Spotiflow

Spotiflow achieves subpixel-accurate spot detection in microscopy images by formulating the task as a multi-scale heatmap and stereographic flow regression problem.

Key Features:

• Deep stereographic flow regression: Spotiflow formulates spot detection as a multi-scale heatmap and stereographic flow regression problem to achieve subpixel-accurate localizations.
• 2D and 3D compatibility: Spotiflow can be used for 2D images and 3D volumetric stacks.
• Cross-condition generalization: Spotiflow can be trained to generalize across different imaging conditions, tissue types and chemical preparations.
• Computational efficiency: Spotiflow is substantially more time- and memory-efficient than existing methods.

Scientific Applications:

• Imaging-based spatial transcriptomics (iST): Spotiflow enables precise detection of millions of transcripts in images with low signal-to-noise ratio, leading to meaningful improvements in biological insights obtained from iST experiments.
• Live imaging: Spotiflow enhances accuracy in live imaging experiments, improving the biological insights obtained.

Methodology:

Spotiflow formulates spot detection as a multi-scale heatmap and stereographic flow regression problem using deep learning.