FLINO: a new method for immunofluorescence bioimage normalization

GammaGateR: semi-automated marker gating for single-cell multiplexed imaging

MOTIVATION

Multiplexed immunofluorescence (mIF) is an emerging assay for multichannel protein imaging that can decipher cell-level spatial features in tissues. However, existing automated cell phenotyping methods, such as clustering, face challenges in achieving consistency across experiments and often require subjective evaluation. As a result, mIF analyses often revert to marker gating based on manual thresholding of raw imaging data.

RESULTS

To address the need for an evaluable semi-automated algorithm, we developed GammaGateR, an R package for interactive marker gating designed specifically for segmented cell-level data from mIF images. Based on a novel closed-form gamma mixture model, GammaGateR provides estimates of marker-positive cell proportions and soft clustering of marker-positive cells. The model incorporates user-specified constraints that provide a consistent but slide-specific model fit. We compared GammaGateR against the newest unsupervised approach for annotating mIF data, employing two colon datasets and one ovarian cancer dataset for the evaluation. We showed that GammaGateR produces highly similar results to a silver standard established through manual annotation. Furthermore, we demonstrated its effectiveness in identifying biological signals, achieved by mapping known spatial interactions between CD68 and MUC5AC cells in the colon and by accurately predicting survival in ovarian cancer patients using the phenotype probabilities as input for machine learning methods. GammaGateR is a highly efficient tool that can improve the replicability of marker gating results, while reducing the time of manual segmentation.

AVAILABILITY AND IMPLEMENTATION

The R package is available at https://github.com/JiangmeiRubyXiong/GammaGateR.

Quantitative performance assessment of Ultivue multiplex panels in formalin-fixed, paraffin-embedded human and murine tumor specimens

We present a rigorous validation strategy to evaluate the performance of Ultivue multiplex immunofluorescence panels. We have quantified the accuracy and precision of four different multiplex panels (three human and one mouse) in tumor specimens with varying levels of T cell density. Our results show that Ultivue panels are typically accurate wherein the relative difference in cell proportion between a multiplex image and a 1-plex image is less than 20% for a given biomarker. Ultivue panels exhibited relatively high intra-run precision (CV ≤ 25%) and relatively low inter-run precision (CV >> 25%) which can be remedied by using local intensity thresholding to gate biomarker positivity. We also evaluated the reproducibility of cell-cell distance estimates measured from multiplex images which show high intra- and inter-run precision. We introduce a new metric, multiplex labeling efficiency, which can be used to benchmark the overall fidelity of the multiplex data across multiple batch runs. Taken together our results provide a comprehensive characterization of Ultivue panels and offer practical guidelines for analyzing multiplex images.

MIM-CyCIF: masked imaging modeling for enhancing cyclic immunofluorescence (CyCIF) with panel reduction and imputation

Cyclic Immunofluorescence (CyCIF) can quantify multiple biomarkers, but panel capacity is limited by technical challenges. We propose a computational panel reduction approach that can impute the information content from 25 markers using only 9 markers, learning co-expression and morphological patterns while concurrently increasing speed and panel content and decreasing cost. We demonstrate strong correlations in predictions and generalizability across breast and colorectal cancer, illustrating applicability of our approach to diverse tissue types.

Semi-automated approaches for interrogating spatial heterogeneity of tissue samples

Tissues are spatially orchestrated ecosystems composed of heterogeneous cell populations and non-cellular elements. Tissue components' interactions shape the biological processes that govern homeostasis and disease, thus comprehensive insights into tissues' composition are crucial for understanding their biology. Recently, advancements in the spatial biology field enabled the in-depth analyses of tissue architecture at single-cell resolution, while preserving the structural context. The increasing number of biomarkers analyzed, together with whole tissue imaging, generate datasets approaching several hundreds of gigabytes in size, which are rich sources of valuable knowledge but require investments in infrastructure and resources for extracting quantitative information. The analysis of multiplex whole-tissue images requires extensive training and experience in data analysis. Here, we showcase how a set of open-source tools can allow semi-automated image data extraction to study the spatial composition of tissues with a focus on tumor microenvironment (TME). With the use of Lunaphore COMET platform, we interrogated lung cancer specimens where we examined the expression of 20 biomarkers. Subsequently, the tissue composition was interrogated using an in-house optimized nuclei detection algorithm followed by a newly developed image artifact exclusion approach. Thereafter, the data was processed using several publicly available tools, highlighting the compatibility of COMET-derived data with currently available image analysis frameworks. In summary, we showcased an innovative semi-automated workflow that highlights the ease of adoption of multiplex imaging to explore TME composition at single-cell resolution using a simple slide in, data out approach. Our workflow is easily transferrable to various cohorts of specimens to provide a toolset for spatial cellular dissection of the tissue composition.

