VORTEX: AI-driven 3D spatial transcriptomics
This Learning Wednesday paper note highlights “AI-driven 3D Spatial Transcriptomics.” The study is relevant to AI-enabled 3D spatial transcriptomics, with a focus on how three-dimensional tissue context can change what researchers see and measure.
Selected notes from the paper
"A comprehensive three-dimensional (3D) map of tissue architecture and gene expression is crucial for illuminating the complexity and heterogeneity of tissues across diverse biomedical applications."
"However, most spatial transcriptomics (ST) approaches remain limited to two-dimensional (2D) sections of tissue."
"Recent three-dimensional (3D) pathology studies, fueled by substantial advances in high-resolution 3D tissue imaging modalities such as micro computed tomography (microCT) or open-top light-sheet microscopy, showed that 3D morphological characterization can lead to better patient prognostication or cancer biomarker discovery."
"While promising, in situ approaches remain limited in terms of capture area and depth, and require long processing times."
"Serial section-based approaches provide discontinuous coverage along the axial dimension (i.e., 2.5D ST characterization) of thick tissues."
"Such approaches are impractical for scaling to whole-volume transcriptomic profiling in terms of cost and effort."
"We present VOlumetrically Resolved Transcriptomics EXpression (VORTEX), an AI framework that leverages 3D tissue morphology and minimal 2D ST to predict volumetric 3D ST."
"VORTEX presents a fundamentally different mechanism for 3D ST prediction from other frameworks."
"Existing works meticulously align multiple 2D tissue sections with 2D ST measurements from the same volume to construct a 2.5D ST heatmap."
"Obtaining a 3D or 2.5D ST profile of a sample still results in high costs and turnaround time from having to sequence a large number of tissue sections."
" VORTEX operates on continuous 3D tissue morphology as input, based on the underlying morphomolecular links learned by the models."
"VORTEX can provide 3D ST for each volume with orders of magnitude less cost and time because it requires significantly fewer ST measurements from a VOI for fine-tuning."
"Fine-tuning the model on a single 2D ST capture area from the VOI can help predict the ST profile for any other tissue regions outside the capture area, across the plane and at varying depths."
"Incorporating depth context enhances the predictive performance, suggesting that the 3D context provides more morphological cues for predicting accurate transcriptomics expression."
"VORTEX enables the integration and reuse of morphomolecular traits both across and within the samples, leveraging 3D tissue morphology as the common binding factor."
Demo: https://lnkd.in/ejRtqNXv
Blog post from Cristina Almagro Pérez: https://lnkd.in/eYQzpyPk
Previous Cell paper from Dr. Song: https://lnkd.in/eBvrVf-3
From an Alpenglow perspective, this paper is useful because it connects AI-enabled 3D spatial transcriptomics with a broader need in 3D spatial biology, measuring tissue architecture across depth while preserving context for quantitative analysis.
