Multiplex Imaging Analysis —Multiplex imaging is an emerging technology that stains multiple proteins on a single slide while preserving tissue architecture. This provides invaluable insights into spatial biology, and the complexity of the immune tumor microenvironment (iTME). Current multiplex technologies can map as many as 100 unique protein markers to identify cells and their functional state within a single slide, providing a highly detailed picture of the iTME.
This information can be used to discover new drug targets, understand drug mechanisms of action and develop predictive biomarkers for immunotherapy and other treatment modalities, making Multiplex Imaging Analysis a valuable tool for cancer researchers. While holding great promise, it is currently not fully utilized in cancer research, as the analysis of multiplex images suffers from several limitations, including inaccurate cell-typing, which is mainly achieved through manual thresholding and clustering-based methods. These methods are laborious, user-dependent, and do not support transfer learning between imaging modalities and antibody panels, which makes them hard to implement as a routine tool for translational and clinical research. Existing methods for the deduction of cell state (phenotypic) marker expression share similar deficiencies. The scientific community clearly needs a more robust and automated pipeline for multiplex imaging analysis.
A deep learning-based framework for cell typing in multiplex imaging
To address these limitations, our team at Nucleai, a leader in AI-powered spatial biology, has trained a novel, end-to-end deep learning (DL) framework for multiplex image analysis, which enables accurate cell typing and phenotypic marker quantification, and which is robust across markers, tumor indications and imaging modalities. In a new study, this innovative approach demonstrated superior performance to conventional analytical methods in classifying cell types and phenotypic markers, thus enabling the identification of multiple cellular and spatial features associated with disease prognosis.
The training of a DL multi-classifier was based on a publicly available dataset (CODEX), consisting of 140 tissue cores from 35 colorectal cancer (CRC) patients, which were stained with 56 protein markers and matched H&E slides, to annotate 12 cell types, as well as positivity for 14 immunomodulatory biomarkers. More than 7,000 cell annotations from 57 tissue cores were used in the training set for the DL multi-classifier.
Performance was evaluated quantitatively on 1,800 annotations from 14 test cores and qualitatively on all cores, by expert pathologists. The models were not trained on the annotations used for performance evaluation.

Compared to current clustering or manual-gating based methods of cell typing, our results demonstrated a 1.5-fold increase in accuracy – from 65% to 90% including high accuracy in rare cell populations such as dendritic cells. Our deep learning analysis pipeline also exhibited very high accuracy in unseen markers and in two cohorts on which the algorithm was not trained upon, one of which was stained with a different imaging technology. Thus, our multiplex analysis pipeline demonstrates marked improvement in accuracy compared to current commonplace analysis methods, and is robust across markers, tumor indications and imaging technologies.

One of the key benefits of the AI-powered spatial biology platform is its ability to analyze multiplexed samples with high precision and accuracy. The platform uses machine learning algorithms to identify and quantify multiple targets simultaneously, leading to a more comprehensive analysis of the biological samples. Additionally, the platform is able to process large amounts of data quickly, reducing the time from months to weeks, and efficiently, making it an attractive solution for high-plex multiplex analysis.
In summary, mIF imaging is a crucial tool in the development of cancer drugs, due to its ability to analyze the tumor microenvironment and its potential to identify biomarkers that could predict clinical outcomes in cancer patients, including response to immunotherapies and overall survival. Despite the success of immunotherapies that harness the patient’s immune system to combat cancer, these treatments are only effective for a small group of patients. To effectively differentiate between those who will respond to immunotherapy and those who won’t, accurate predictive biomarkers are crucial. Biopharmaceutical companies seeking to capitalize on this rapidly growing field can benefit from Nucleai’s comprehensive, multiplex deep learning-based solution.