S. Korovin
Please Note
4 records found
1
Publisher Correction
CLEAR-IT, a framework for contrastive learning to capture the immune composition of tumor microenvironments (Communications Biology, (2026), 9, 1, (796), 10.1038/s42003-026-09984-2)
Correction to: Communications Biologyhttps://doi.org/10.1038/s42003-026-09984-2, published online 16 April 2026 In the version of this article initially published, there were errors in the Fig. 1g x-axis labels, where the labels currently reading “CD56” and “CD68” from left-to-right were originally interchanged. The figure is now amended in the HTML and PDF versions of the article.
Forensic microtrace investigation relies on time- and labour-intensive microscopic analyses. To aid forensic experts in their investigations, an image recognition model for microtrace localisation and classification is needed. In this work, we use deep learning to automate trace recognition in images captured with automated microscopy. We localise and classify fibres, hairs, skin, glass and sand in microscopy scans through pixel-wise classification of tape-lift samples. As deep learning requires extensive amounts of annotated training data, we additionally investigate various pretraining strategies to minimise the required annotation workload. We compare ImageNet pretraining, pretraining with self-supervised learning and a sequential application of these approaches. We find that pretrained models are able to reduce the required annotated data twofold compared to models trained from scratch while retaining the prediction accuracy. While our ImageNet-pretrained models outperform our self-supervised-pretrained models, we achieve the highest accuracy by combining the two approaches, resulting in a factor 4 reduction of manual annotated microtraces or a 65 % improvement in recognition and localisation accuracy (mean intersection over union increases from 0.34 to 0.56 due to pretraining) when training on only 2.2 dm2 of annotated tape lift scans. The developed models offer a solid fundament for automated analysis of forensic microtrace scans.
Accurate phenotyping of cells in the tumor microenvironment is essential for understanding cancer biology but typically requires precise cell segmentation, limiting scalability. Here, we introduce Contrastive Learning Enabled Accurate Registration of Immune and Tumor cells (CLEAR-IT), a self-supervised framework that learns cell-level features from multiplexed images using only cell locations. CLEAR-IT encoders achieve strong linear evaluation performance, improve substantially with hyperparameter optimization, and maintain high accuracy across imaging modalities and with up to 90% fewer labels. When substituted for handcrafted features in a state-of-the-art classifier, CLEAR-IT features yield higher performance, and their combination enables comparable accuracy with less than half of the labeled data otherwise required. The learned representations also support prognostic modeling: using annotations from a single patient, CLEAR-IT-based phenotyping identifies survival-associated tissue features that generalize across two cohorts and modalities. CLEAR-IT provides a segmentation-light, label-efficient approach for scalable cell phenotyping and enhances existing workflows in digital pathology and tumor microenvironment analysis.
The spatial resolution of optical imaging systems is fundamentally restricted by the diffraction limit. However, in widefield live-cell microscopy, the achievable resolution is further constrained by specimen motion, which indicates the existence of a fundamental spatio-temporal resolution trade-off between signal accumulation during full frame integration and the resulting motion blur. To improve the fidelity with which moving objects can be imaged, a quantitative understanding of this spatio-temporal trade-off is necessary. Here, we present a systematic analysis of motion-induced resolution dynamics measured with the spectral signal-to-noise ratio (SSNR). We developed a simulation framework that models the image formation of objects undergoing arbitrary motion to evaluate the degradation of the spatial resolution under translational, diffusive, and rotational dynamics. Our results demonstrate that for translating objects, the spatial resolution is anisotropically reduced as a function of the orientation of the object relative to the motion vector, leading to the SSNR degrading by up to 50% and the resolution by up to 40% for a 90° change in the motion direction. This motion-blur-induced spectral degradation was further observed on simulated points under Brownian motion and dynamic Drosophila data. Moreover, we show that for rotational motion, conventional radially averaged metrics such as the Fourier ring correlation cannot quantify the effects of angular blur, which the SSNR quantifies accurately. These findings underscore the necessity of an object-oriented imaging approach, in which acquisition parameters such as exposure time are tuned to specific biological spatio-temporal characteristics to optimize the trade-off between motion blur and spatial fidelity.