Nature Methods: A new way to construct detailed maps of human tissue

Researchers at Weill Cornell Medical College have developed a computational method to map the structure of
human tissue in unprecedented detail.
Their approach promises to accelerate the study of organ-level cellular interactions and potentially provide powerful new diagnostic strategies
for a wide range of diseases.

The method stemmed from scientists' frustration
with the gap between classical microscopy and modern single-cell molecular analysis.
"If you look at tissue under a microscope, you see a bunch of cells clumping together spatially — you can see that tissue in the image almost immediately," said Junbum Kim, lead author of the study, a graduate student
in physiology and biophysics at Weill Cornell Medical College.
"Now, cell biologists have the ability to examine individual cells in great detail, down to the genes expressed by each cell, so they focus on the cell, not the tissue structure
," he said.

However, "it is crucial for researchers to learn more about the details of the organizational structure; Fundamental changes in the relationships between cells within tissues drive the function of healthy and diseased organs," said senior author Dr Olivier Elemento, director
of the England Institute of Precision Medicine.

However, manually combining single-cell data with tissue charts is slow and tedious
.
Machine learning algorithms have shown some potential to automate processes, but they are limited
by the data used to train them.
To solve this problem, Kim and his colleagues developed an unsupervised computational strategy that uses a combination of single-cell gene expression profiles and cell locations to define structural regions
within tissues.

Co-senior author André Renderiro, Ph.
D.
, a postdoctoral researcher at Weill Cornell Medicine and currently principal investigator at the Center for Molecular Medicine of the Austrian Academy of Sciences in Vienna, Austria, likens this new approach to mapping cities like New York: "One way is to go to every intersection and count every kind of building: it's residential, it's commercial.
.
.
Is it a shop or a restaurant?" Put all this data into one matrix, put the location of the building into another matrix, and then you can combine the two matrices to look for patterns
.

"Essentially, we can make a rough idea of the location and boundaries of different neighborhoods based on the abundance of residential and commercial buildings — just as anyone walking on the Upper East Side, Midtown, or Downtown does based on their observations.
"

Using the new method, the researchers generated detailed maps of several types of tissues, identifying and quantifying new aspects of microscopic anatomy — patterns that emerge at small scales as cells interact and determine the final function of
the tissue.
They also teamed up with a colleague at the University of North Carolina at Chapel Hill who studies lung disease to demonstrate that their technique can map subtle differences
between different disease states in tissues.

While cancer and other chronic diseases often cause significant changes in tissue structure, detailed microanatomy can also help diagnose and treat more serious diseases
.
Lendello noted that severe COVID-19 is an example, "where a lot of immune cells move around and lung tissue changes very
dramatically.
" The team is now applying their new technology to a wide range of organizations to understand how organizational changes can lead to functional in healthy states and dysfunction in times of
disease.