Manual differential leukocyte counting is a critical yet time-consuming and observer-dependent process in clinical hematology. This study presents a comparative analysis of You Only Look Once v11 (YOLOv11) and Vision Transformer (ViT) architectures for the classification of 14 leukocyte types and artifacts using a private clinical dataset. We further investigated the impact of HistAuGAN, a domain-specific data augmentation strategy designed to simulate real-world staining variability. Across experimental settings, ViT models achieved higher overall performance than YOLOv11 variants, and the application of HistAuGAN led to systematic improvements in both architectural families. The best-performing configuration, trained on the HistAuGAN-augmented dataset, achieved a macro F1-score of 98.36% and an overall accuracy of 99.75% on the validation set. To assess generalization capacity, this configuration was additionally evaluated on the public PBC and LISC datasets, demonstrating meaningful cross-dataset performance without architectural modification. Model interpretability was examined through attention- and activation-based saliency analyses, indicating that predictions were primarily driven by morphologically relevant leukocyte regions rather than background structures. These findings suggest that combining global-context modeling with domain-informed augmentation provides a robust and clinically coherent framework for fine-grained leukocyte classification.
Manual differential leukocyte counting is a critical but time-consuming and subjective process. This study provides a rigorous comparative analysis of You OnlyLook Once v11 (YOLOv11) and Vision Transformer (ViT) architectures for classifying 14 types of leukocytes and artifacts from a private clinical dataset. We also evaluated the impact of HistAuGAN, a data augmentation technique that simulates real world staining variability, by training and evaluating models from both families with and without its application. The results demonstrated the consistent superiority of the ViT architecture over YOLOv11 in all experimental settings. Furthermore, the use of HistAuGAN promoted a universal and significant performance improvement across all tested models. The top performing configuration, the ViT-Base model trained with HistAuGAN, achieved a macro F1-Score of98.36% and an overall accuracy of 99.75% on the test set. We conclude that the synergy between an architecture capable of learning global features (ViT) and adomain-specific data augmentation technique that addresses practical challenges represents a state-of-the-art strategy, establishing a new performance benchmark for high granularity leukocyte classification and reinforcing the potentialof artificial intelligence to transform diagnostic hematology.