MARC: Morphology-Aware Regression of Consensus for Cell Segmentation in Subcellular Spatial Transcriptomics
Authors: Xinyu Shu, Andrew Zhang, Jean Yang, Jinman Kim
Organizations: School of Computer Science, The University of Sydney, Australia · Sydney Precision Data Science Centre, The University of Sydney, Australia · School of Mathematics and Statistics, The University of Sydney, Australia · Charles Perkins Centre, The University of Sydney, Australia
Accurate cell segmentation remains a major bottleneck in subcellular spatial transcriptomics (SST), in which morphological images and spatially resolved RNA transcripts are used to partition tissues into individual cellular instances. As segmentation serves as the foundation for constructing cell-level representations, boundary errors can lead to incorrect transcript assignments and compromise downstream analyses. However, reliable ground-truth boundaries are unavailable because they must be inferred from incomplete morphological and transcript signals. Furthermore, manual annotation of a large number of cells is time-consuming. Agreement among complementary segmentation methods provides a practical surrogate for identifying well-supported and ambiguous regions, but explicit consensus construction requires executing multiple computationally intensive pipelines. In this study, we propose MARC (Morphology-Aware Regression of Consensus), a framework that predicts a multi-method consensus-support map for SST segmentation. MARC is trained with leave-one-method-out consensus pseudo-targets and a Foreground-Union Consensus Loss that focuses supervision on candidate and consensus foreground. We evaluated MARC on 4,642 held-out tiles from Xenium kidney tissue, achieving a mean Dice score of 0.90, a mean intersection-over-union of 0.82, and a mean cell-level Spearman correlation of 0.79 against explicitly computed cross-method consensus maps. We demonstrate that the predicted consensus maps localise weakly supported regions while preserving consensus-based rankings and identifying low-consensus cells for manual review. These results show that MARC closely approximates explicit cross-method consensus without multi-method inference and therefore has the potential to facilitate robust, consensus-aware evaluation of cell segmentation in large-scale SST studies.
Figures & tables
Figure 1: Overview of explicit multi-method consensus construction and MARC. (a) Explicit consensus averages outputs from multiple segmentation pipelines using method-specific SST inputs, making consensus construction computationally intensive. (b) MARC predicts pixel-level consensus support from a candidate mask Fk , DAPI morphology D , and transcript information T , and aggregates the prediction into cell-level scores. MARC requires no segmentation ensemble at inference. The same candidate cell is outlined in red.
Figure 2: Architecture of MARC for predicting pixel-level consensus support from a candidate mask, DAPI morphology, and transcript density.
Candidate input
Foreground-union L1 ↓
Dice ↑
IoU ↑
Sensitivity ↑
Precision ↑
Cell Spearman correlation ρ↑
Cellpose-SAM
0.0832
0.9007
0.8193
0.9020
0.8994
0.8172
BIDCell
0.1060
0.9029
0.8230
0.9298
0.8774
0.6251
ProSeg
0.1240
0.8844
0.7927
0.9073
0.8626
0.8802
Xenium 10x
0.0793
0.9127
0.8394
0.9283
0.8977
0.8396
Avg. over methods
0.0981
0.9002
0.8186
0.9168
0.8843
0.7905
Table 1: Performance of MARC predictions against explicit consensus targets. The best result for each metric is shown in bold.
Figure 3: Predicted versus target cell-level consensus-support scores for each candidate method on the held-out test region.
Configuration
Method
Dice ↑
IoU ↑
Sensitivity ↑
Precision ↑
Cell Spearman correlation ρ↑
Candidate-copy baseline
BIDCell
0.6990
0.5372
0.5474
0.9665
– ∗
Cellpose-SAM
0.8878
0.7982
0.8794
0.8963
– ∗
ProSeg
0.8028
0.6705
0.9608
0.6894
– ∗
Xenium 10x
0.8863
0.7958
0.9070
0.8665
– ∗
Avg. over methods
0.8189
0.7004
0.8237
0.8547
– ∗
MARC: candidate mask only
BIDCell
0.8364
0.7189
0.8031
0.8727
0.2493
Table 2: Input-channel ablation evaluated against explicit consensus targets.
