Novelty Detection of Leukocyte Image via Mean-Shifted Feature and Directly Optimized Subspace
Wei Li, Taotao Lai, Guanghai Liu, Haoyi Fan, Zuoyong Li · IEEE Transactions on Instrumentation and Measurement · 2023
Novelty detection of leukocyte images aims to learn effective data description from in-distribution leukocyte samples and detect out-of-distribution ones that deviate from the expected patterns. Recently, the methods of fine-tuning on the pretrained model have demonstrated excellent performance in novelty detection. However, those methods are prone to collapse or feature deterioration due to the training data of limited categories and tiny differences in medical images such as leukocyte images. To alleviate the above issues, we propose an unsupervised feature adaptation method based on the mean-shifted feature (MSF) and the directly optimized subspace (SUB) for leukocyte novelty detection, named mean-shifted subspace (SubMSF). Specifically, we first use MSFs to alleviate the catastrophic collapse when fine-tuning the pretrained model via contrastive learning. Then, we combine the MSF with the SUB to recognize tiny differences between leukocytes. Compared with several state-of-the-art methods, the results on real-world leukocyte datasets demonstrate the effectiveness of the proposed SubMSF.