针对白血病诊断中白细胞分类难题，提出多阶段微调与头多样性集成方法。
Multi-Stage Fine-Tuning of Pathology Foundation Models with Head-Diverse Ensembling for White Blood Cell Classification
- 分阶段微调DINOBloom-base模型，训练多种分类头以实现类别特异性优化。
- 不同分类头在不同成熟阶段表现最优：MLP头对早幼粒细胞F1达0.733。
- 通过头多样性集成提升精度，识别出标注错误或形态模糊的疑难样本。
从外周血涂片中分类白血球对白血病诊断至关重要。然而，自动化方法仍面临类别不平衡、域偏移及形态连续性混淆等挑战，相邻成熟阶段特征细微重叠。本文提出针对WBCBench 2026挑战（ISBI 2026）的13类白细胞分类多阶段微调方法。最优模型为微调后的DINOBloom-base，其上训练了线性、余弦及多层感知机（MLP）三类分类头。余弦头在成熟粒细胞边界表现最佳（带状中性粒细胞BNE F1=0.470），线性头在较不成熟粒细胞类中表现更优（杆状核粒细胞MMY F1=0.585），MLP头在最不成熟粒细胞中表现最佳（早幼粒细胞PMY F1=0.733），揭示了类别特异性专长。基于此，构建头多样性集成：以MLP头为主预测器，仅当其他两头一致时替换预设四组混淆对中的预测结果。进一步发现，所有模型均误判的案例显著富集于可能的标注错误或固有形态模糊情况。
原文摘要 · Abstract (English)
The classification of white blood cells (WBCs) from peripheral blood smears is critical for the diagnosis of leukemia. However, automated approaches still struggle due to challenges including class imbalance, domain shift, and morphological continuum confusion, where adjacent maturation stages exhibit subtle, overlapping features. We present a multi-stage fine-tuning methodology for 13-class WBC classification in the WBCBench 2026 Challenge (ISBI 2026). Our best-performing model is a fine-tuned DINOBloom-base, on which we train multiple classifier head families (linear, cosine, and multilayer perceptron (MLP)). The cosine head performed best on the mature granulocyte boundary (Band neutrophil (BNE) F1 = 0.470), the linear head on more immature granulocyte classes (Metamyelocyte (MMY) F1 = 0.585), and the MLP head on the most immature granulocyte (Promyelocyte (PMY) F1 = 0.733), revealing class-specific specialization. Based on this specialization, we construct a head-diverse ensemble, where the MLP head acts as the primary predictor, and its predictions within the four predefined confusion pairs are replaced only when two other head families agree. We further show that cases consistently misclassified by all models are substantially enriched for probable labeling errors or inherent morphological ambiguity.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。