UKSSL: Underlying Knowledge Based Semi-Supervised Learning for Medical Image Classification

Abstract

Deep learning techniques have made significant progress in medical image analysis. However, obtaining ground truth labels for unlabeled medical images is challenging as they often outnumber labeled images. Thus, training a high-performance model with limited labeled data has become a crucial challenge. Methods: This study introduces an underlying knowledge-based semi-supervised framework called UKSSL, consisting of two components: MedCLR extracts feature representations from the unlabeled dataset; UKMLP utilizes the representation and finetunes it with the limited labeled dataset to classify the medical images. Results: UKSSL evaluates on the LC25000 and BCCD datasets, using only 50% labeled data. It gets precision, recall, F1-score, and accuracy of 98.9% on LC25000 and 94.3%, 94.5%, 94.3%, and 94.1% on BCCD, respectively. These results outperform other supervised-learning methods using 100% labeled data. Conclusions: The UKSSL can efficiently extract underlying knowledge from the unlabeled dataset and perform better using limited labeled medical images.

Efficient spine segmentation network based on multi‐scale feature extraction and multi‐dimensional spatial attention

“…Thus, our research progress is influenced by these practical constraints. In the future studies, a shift in training methods from supervised to weakly supervised or even unsupervised is imperative 44–46 . We need to intensify research in this direction to reduce reliance on extensively annotated data.…”

CGS‐Net: A classification‐guided framework for automated infection segmentation of COVID‐19 from CT images

“…The article provides a rigorous definition and comprehensive description of weakly supervised learning, introducing several commonly used weakly supervised learning algorithms. Ren et al 35 also introduced a knowledge‐based semi‐supervised learning algorithm for medical image classification. This algorithm leverages unlabeled data for pre‐training and incorporates a knowledge‐based distillation technique to transfer knowledge from the teacher model to the student model.…”

A systematic review of deep learning methods for the classification and segmentation of prostate cancer on magnetic resonance images

“…13 Contrastive learning techniques, including UKSSL (Underlying Knowledge-based Semi Supervised Learning), enable the extraction of valuable insights from limited labeled datasets, thereby enhancing the robustness and generalizability of classification models. 14 By harnessing the power of DL, these models not only enhance diagnosis accuracy but also facilitate personalized treatment planning and therapeutic decision-making in PCa management. Overall, DL-based classification techniques represent a paradigm shift in PCa research, offering unprecedented opportunities for improving patient outcomes and advancing the field of oncology.…”

“…The integration of DL architectures such as CNNs and transfer learning algorithms like EfficientNet, DenseNet, and Xception has significantly improved classification accuracy rates, particularly in analyzing complex 3D MRI scans for detecting malignant lesions 13 . Contrastive learning techniques, including UKSSL (Underlying Knowledge‐based Semi Supervised Learning), enable the extraction of valuable insights from limited labeled datasets, thereby enhancing the robustness and generalizability of classification models 14 . By harnessing the power of DL, these models not only enhance diagnosis accuracy but also facilitate personalized treatment planning and therapeutic decision‐making in PCa management.…”