Out-of-distribution evaluations of channel agnostic masked autoencoders in fluorescence microscopy
This work addresses the problem of model generalization for researchers in fluorescence microscopy and high-content screening, offering a more interpretable evaluation method, though it is incremental in improving existing autoencoder approaches.
The paper tackled the challenge of evaluating computer vision models in high-content screening under distribution shifts by proposing an evaluation scheme that isolates specific sources of shift using the JUMP-CP dataset, and introduced Campfire, a channel-agnostic masked autoencoder that generalizes to out-of-distribution experimental batches, perturbagens, and fluorescent markers, achieving successful transfer learning across cell types.
Developing computer vision for high-content screening is challenging due to various sources of distribution-shift caused by changes in experimental conditions, perturbagens, and fluorescent markers. The impact of different sources of distribution-shift are confounded in typical evaluations of models based on transfer learning, which limits interpretations of how changes to model design and training affect generalisation. We propose an evaluation scheme that isolates sources of distribution-shift using the JUMP-CP dataset, allowing researchers to evaluate generalisation with respect to specific sources of distribution-shift. We then present a channel-agnostic masked autoencoder $\mathbf{Campfire}$ which, via a shared decoder for all channels, scales effectively to datasets containing many different fluorescent markers, and show that it generalises to out-of-distribution experimental batches, perturbagens, and fluorescent markers, and also demonstrates successful transfer learning from one cell type to another.