45 Dermatological and Surgical Assistance Program Computer Vision News In our last webinar, we covered several topics about annotations regarding the integration of non-experts knowledge, the trade-off between detailed expert annotations and their cost and how to build a medical image dataset for skin lesion classification. Andre Pacheco, an Assistant Professor at the Federal University of Espírito Santo (UFES), presented PAD-UFES-20 dataset. The Dermatological and Surgical Assistance Program (in Portuguese: Programa de Assistência Dermatologica e Cirurgica - PAD) at UFES is a non-profit program that provides free skin lesion treatment, in particular, to low-income people who cannot afford private treatment. Due to historical factors, the state of Espírito Santo witnessed an influx of thousands of European immigrants during the 19th century. Given Brazil's tropical climate, many of these immigrants and their descendants did not acclimatize well to this environment. Consequently, there is a notable prevalence of skin lesions and cancer in this region, and the PAD plays a pivotal role in providing support to these individuals. Andre explained the challenges for building the dataset: ❖ Convince the doctors to collaborate. ❖ Design and develop applications to collect and store sensible data. ❖ Train doctors and students to use the app. ❖ Coordinate the data collection. ❖ Data quality selection. To learn more about the PAD-UFES project, the skin lesion dataset, and the challenges they’veovercome; check out Andre’s talk in the video above. Veronika and Amelia had three successful editions so far (in February, June, and September 2023) with 10 speakers in total. The videos are available on their YouTube playlist.
RkJQdWJsaXNoZXIy NTc3NzU=