IST-editing: Infinite spatial transcriptomic editing in a generated gigapixel mouse pup

Jiqing Wu, Ingrid Berg, Viktor Koelzer
Proceedings of The 7nd International Conference on Medical Imaging with Deep Learning, PMLR 250:1707-1724, 2024.

Abstract

Advanced spatial transcriptomics (ST) techniques provide comprehensive insights into complex organisms across multiple scales, while simultaneously posing challenges in biomedical image analysis. The spatial co-profiling of biological tissues by gigapixel whole slide images (WSI) and gene expression arrays motivates the development of innovative and efficient algorithmic approaches. Using Generative Adversarial Nets (GAN), we introduce **I**nfinite **S**patial **T**ranscriptomic **e**diting (IST-editing) and establish gene expression-guided editing in a generated gigapixel mouse pup. Trained with patch-wise high-plex gene expression (input) and matched image data (output), IST-editing enables the seamless synthesis of arbitrarily large bioimages at inference, *e.g.*, with a $106496 \times 53248$ resolution. After feeding edited gene expression values to the trained model, we simulate cell-, tissue- and animal-level morphological transitions in the generated mouse pup. Lastly, we discuss and evaluate editing effects on interpretable morphological features. The code and generated WSIs are publicly accessible via https://github.com/CTPLab/IST-editing.

Cite this Paper

BibTeX
@InProceedings{pmlr-v250-wu24a, title = {IST-editing: Infinite spatial transcriptomic editing in a generated gigapixel mouse pup}, author = {Wu, Jiqing and Berg, Ingrid and Koelzer, Viktor}, booktitle = {Proceedings of The 7nd International Conference on Medical Imaging with Deep Learning}, pages = {1707--1724}, year = {2024}, editor = {Burgos, Ninon and Petitjean, Caroline and Vakalopoulou, Maria and Christodoulidis, Stergios and Coupe, Pierrick and Delingette, Hervé and Lartizien, Carole and Mateus, Diana}, volume = {250}, series = {Proceedings of Machine Learning Research}, month = {03--05 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v250/main/assets/wu24a/wu24a.pdf}, url = {https://proceedings.mlr.press/v250/wu24a.html}, abstract = {Advanced spatial transcriptomics (ST) techniques provide comprehensive insights into complex organisms across multiple scales, while simultaneously posing challenges in biomedical image analysis. The spatial co-profiling of biological tissues by gigapixel whole slide images (WSI) and gene expression arrays motivates the development of innovative and efficient algorithmic approaches. Using Generative Adversarial Nets (GAN), we introduce **I**nfinite **S**patial **T**ranscriptomic **e**diting (IST-editing) and establish gene expression-guided editing in a generated gigapixel mouse pup. Trained with patch-wise high-plex gene expression (input) and matched image data (output), IST-editing enables the seamless synthesis of arbitrarily large bioimages at inference, *e.g.*, with a $106496 \times 53248$ resolution. After feeding edited gene expression values to the trained model, we simulate cell-, tissue- and animal-level morphological transitions in the generated mouse pup. Lastly, we discuss and evaluate editing effects on interpretable morphological features. The code and generated WSIs are publicly accessible via https://github.com/CTPLab/IST-editing.} }
Endnote
%0 Conference Paper %T IST-editing: Infinite spatial transcriptomic editing in a generated gigapixel mouse pup %A Jiqing Wu %A Ingrid Berg %A Viktor Koelzer %B Proceedings of The 7nd International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2024 %E Ninon Burgos %E Caroline Petitjean %E Maria Vakalopoulou %E Stergios Christodoulidis %E Pierrick Coupe %E Hervé Delingette %E Carole Lartizien %E Diana Mateus %F pmlr-v250-wu24a %I PMLR %P 1707--1724 %U https://proceedings.mlr.press/v250/wu24a.html %V 250 %X Advanced spatial transcriptomics (ST) techniques provide comprehensive insights into complex organisms across multiple scales, while simultaneously posing challenges in biomedical image analysis. The spatial co-profiling of biological tissues by gigapixel whole slide images (WSI) and gene expression arrays motivates the development of innovative and efficient algorithmic approaches. Using Generative Adversarial Nets (GAN), we introduce **I**nfinite **S**patial **T**ranscriptomic **e**diting (IST-editing) and establish gene expression-guided editing in a generated gigapixel mouse pup. Trained with patch-wise high-plex gene expression (input) and matched image data (output), IST-editing enables the seamless synthesis of arbitrarily large bioimages at inference, *e.g.*, with a $106496 \times 53248$ resolution. After feeding edited gene expression values to the trained model, we simulate cell-, tissue- and animal-level morphological transitions in the generated mouse pup. Lastly, we discuss and evaluate editing effects on interpretable morphological features. The code and generated WSIs are publicly accessible via https://github.com/CTPLab/IST-editing.
APA
Wu, J., Berg, I. & Koelzer, V.. (2024). IST-editing: Infinite spatial transcriptomic editing in a generated gigapixel mouse pup. Proceedings of The 7nd International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 250:1707-1724 Available from https://proceedings.mlr.press/v250/wu24a.html.

Related Material