Wenderoth Laura, Asemissen Anne-Marie, Modemann Franziska, Nielsen Maximilian, Werner René
Institute for Applied Medical Informatics, University Medical Center Hamburg-Eppendorf, Christoph-Probst-Weg 1, 20251 Hamburg, Germany; Institute of Computational Neuroscience, University Medical Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany; Center for Biomedical Artificial Intelligence (bAIome), University Medical Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany.
II. Department of Medicine, University Medical Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany.
Comput Methods Programs Biomed. 2025 Mar;260:108560. doi: 10.1016/j.cmpb.2024.108560. Epub 2024 Dec 9.
Classification of peripheral blood and bone marrow cells is critical in the diagnosis and monitoring of hematological disorders. The development of robust and reliable automatic classification systems is hampered by data scarcity and limited model generalizability across laboratories. The present study proposes the integration of self-supervised learning (SSL) into cell classification pipelines to address these challenges.
The experiments are based on four public hematological single cell image datasets: one bone marrow and three peripheral blood datasets. The cell classification pipeline consists of two parts: (1) SSL-based image feature extraction without the use of image annotations, and (2) a lightweight machine learning classifier applied to the SSL features and trained on only a small number of annotated images.
Direct transfer of SSL models trained on bone marrow data to peripheral blood data resulted in higher balanced classification accuracy than the transfer of supervised deep learning counterparts for all blood datasets. After adaptation of the lightweight machine learning classifier with 50 labeled samples per class of the new dataset, the SSL pipeline surpasses supervised deep learning classification performance for one dataset and classes with rare or atypical cell types and performs similarly on the other datasets.
The results demonstrate that SSL enables (1) extraction of meaningful cell image features without the use of cell class information; (2) efficient transfer of knowledge between bone marrow and peripheral blood cell domains; and (3) efficient model adaptation to new datasets using only a few labeled data samples.
外周血细胞和骨髓细胞的分类在血液系统疾病的诊断和监测中至关重要。数据稀缺以及跨实验室模型通用性有限阻碍了强大且可靠的自动分类系统的发展。本研究提出将自监督学习（SSL）整合到细胞分类流程中以应对这些挑战。
实验基于四个公开的血液学单细胞图像数据集：一个骨髓数据集和三个外周血数据集。细胞分类流程由两部分组成：（1）基于SSL的图像特征提取，无需使用图像标注；（2）一个轻量级机器学习分类器，应用于SSL特征，并仅在少量标注图像上进行训练。
在所有血液数据集上，将在骨髓数据上训练的SSL模型直接转移到外周血数据上，比转移监督深度学习模型具有更高的平衡分类准确率。在使用新数据集每个类别50个标记样本对轻量级机器学习分类器进行适配后，SSL流程在一个数据集以及具有罕见或非典型细胞类型的类别上超越了监督深度学习分类性能，在其他数据集上表现相似。
结果表明，SSL能够（1）在不使用细胞类别信息的情况下提取有意义的细胞图像特征；（2）在骨髓和外周血细胞领域之间有效地转移知识；（3）仅使用少量标记数据样本就能够有效地使模型适应新数据集。
