Diffusion-Based Synthetic Brightfield Microscopy Images for Enhanced Single Cell Detection
This work addresses the challenge of limited annotated data for biological researchers using microscopy, offering a cost-effective augmentation method, though it is incremental as it applies existing diffusion techniques to a specific domain.
The paper tackled the problem of data scarcity and annotation bottlenecks in single cell detection for brightfield microscopy by generating synthetic images using a U-Net based diffusion model. The result showed that training object detection models like YOLOv8 with synthetic data improved detection accuracies, and a human expert survey indicated high realism with 50% accuracy in distinguishing synthetic from real images.
Accurate single cell detection in brightfield microscopy is crucial for biological research, yet data scarcity and annotation bottlenecks limit the progress of deep learning methods. We investigate the use of unconditional models to generate synthetic brightfield microscopy images and evaluate their impact on object detection performance. A U-Net based diffusion model was trained and used to create datasets with varying ratios of synthetic and real images. Experiments with YOLOv8, YOLOv9 and RT-DETR reveal that training with synthetic data can achieve improved detection accuracies (at minimal costs). A human expert survey demonstrates the high realism of generated images, with experts not capable to distinguish them from real microscopy images (accuracy 50%). Our findings suggest that diffusion-based synthetic data generation is a promising avenue for augmenting real datasets in microscopy image analysis, reducing the reliance on extensive manual annotation and potentially improving the robustness of cell detection models.