Program

Poster Presentations

Day 2, June 11（Thu.）　　Room P (5F 501+502)

2P-28

PDF

A Pixel-UMAP Pipeline Toward Integrated Analysis of Targeted DESI-TQ-MSI Data

(¹Kyoto Univ., ²Kwansei Gakuin Univ.)
^oRyoma Kizu^1,2, Ikuko Yao², Yuki Sugiura¹

Conventional LC/IC–MS loses spatial context in developmental metabolomics. We developed pixel-UMAP for DESI-TQ-MSI to enable interpretable, low-dimensional mapping of pixel-level multivariate targeted metabolite images. The pipeline performs removal of off-tissue pixels, inter-section batch correction, and ROI-focused re-clustering followed by re-UMAP to resolve local tissue structures at higher resolution. Pixel-UMAP extracts organ- and compartment-specific metabolic signatures across multiple sections while reducing technical artifacts. Embedding ¹³C⁶-glucose tracing MSI into the same space visualized spatial domains with characteristic central carbon metabolism. The workflow supports systematic spatial metabolomics and comparative mapping across stages.

---