Figure 4: Qualitative consensus-support predictions on the held-out test region. Rows show representative candidate cells from each segmentation method, while columns show the DAPI morphology, transcript-density map, candidate foreground mask, predicted pixel-level consensus, and predicted cell-level consensus map. Red outlines delineate the representative candidate cell in each row. Legends in the first column report the segmentation method, cell identifier, area in pixels, predicted cell-level consensus score, and crop coordinates. Brighter values in the output maps indicate stronger predicted cross-method consensus support.
Spatial transcriptomics jointly captures molecular state, tissue morphology, and spatial organization, yet spatial-domain methods are typically compared only at late training checkpoints, obscuring differences in convergence efficiency. We introduce MP-MNCA, a morphology-prior masked neighbor cross-attention framework that preserves the full 3,000 dimensional highly-variable-gene representation while selectively routing molecular information across spatial neighbors, with histological similarity incorporated as a continuous attention prior under a self-supervised masked neighbor-contrastive objective. Across 12 heterogeneous human dorsolateral prefrontal cortex (DLPFC) sections, MP-MNCA is evaluated against established spatial-domain baselines using both convergence trajectories and epoch-wise spatial-domain visualizations. Our results show that MP-MNCA (i) reaches coherent spatial-domain structure in substantially fewer training epochs, as reflected quantitatively by convergence curves and qualitatively by progressive spatial grids, and (ii) maintains strong clustering quality, achieving a mean peak refined ARI of 0.546 versus 0.549 for the strongest baseline (STAIG), while attaining the highest mean refined NMI of 0.657 versus 0.640. Moreover, MP-MNCA outperforms STAIG on 7 of 12 sections in ARI and 11 of 12 in NMI, demonstrating faster convergence without sacrificing spatial-domain quality.
Spatial transcriptomics (ST) links tissue morphology with molecular programs, motivating multimodal pretraining methods that align histology images with gene expression. However, existing approaches suffer from two key limitations: spatially informative gene selection is often dominated by ubiquitous housekeeping genes, leading to weakly discriminative representations, and independent spot-patch alignment fails to capture spatial dependencies that are critical for tissue organization. To address these challenges, we introduce PaSTel, a hierarchical multimodal pretraining framework that integrates biological priors at three levels. At the spot level, TF-IDF reweighting is used to identify spatially informative genes; at the functional level, curated KEGG pathways serve as anchors for encoding global biological semantics; and at the regional level, spatial clustering aggregates neighboring spots to model meso-scale tissue structure. Across multiple downstream tasks, PaSTel consistently outperforms existing vision and vision-omics encoders, demonstrating that incorporating multiscale biological priors yields more informative and transferable representations for spatial transcriptomics.
Azim Dehghani Amirabad, Junchao Zhu, Pushpak Pati +3
Predicting gene expression from H&E-stained histology images offers a scalable alternative to costly spatial transcriptomics, yet most existing methods operate at the spot level, where signals from multiple cells are aggregated and critical cellular heterogeneity is obscured. Extending this paradigm to single-cell resolution is non-trivial. Naively applying pathology foundation models faces a scale mismatch: their patch-level representations mix multiple cells, whereas per-cell cropping or resizing distorts morphology and removes local context. Conversely, segmentation-based models without strong pretrained visual encoders often lack the morphological representation capacity needed for accurate molecular prediction and inherit errors from imperfect cell boundary masks. Here, we present CELLO, an efficient end-to-end framework that performs a single pathology foundation model forward pass per image and uses grid sampling to extract location-specific features for all cells simultaneously. We further introduce a distance-decay cross-attention module that refines each cell representation using spatially biased local morphological context. Using 52 public Xenium-H&E pairs from HEST-1k that span 12 organs and approximately 10 million cells, CELLO improves the average predictive accuracy over the evaluated baselines while reducing the mean whole-slide inference time compared to DeepSpot2Cell, a 14.0x speed-up on average that excludes upstream cell segmentation. Our work establishes a scalable foundation for single-cell gene expression prediction from H&E images.
Zijun Gao, Chunbin Gu, Jinxi Xiang +2
The Chinese University of Hong Kong · Stanford University School of Medicine