GIST: Guiding-Image Spatial Transcriptomics (SJ-21-0037)

St. Jude Reference #SJ-21-0037


Researchers at St. Jude created an analytical algorithm and companion software for multi-modal analysis of spatial transcriptomics data with paired histopathological images. The invention builds on cutting edge probabilistic statistical modelling and deep learning technology to improve outcomes for pathology pipelines where applications like identification of various cell types in heterogenous tissue and immune infiltration in cancer tissues of critical diagnostic and prognostic value. The invention provides avenues to improve and expedite such clinical pipelines. Because accurate determination of cell type composition underlies most spatial transcriptomics analysis, the invention also has the potential to improve all applications of spatial transcriptomics data.                                      

Existing analysis and workflow technology is limited but way of reliance on only one data source – either image or transcriptomic data; however, this invention relies on a multimodal approach that leverages strengths from both data modalities in an integrative manner. A probabilistic framework offers the ability to quantify uncertainty in prediction, and a modular design allows the algorithm to continuously improve as more datasets are made available.


Analytical algorithm, software, multi-modal analysis, spatial transcriptomics data, histopathological images, probabilistic statistical modelling, deep learning, improve outcomes, pathology pipeline, heterogenous tissue, immune infiltration, cancer tissues, spatial transcriptomics data, multimodal approach, continuous improvement.

Granted Patents or Published Applications

Unpublished application pending.

Related Scientific References

Asif Zubair, Richard H Chapple, Sivaraman Natarajan, William C Wright, Min Pan, Hyeong-Min Lee, Heather Tillman, John Easton, Paul Geeleher; “Cell type identification in spatial transcriptomics data can be improved by leveraging cell-type-informative paired tissue images using a Bayesian probabilistic model,” Nucleic Acids Res. 2022 Aug 12;50(14):e80. doi: 10.1093/nar/gkac320. 

Licensing Opportunities

We are seeking partners to commercialize this software.

Contact the Office of Technology Licensing (Phone: 901-595-2342, Fax: 901-595-3148) for more information.