ShapeUpLMD: Automated spatial optimization of laser microdissection for tissue sampling

Abstract Laser microdissection (LMD) enables enrichment of defined cellular populations from heterogeneous tissues, providing histologically-resolved molecular profiles with spatial resolution. However, the absence of standardized methods for generating optimized regions of interest (ROIs) limits reproducibility and scalability for spatial multi-omic workflows. To address this gap, we developed ShapeUpLMD, an open-source software tool and integrated workflow that automates the optimization of spatially defined ROIs for LMD, directly from digital pathology annotations. Fresh-frozen uterine serous carcinoma (USC) tumors (n = 11) were sectioned onto polyethylene naphthalate slides, scanned and subsequently annotated by an expert pathologist. Tumor and non-tumor ROIs were used to train tumor-specific classifiers in HALO (Indica Labs). Classified tumor or unbiased whole tissue ROIs were refined and optimized by ShapeUpLMD prior to automated collection on a Leica LMD7 microscope. Across the cohort, predicted tumor ROIs increased effective tumor purity by 86 ± 6% relative to whole tissue while maintaining high spatial concordance between predicted and collected ROIs (accuracy = 0.9 ± 0.05). Data-independent acquisition mass spectrometry quantified upwards of ∼6,500 proteins across spatially resolved ROIs, revealing regionally coherent clustering of adjacent tumor regions and abundance patterns consistent with tumor and non-tumor cell admixture in an unbiased spatial proteomic application. ShapeUpLMD provides an automated, reproducible, and scalable framework that bridges digital pathology with LMD enabling high-fidelity spatial enrichment for multi-omic analyses. This workflow increases throughput, reduces inter-operator variability, and supports standardized, regionally resolved tissue collection for spatial systems biology applications. The software is available at https://github.com/GYNCOE/ShapeUpLMD .