Researchers at Rice University and the University of Texas MD Anderson Cancer Center have announced a new microscope powered by AI. The researchers believe the AI-powered microscope could check cancer margins in minutes. When surgeons have to remove cancer, the biggest question is, did they get it all, and the margins are examined in detail to determine if all the cancer was removed. The new microscope was created to quickly and inexpensively examine large tissue sections, potentially during surgery, to discover that critical answer.
The microscope is designed to image relatively thick pieces of tissue with cellular resolution and can allow surgeons to inspect the margins of tumors within minutes of cutting them out. Rice University engineers and applied physicists created the telescope and talked about it in a study published this week. The microscope is called deep learning extended depth-of-field microscope or DeepDOF and leverages deep learning to train a computer algorithm to optimize image collection and image post-processing.
The researchers say that a typical microscope has a trade-off between spatial resolution and depth-of-field. That means only objects at the same distance from the lens can be clearly focused. Features even a few millionths of a meter closer or further from the microscope’s objective are blurry. That means microscope samples have to be very thin and mounted on glass slides.
The need for the samples to be extremely thin means removed tissue has to be sent to the lab where it’s frozen or prepared with chemicals and cut into razor-thin slices for slides. That is a time-consuming process requiring specialized equipment. It’s very rare for hospitals to have the ability to make slides for examination during surgery.
DeepDOF uses a standard optical microscope combined with an inexpensive optical facemask that costs less than $10 to image whole pieces of tissue. It’s able to provide depth-of-field as much as five times greater than state-of-the-art microscopes today. The facemask is placed over the microscope’s objective to module the light that enters the microscope. Modulation allows for better control of depth-dependent blur in images captured by the microscope.
The control allows the deblurring algorithms applied to the captured images to faithfully recover high-frequency texture information over a much wider range of depths than conventional microscopes. The microscope can capture and process images in as little as two minutes.