Machine learning meets classical computer vision for accurate cell identification
Karimi, E.; Rezanejad, M.; Fiset, B.; Perus, L.; McDowell, S. A. C.; Arabzadeh, A.; Beugnot, G.; Siegel, P.; Guiot, M.-C.; Quail, D. F.; Siddiqi, K.; Walsh, L. A.
Show abstract
High-parameter multiplex immunostaining techniques have revolutionized our ability to image healthy and diseased tissues with unprecedented depth; however, accurate cell identification and segmentation remain significant downstream challenges. Identifying individual cells with high precision is a requisite to reliably and reproducibly interpret acquired data. Here we introduce CIRCLE, a cell identification pipeline that combines classical and modern machine learning-based computer vision algorithms to address the shortcomings of current cell segmentation tools for 2D images. CIRCLE is a fully automated hybrid cell detection model, eliminating subjective investigator bias and enabling high-throughput image analysis. CIRCLE accurately distinguishes cells across diverse tissues microenvironments, resolves low-resolution structures, and can be applied to any 2D image that contains nuclei. Importantly, we quantitatively demonstrate that CIRCLE outperforms current state-of-the-art image segmentation tools using multiple accuracy measures. As high-throughput multiplex imaging grows closer toward standard practice for histology, integration of CIRCLE into analysis protocols will deliver unparalleled segmentation quality.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
- Single-Cell Multi-Modal GAN (scMMGAN) reveals spatial patterns in single-cell data from triple negative breast cancer 94%
- Generating realistic neurophysiological time series with denoising diffusion probabilistic models 93%
- Bi-level Graph Learning Unveils Prognosis-Relevant Tumor Microenvironment Patterns in Breast Multiplexed Digital Pathology 92%
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.