Growth threshold for pseudo labeling and pseudo label dropout for semi-supervised medical image classification

Open Webpage

Shaofeng Zhou, Shengwei Tian, Long Yu, Weidong Wu, Dezhi Zhang, Zhen Peng, Zhicheng Zhou

Published: 01 Jan 2024, Last Modified: 11 Apr 2025Eng. Appl. Artif. Intell. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Novel semi-supervised medical image classification framework with three loss functions proposed.•Growth Threshold for Pseudo Labeling introduced, improving utilization of unlabeled samples.•Pseudo Label Dropout proposed to alleviate category imbalance.•GTPL & PLD evaluated on ISIC 2018 & 2019, showing improved performance.