Spatial Transcriptomics As Rasterized Image Tensors (STARIT) characterizes cell states with subcellular molecular heterogeneity.

STARIT reveals subcellular heterogeneity enables identification of cell states beyond gene counts in imSRT data.

• Why it matters: Understanding subcellular molecular variation is crucial for accurately defining cell states, but current methods overlook this detail, limiting biological insights.
• What they did: The approach converts transcript data into image-based tensors, allowing deep learning models to analyze subcellular localization patterns in both simulated and real datasets.
• The result: STARIT successfully distinguishes cell types and states based on subcellular transcript localization, offering a standardized framework to uncover heterogeneity missed by traditional gene count methods.