Class Imbalanced

Class imbalance, where one or more classes in a dataset are significantly under-represented compared to others, poses a major challenge in machine learning, particularly for medical image analysis. Current research focuses on developing techniques to mitigate the bias towards majority classes, employing methods such as novel loss functions (e.g., contour-weighted loss, data imbalance loss), self-supervised learning, ensemble methods, and metric learning integrated into two-stage frameworks. These advancements aim to improve the accuracy and robustness of models in applications like medical image segmentation and classification, leading to more reliable diagnoses and improved healthcare outcomes.