Spatial biology tools for the advancement of cancer | Tribun Health
Written by Tribun Health on
Share this articlesubscribe to our blog
[SEPTEMBER 19TH, 2023]
Trillions of cells in our bodies collaborate seamlessly to maintain our well-being. These cells serve as the essential foundation of life and perform precise, regulated functions vital to our survival. Yet, on rare occasions, a cell may break away from its usual behavior, dividing uncontrollably and possibly spreading to other areas of the body, resulting in cancer. (1)
The region surrounding the tumor forms a complex ecosystem known as the tumor microenvironment (TME). In it, there's a mix of cellular and non-cellular elements, like the extra-cellular matrix (ECM), diverse tissue cell types, immune cells, endothelial cells, fibroblasts, blood vessels, and signaling molecules. All these elements interact with each other and the tumor, ultimately supporting its growth.
CRUCIAL INSIGHTS: TUMOR MICROENVIRONMENT & NOVEL TARGETED THERAPIES
Understanding the intricate spatial environment of cells and tissues and their interplay has become increasingly important in our quest to better understand and treat cancer whilst improving patient outcomes. Consequently, a plethora of spatial biology tools and techniques have been and are continuously being developed, with varying levels of complexity rendering their use more difficult.
Over the past decade, we have observed a surge in spatial proteomics techniques, which aim to characterize the abundance and spatial distribution of proteins and their post-translational modifications in the TME (2). One of these techniques, multiplex immunofluorescence (mIF), is particularly useful in revealing the heterogeneity of the cell types and immune cell populations of the TME without losing their spatial distribution. The diversity in cell populations and their states are translated by the expression of a large panel of biomarkers which can be visualized in the mIF images. Given the multitude of biomarkers that can be investigated concurrently, it is essential to employ robust image analysis tools that can effectively emphasize vital insights amidst the rich information contained in the mIF images. Such complexities exceed the capabilities of standard image analysis tools and require more than just human visual interpretation to accurately decipher the high dimensional data.
Image Source: High dimensional multiplex immunofluorescence breast cancer image – courtesy of the Centre de Recherche et de Cancérologie Lyon.
MIF IMAGES COMBINED WITH DEEP LEARNING TO UNRAVEL THE COMPLEX TME
The use of artificial intelligence (AI)-based tools has greatly facilitated and automated the objective quantification of complex images. By mimicking human actions, AI facilitates a more effective and rapid extraction of data from images. In the past, AI often relied on techniques that used handcrafted features or thresholds to analyze images. However, these were often difficult to define or calibrate, varying per image and per region.
Deep Learning (DL) offers a solution to this issue. It operates as a comprehensive, end-to-end, trainable system. Such an approach saves time and removes the need for the definition of precise features for algorithm training. It also allows for the discovery of new features, which can better define and explain the TME. Within the realm of mIF image analysis, the objectives encompass intricate tasks such as extracting information about single-cell interactions, distal analysis, biomarker expression, and cell infiltration within tumors.
Tribun Health has developed a standard workflow for multiplexed fluorescence (mIF) image analysis, along with a strong pipeline of models based on DL, to understand the complex information contained within such images. All cellular positions within the tissue are first detected, typically using the DAPI staining. Each cell is then classified as positive per biomarker. Certain cells might exhibit the simultaneous expression of multiple biomarkers, unveiling phenotypes of interest. Tribun Health’s DL models are based on Convolutional Neural Networks (CNN) which achieve state-of-the-art results in most image classification tasks. As a result, these algorithms acquire the capability to simultaneously learn both the unique biomarker intensity and the shape, facilitating the accurate identification of positive cells.
Tribun Health’s workflow for multiplexed image analysis
Spatial biology tools provide an extensive overview of the cellular interactions, the immune contexture within tumors, and their spatial organization. Such tools are fundamentally transforming our understanding of the TME. They not only enable the optimization of existing cancer treatments, but also pave the way for the development of novel therapy strategies. New immunotherapy targets can be identified, and patient response predictions can be improved, making precision medicine a reality for the benefit of the patient.
Ready to explore options for your multiplex immunofluorescence image analysis projects? Request a consultation
(1) Baghban, R., Roshangar, L., Jahanban-Esfahlan, R. et al. Tumor microenvironment complexity and therapeutic implications at a glance. Cell Commun Signal 18, 59 (2020). https://doi.org/10.1186/s12964-020-0530-4
(2) Mao Y, Wang X, Huang P, Tian R. Spatial proteomics for understanding the tissue microenvironment. Analyst. 2021 Jun 14;146(12):3777-3798. doi: 10.1039/d1an00472g. PMID: 34042124.