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

Abstract

Cyclic Immunofluorescence (CyCIF) has emerged as a powerful technique that can measure multiple biomarkers in a single tissue sample but it is limited in panel size due to technical issues and tissue loss. We develop a computational model that imputes a surrogate in silico high-plex CyCIF from only a few experimentally measured biomarkers by learning co-expression and morphological patterns at the single-cell level. The reduced panel is optimally designed to enable full reconstruction of an expanded marker panel that retains the information from the original panel necessary for downstream analysis. Using a masked image modeling approach based on the self-supervised training objective of reconstructing full images at the single-cell level, we demonstrate significant performance improvement over previous attempts on the breast cancer tissue microarray dataset. Our approach offers users access to a more extensive set of biomarkers beyond what has been experimentally measured. It also allows for allocating resources toward exploring novel biomarkers and facilitates greater cell type differentiation and disease characterization. Additionally, it can handle assay failures such as low-quality markers, technical noise, and/or tissue loss in later rounds as well as artificially upsample to include additional panel markers.

Explore & actuate: the future of personalized medicine in oncology through emerging technologies

“…Orion augments existing histopathological analysis of the tissue biopsies by adding complementary spatial information through high-plex immunofluorescence imaging resulting in better interpretable information for both human experts and machine learning (ML) [64 ▪▪ ]. Bioinformatic tools can compensate for limitations of existing platforms; MLSpatial predicts spatial contexts with just scRNA-seq data while MIM-CyCIF predicts other marker expressions beyond the limits of CyCIF [65 ▪▪ ,66 ▪▪ ].…”

Imputing Single-Cell Protein Abundance in Multiplex Tissue Imaging

“…Computationally increasing—or imputing—additional data by filling in missing data with predicted values is already common in other molecular assays, such as single cell RNA sequencing (scRNA) (Chen et al, 2022; Gong et al, 2018; He et al, 2020; Kharchenko et al, 2014; Talwar et al, 2018; Tran et al, 2022; van Dijk et al, 2018; Xu et al, 2021), bulk genomics (Qiu et al, 2020), and bulk transcriptomics (Patruno et al, 2021). While imputation has been applied to MTI images (Sims & Chang, 2023; Ternes et al, 2021), to the best of our knowledge imputation on MTI single-cell datasets has not been explored. Imputing single-cell data is especially valuable because single-cell datasets require fewer computational resources to process than images and can be readily integrated with other molecular datasets.…”