Classification and Segmentation of Vulvovaginal Candidiasis in Microscopic Leucorrhea Images Based on Combined Deep Learning Model
Yiyao Ma, Yifei Xu, Wei Li
Vulvovaginal Candidiasis (VVC) is a common and serious gynecological disease. Early diagnosis and treatment are of great significance to women's health. However, most hospitals still use manual diagnosis method, which is not only inefficient but also unstable. This paper proposes a VVC image classification and recognition method based on computer vision and deep learning. Our models can greatly reduce the workload of doctors and improve detection efficiency and stability.After testing on 480 samples, our model has reached 92\% accuracy, 93\% recall and 97\%AUC with 23M parameters. The overall performance is superior to the best baseline model that we obtain 93\% accuracy, 92\% recall and 96\%AUC with 56M parameters. Besides, we are the first known paper to propose detection targets for pathogenic bacteria, using different colored rectangles to encircle different types of bacteria.
Wednesday 6th July
Poster Session 1.1 - onsite 15:20 - 16:20, virtual 11:00 - 12:00 (UTC+2)