Artificial Intelligence (AI)-driven Laser Microdissection (LMD) microscope for spatial 'omics

Over the past few years spatially resolved, high-dimensional molecular analysis of tissue architecture has revolutionised our insight into how cellular heterogeneity regulates development, homeostasis and disease. In 2020 Spatial Transcriptomics was selected as "Method of the Year" by Nature Methods, highlighting the enormous potential of integrating spatial with 'omics-level molecular data.

Transcriptomics methods using in situ hybridisation/barcoding/sequencing have been at the vanguard of the field and turnkey systems have been developed by companies such as 10x Genomics and NanoString. More recently, artificial intelligence (AI) assisted cell identification and characterisation, coupled with automated laser microdissection (LMD), has transformed our ability to isolate and capture single cells from tissue sections, thus enabling ex-situ methods for specimen characterisation.

LMD used to be a labour-intensive low throughput method where Regions of Interest (ROI) had to be identified by the user, cut out by manually directing the laser and capturing the section in single tubes. Cutting-edge instruments can now perform these tasks autonomously. Machine-learning-based image processing is first used to identify and classify cells in images taken by slide scanners.

The software then identifies individual cells or clusters of cells to be captured and characterised ex situ by 'omics methods. Finally, molecular and spatial data are integrated to generate a spatially resolved map of molecular characteristics that can be further interrogated.

This bid brings together four facilities at the Institute of Genetics and Cancer (IGC), the Advanced Imaging Resource (AIR), the Host and Tumour Profiling Unit (HTPU) and the IGC Mass spectrometry facility (IGC-MS) and Edinburgh Pathology (EP). Together, we have the expertise to operate and implement such a technically demanding sample processing and analysis pipeline.

In addition, all four facilities collaborate with the IGC Bioinformatics core which will support data processing and integration.

The system will be initially used by a consortium of 14 research groups from Edinburgh and Glasgow to investigate cellular/tissue/tumour heterogeneity in human and murine tissues. The work we are doing will shed light on how mutational, protein, metabolite and transcript dynamics are spatially regulated. Importantly, we wish to interrogate how changes at the cellular level shape the response of the tissue or organism to environmental challenges, tumour progression and inflammation.