Deep convolutional generative adversarial networks in retinitis pigmentosa disease images augmentation and detection
Artykuł w czasopiśmie
MNiSW
100
Lista 2024
| Status: | |
| Autorzy: | Powroźnik Paweł, Skublewska-Paszkowska Maria, Nowomiejska Katarzyna, Aristidou Andreas , Panayides Andreas, Rejdak Robert |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Rok wydania: | 2025 |
| Wersja dokumentu: | Elektroniczna |
| Język: | angielski |
| Numer czasopisma: | 2 |
| Wolumen/Tom: | 19 |
| Strony: | 321 - 340 |
| Scopus® Cytowania: | 0 |
| Bazy: | Scopus | BazTech |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | NIE |
| Publikacja OA: | TAK |
| Licencja: | |
| Sposób udostępnienia: | Otwarte czasopismo |
| Wersja tekstu: | Ostateczna wersja opublikowana |
| Czas opublikowania: | W momencie opublikowania |
| Data opublikowania w OA: | 1 stycznia 2025 |
| Abstrakty: | angielski |
| Large medical datasets are crucial for advancing contemporary medical practices that incorporate computer vision and machine learning techniques. These records serve as indispensable resources for identifying patterns that assist healthcare professionals in diagnosing rare diseases and enhancing patient outcomes. Moreover, these datasets drive research into the causes and progression of such diseases, potentially leading to innovative therapeutic strategies. However, the acquisition of such data poses significant challenges due to privacy and ethical concerns, as well as the rarity of certain conditions. Therefore, it is imperative to both collect new medical data and develop tools that facilitate the enhancement of existing datasets while preserving the accurate characteristics of the diseases. This study focuses on leveraging deep convolutional generative adversarial networks (DCGAN) to expand a dataset containing images of retinitis pigmentosa, a rare eye condition affecting the retina. Our study showcases that integrating xtreme gradient boosting (XGBoost) within the DCGAN framework enhances the clarity and quality of these augmented images. By employing hybrid VGG16 alongside XGBoost techniques during training, we observe improvements in detection accuracy. The outcomes of the proposed method are highly promising, with the model achieving all key per- formance metrics surpassing the 90% threshold as well as improving baseline classification accuracy by almost 19%. |
