Continuous Magnification Training Improves Embedding Quality in Histopathological Self-Supervised Learning

Current histopathological foundation models are trained on discrete standard microscope magnifications (0.25, 0.5, 1.0, 2.0 microns per pixel). We use the unsupervised RankMe met ric to show that this can affect embedding space quality at magnifications outside their training distribution, with rank scores dropping at inter mediate scales. We introduce continuous mag nification training, where patches are sampled from a continuous distribution during training, and show that this eliminates the irregularities in the embedding space.

BIFOLD AUTHORS

Alexander Möllers