Spatial transcriptomics analyses often lose a substantial fraction of molecular signals because standard cell segmentation and deconvolution methods fail to capture expression from fragile cell types, subcellular structures, or extracellular regions. This study introduces RESCUE, a computational approach that recovers these unattributed spatial expression patterns, revealing biologically meaningful signals across diverse datasets and enabling more complete and accurate tissue interpretation.
- Young Joo Lee
- Seokjin Yeo
- Sihai Dave Zhao