method on 38,000 skin lesion images collected from 4 publicly available datasets. FedPerl achieves superior performance over the baselines and state-of-the-art SSFL by 15.8%, and 1.8% respectively. Further, FedPerl shows less sensitivity to noisy clients.

show abstract

supporting

confidence: 88%