Carolina Wählby

This paper develops a vision transformer model for rapid bacterial species identification from time-lapse images without fluorescent labels, achieving high accuracy on clinical isolates.

SFUMATO provides a Bayesian probabilistic framework for spatial transcriptomics that captures uncertainty and enables continuous visualization of transcriptional domains.

The paper predicts food allergies and postpartum depression using microbiome time series data with imputation and logarithmic transformation, achieving balanced accuracies of 0.72 and 0.62 respectively.

This study uses microfluidics, time-lapse microscopy, and deep learning to rapidly identify seven bacterial species with high accuracy in one hour.

This paper develops a method to rapidly detect bacteria in blood using smart centrifugation, microfluidic trapping, and deep learning for sepsis diagnostics.

This paper introduces misPLA, a method for spatially mapping protein interactions and activity states in cancer models, revealing shared and disease-specific signaling patterns.

PIP{Sigma}X provides an automated, user-friendly pipeline for analyzing multiplexed tissue images, enabling robust spatial analysis and integration with visualization tools.

This study evaluates the Xenium In Situ spatial transcriptomics platform using mouse brain and human breast cancer datasets, comparing it to other technologies and benchmarking analysis tools.

Points2Regions clusters spatial transcriptomics data to identify biologically relevant regions across multiple scales efficiently and interactively.

The paper uses deep learning to classify bacteria species from phase-contrast microscopy images without staining or fixation, achieving over 98% accuracy.

TissUUmaps 3 improves interactive visualization and exploration of large-scale spatial omics data through optimizations and new modules for analysis and quality assessment.

The paper introduces three plugins for visualization and quality control in large-scale multiplex tissue analysis to improve accuracy and user trust in automated approaches.

This study compares the predictive performance of models trained on fluorescence and brightfield images for drug mechanism of action prediction, finding comparable results.

SimSearch enables fast ROI detection in microscopy images through user-guided deep learning with minimal interaction and prior information.

The paper creates a comprehensive topographic atlas of early human lung development, identifying cell states, trajectories, and interactions to understand cell heterogeneity origins.

The paper uses a de novo spatial modeling approach to identify novel cell populations in the developing human heart, revealing previously unreported molecular diversity.

This paper presents a machine learning approach for cell classification in glioma tissue and identifies cell niches using graph neural networks.

The paper introduces the HASTE Toolkit, a cloud-native toolkit for intelligent data pipelines that prioritize resource usage based on data content, demonstrating it in microscopy imaging case studies.

ISTDECO improves gene expression decoding in IST data by combining spot localization and barcode decoding into a single algorithm, achieving state-of-the-art performance in noisy and crowded conditions.

This paper introduces a graph-based method for analyzing in situ sequencing data to identify spatial compartments in mouse brain tissue, linking them to known brain regions and morphology.

Spage2vec is an unsupervised method for analyzing spatial transcriptomic data, revealing gene expression patterns in mouse brain tissue without prior segmentation or biological knowledge.

The paper predicts cell migration modes in microscopy images using weakly annotated data and Bayesian deep learning, achieving 87.91% sensitivity.

The paper develops a zebrafish pipeline to study dyslipidemia and atherosclerosis, identifying potential causal genes and showing drug effects.