Dual-modality imaging of immunofluorescence and imaging mass cytometry for whole-slide imaging and accurate segmentation

Imaging mass cytometry (IMC) is a powerful technique capable of detecting over 30 markers on a single slide. It has been increasingly used for single-cell-based spatial phenotyping in a wide range of samples. However, it only acquires a rectangle field of view (FOV) with a relatively small size and low image resolution, which hinders downstream analysis. Here, we reported a highly practical dual-modality imaging method that combines high-resolution immunofluorescence (IF) and high-dimensional IMC on the same tissue slide. Our computational pipeline uses the whole-slide image (WSI) of IF as a spatial reference and integrates small-FOV IMC into a WSI of IMC. The high-resolution IF images enable accurate single-cell segmentation to extract robust high-dimensional IMC features for downstream analysis. We applied this method in esophageal adenocarcinoma of different stages, identified the single-cell pathology landscape via reconstruction of WSI IMC images, and demonstrated the advantage of the dual-modality imaging strategy.

GammaGateR: semi-automated marker gating for single-cell multiplexed imaging

Motivation

Multiplexed immunofluorescence (mIF) is an emerging assay for multichannel protein imaging that can decipher cell-level spatial features in tissues. However, existing automated cell phenotyping methods, such as clustering, face challenges in achieving consistency across experiments and often require subjective evaluation. As a result, mIF analyses often revert to marker gating based on manual thresholding of raw imaging data.

Results

To address the need for an evaluable semi-automated algorithm, we developed GammaGateR, an R package for interactive marker gating designed specifically for segmented cell-level data from mIF images. Based on a novel closed-form gamma mixture model, GammaGateR provides estimates of marker-positive cell proportions and soft clustering of marker-positive cells. The model incorporates user-specified constraints that provide a consistent but slide-specific model fit. We compared GammaGateR against the newest unsupervised approach for annotating mIF data, employing two colon datasets and one ovarian cancer dataset for the evaluation. We showed that GammaGateR produces highly similar results to a silver standard established through manual annotation. Furthermore, we demonstrated its effectiveness in identifying biological signals, achieved by mapping known spatial interactions between CD68 and MUC5AC cells in the colon and by accurately predicting survival in ovarian cancer patients using the phenotype probabilities as input for machine learning methods. GammaGateR is a highly efficient tool that can improve the replicability of marker gating results, while reducing the time of manual segmentation.

Availability and Implementation

The R package is available at https://github.com/JiangmeiRubyXiong/GammaGateR.

MIM-CyCIF: Masked Imaging Modeling for Enhancing Cyclic Immunofluorescence (CyCIF) with Panel Reduction and Imputation

CyCIF can quantify multiple biomarkers, but panel capacity is limited by technical challenges. We propose a computational panel reduction approach that can impute the information content from 25 markers using only 9 markers, learning co-expression and morphological patterns while concurrently increasing speed and panel content and decreasing cost. We demonstrate strong correlations in predictions and generalizability across breast and colorectal cancer, illustrating applicability of our approach to diverse tissue types.

MIM-CyCIF: Masked Imaging Modeling for Enhancing Cyclic Immunofluorescence (CyCIF) with Panel Reduction and Imputation

CyCIF quantifies multiple biomarkers, but panel capacity is compromised by technical challenges including tissue loss. We propose a computational panel reduction, inferring surrogate CyCIF data from a subset of biomarkers. Our model reconstructs the information content from 25 markers using only 9 markers, learning co-expression and morphological patterns. We demonstrate strong correlations in predictions and generalizability across breast and colorectal cancer tissue microarrays, illustrating broader applicability to diverse tissue types.

