Scientists have developed a new microscope capable of rapidly imaging thick tissue samples with cellular resolution — enabling surgeons to confirm the margins of tumors mere minutes after their surgical removal, according to a recent study published in the journal Proceedings of the National Academy of Sciences.
New AI microscope confirms tumor removal in minutes
"The main goal of the surgery is to remove all the cancer cells, but the only way to know [whether or not] you got everything is to look at the tumor under a microscope," said a doctoral student in computer engineering at Rice University Mary Jin, who was also co-lead author of the new study.
"Today, you can only do that by first slicing the tissue into extremely thin sections and then imaging those sections separately," said Jin. "This slicing process requires expensive equipment and the subsequent imaging of multiple slices is time-consuming. Our project seeks to basically image large sections of tissue directly, without any slicing."
Most microscopes trade resolution for depth-of-field
Called the deep-learning extended depth-of-field microscope, or DeepDOF, the AI device trains computer algorithms to optimize both image post-processing and the initial image collection process.
With most microscopes, use is a zero-sum game between spatial resolution and depth-of-field, which means only objects the same distance from the lens may be viewed with a clear focus, reports Futurity.
Today, slides examine tumor margins, and preparing them is not a simple proposition. Removed tissue is typically sent to a hospital lab, where health experts either freeze it or apply a chemical treatment before making razor-thin slices to mount and study in slides.
DeepDOF could improve cancer patients' surgery outcomes
However, this process takes inordinate amounts of time, and calls for specialized equipment, not to mention specialists (which also cost money). It's not common for hospitals to examine slides mid-surgery, and many hospitals throughout the world simply lack the requisite tools to try if they could.
"Current methods to prepare tissue for margin status evaluation during surgery have not changed significantly since first introduced over 100 years ago," said co-author Ann Gillenwater, a head and neck surgery professor at University of Texas' MD Anderson. "By bringing the ability to accurately assess margin status to more treatment sites, the DeepDOF has potential to improve outcomes for cancer patients treated with surgery."
DeepDOF's design centers around post-processing algorithm
Co-corresponding author Ashok Veeraraghavan — who is also Jin's doctoral thesis advisor — said DeepDOF employs a standard optical microscope in tandem with an inexpensive optical phase mask, which lowers the price of imaging whole tissue pieces to less than $10. This opens the door to field depths of up to five times that of present-day microscopes.
"Traditionally, imaging equipment like cameras and microscopes are designed separately from imaging processing software and algorithms," said Yubo Tang, co-lead author and postdoctoral research associate in the lab of co-corresponding author Rebecca Richards-Kortum. "DeepDOF is one of the first microscopes that's designed with post-processing algorithm in mind."
AI transforming future of medical industry
The phase mask is positioned above the microscope's objective to modify the light passing into the microscope.
"The modulation allows for better control of depth-dependent blur in the images captured by the microscope," Said Veeraghavan — an associate professor of computer and electrical engineering, to Futurity. "That control helps ensure that the deblurring algorithms that are applied to the captured images are faithfully recovering high-frequency texture information over a much wider range of depths than conventional microscopes."
Artificial intelligence is taking nearly every industry and scientific field into the 21st century. From the cutting edge of quantum chemistry to new and groundbreaking ways of diagnosing dangerous cancer tumors — machine and deep learning have come far. But the middle ground between diagnosis and treatment is beginning to open up to AI applications and, thanks to this new study from Rice University, can help surgeons determine if their work was a success, before they close the patient's body.