Statistical Analysis of Multiplex Immunofluorescence and Immunohistochemistry Imaging Data

Advances in multiplexed single-cell immunofluorescence (mIF) and multiplex immunohistochemistry (mIHC) imaging technologies have enabled the analysis of cell-to-cell spatial relationships that promise to revolutionize our understanding of tissue-based diseases and autoimmune disorders. Multiplex images are collected as multichannel TIFF files; then denoised, segmented to identify cells and nuclei, normalized across slides with protein markers to correct for batch effects, and phenotyped; and then tissue composition and spatial context at the cellular level are analyzed. This chapter discusses methods and software infrastructure for image processing and statistical analysis of mIF/mIHC data.

Next-Generation Pathology Using Multiplexed Immunohistochemistry: Mapping Tissue Architecture at Single-Cell Level

Single-cell omics aim at charting the different types and properties of all cells in the human body in health and disease. Over the past years, myriads of cellular phenotypes have been defined by methods that mostly required cells to be dissociated and removed from their original microenvironment, thus destroying valuable information about their location and interactions. Growing insights, however, are showing that such information is crucial to understand complex disease states. For decades, pathologists have interpreted cells in the context of their tissue using low-plex antibody- and morphology-based methods. Novel technologies for multiplexed immunohistochemistry are now rendering it possible to perform extended single-cell expression profiling using dozens of protein markers in the spatial context of a single tissue section. The combination of these novel technologies with extended data analysis tools allows us now to study cell-cell interactions, define cellular sociology, and describe detailed aberrations in tissue architecture, as such gaining much deeper insights in disease states. In this review, we provide a comprehensive overview of the available technologies for multiplexed immunohistochemistry, their advantages and challenges. We also provide the principles on how to interpret high-dimensional data in a spatial context. Similar to the fact that no one can just “read” a genome, pathological assessments are in dire need of extended digital data repositories to bring diagnostics and tissue interpretation to the next level.

mxnorm: An R Package to Normalize Multiplexed Imaging Data

Multiplexed imaging is an emerging single-cell assay that can be used to understand and analyze complex processes in tissue-based cancers, autoimmune disorders, and more. These imaging technologies, which include co-detection by indexing (CODEX), multiplexed ion beam imaging (MIBI), and multiplexed immunofluorescence imaging (MxIF), provide detailed information about spatial interactions between cells (Angelo et al., 2014; Gerdes et al., 2013; Goltsev et al., 2018). Multiplexed imaging experiments generate data across hundreds of slides and images, often resulting in terabytes of complex data to analyze through imaging analysis pipelines. Methods are rapidly developing to improve particular parts of the pipeline, including software packages in R and Python like spatialTime, imcRtools, MCMICR0, and Squidpy (Creed et al., 2021; Palla et al., 2021; Schapiro et al., 2021; Windhager et al., 2021). An important, but understudied component of this pipeline is the analysis of technical variation within this complex data source - intensity normalization is one way to remove this technical variability. The combination of disparate pre-processing pipelines, imaging variables, optical effects, and within-slide dependencies create batch and slide effects that can be reduced via normalization methods. Current state-of-the-art methods vary heavily across research labs and image acquisition platforms, without one singular method that is uniformly robust - optimal statistical methods seek to improve similarity across images and slides by removing this technical variability while maintaining the underlying biological signal in the data. mxnorm is open-source software built with R and S3 methods that implements, evaluates, and visualizes normalization techniques for multiplexed imaging data. Extending methodology described in Harris et al. (2022), we intend to set a foundation for the evaluation of multiplexed imaging normalization methods in R. This easily allows users to extend normalization methods into the field, and provides a robust evaluation framework to measure both technical variability and the efficacy of various normalization methods. One key component of the R package is the ability to supply user-defined normalization methods and thresholding algorithms to assess normalization in multiplexed imaging data. Core features, usage details, and extensive tutorials are available in the package documentation and vignette on CRAN and the software repository.