Aging Population Boosts Medical Microscopy Vision Markets
| By: Winn Hardin, Contributing Editor
Pathology, or using a microscope to diagnose disease by evaluating blood and tissue samples, is one of the most widely used tools in the physician’s arsenal. And while the microscope has been around for hundreds of years, these systems will have to evolve if society is to stay on top of demand while keeping medical costs in check.
Machine vision is helping to solve this modern challenge by using the latest sensor and automation technology to increase the throughput and accuracy of microscope-based pathology systems.
Building a Faster Microscope
“We’re seeing new opportunities for using high-speed CMOS sensors in microscopy, digital pathology, and fluorescence applications,” explains Prof. Dr. Joost van Kuijk, Vice President Marketing and Technology for Adimec (Eindhoven, The Netherlands).
Traditionally, “microscopy” and “speed” are conflicting terms. For a variety of reasons, including photobleaching, damage to tissue samples, etc., microscope cameras require the highest possible quality and longer exposures with the lowest possible noise floor to produce the best possible image contrast at low light levels. But for hospitals and point-of-care operations, as well as the suppliers of automated pathology systems, running hundreds of samples through a workstation with long exposures is a bad business case.
“These systems traditionally use cooled, scientific-grade CCD cameras, perhaps at 8 megapixels and running 15 fps,” adds Marcel Dijkema, Adimec’s Strategic Product Manager. “Scientific-grade cameras are great for prototyping and developing the detection algorithm, but once you’re past the prototype stage, you need to build cost-effective machines that can process more samples faster, and that’s part of the demand for CMOS. As demand grows beyond the specialized medical lab to hospitals and remote locations, OEMs can’t put $20,000 CCD cameras into $100,000 systems. It’s just not economically feasible. They want systems that are faster, cheaper, and smaller. In other words, they want the best CCD image quality at the speed of CMOS technology. For example, a 4, 8 or 12-megapixel CMOS cameras at speeds beyond 100 images per second are fantastic.”
The CCD vs. CMOS competition has raged for years. Today, CMOS sensors are quickly approaching the image quality of CCD sensors. However, when it comes to medical imaging, it typically comes down to matching the specific sensor response to the specific needs of the microscopy system.
CCD has led CMOS in dynamic range and uniformity. As the CMOS sensor grows in size, the greater the chance of non-uniform response from individual pixels. Non-uniform pixels can lead to artifacts and misdiagnosis in microscope-based pathology and fluorescence systems. New CMOS sensor designs mitigate non-uniformity on the sensor, which helps to produce a better image.
Also, while CMOS has the greater theoretical advantage when it comes to dynamic range, in practical application, dynamic range is limited by the electron noise or noise floor of the sensor itself. In this area, CCD has led, but that’s changing.
“These microscopy systems are asking for a minimum of 60 dB, but you need to able to capture it relatively noise free,” adds Adimec’s Dr. van Kuijk. “There are many things we can do to correct for the imperfections of the image sensors in the camera but it all starts with a good pixel design for CMOS in addition to excellent analog circuit designs. Once you have that we can make a perfect camera that will perform for many years and comply with all rules and regulations such as FDA.”
In the end, the camera you choose really depends on the specific needs of the application. “We continuously stay up to date with the development of new image sensors by CMOS as well as CCD design companies. Choosing the best sensor for the right application and converting them to usable, reliable and reproducible cameras. That’s our game, “adds Adimec’s Dijkema.
While white cells in a blood sample may work fine at 4 megapixels, many other diagnoses require very high resolution across larger, non-fluid tissue samples. Much like inspecting a silicon wafer at high resolution, these systems need image processing to guide the mechanical scanning system and register individual images of a tissue sample into a massive image, measuring up in the hundreds of gigapixels.
“We recently helped develop a system that takes several thousand 8-megapixel images of a 15x15mm sample,” says Kerry Van Iseghem, Founder of Imaging Solutions Group (Fairport, New York). “This system will actually be used in hospitals, not just R&D labs. There are all kinds of things you can discover from images like this if you have enough resolution and can reasonably automate it. ISG built the camera, which uses FPGAs to do on-the-fly processing inside the camera using a host of algorithms.”
ISG has also worked recently on a new class of disposable endoscopes for orthoscopic surgery. Traditionally, a doctor would feed a fiber optic probe inside the body and turn the probe around until the image was at the proper orientation. This can lead to additional tissue damage inside the body and prolong healing times but is necessary for the doctor to remember what end is up during surgery.
The new system, developed in conjunction with Jeff Adair, CEO and Chief Technology Officer of Micro-Imaging Solutions LLC, uses a remote disposable sensor rather than optical fiber. This allows the system to go to higher resolutions without the cost of aligning individual fibers to each pixel; plus, throwing away the sensor and replacing it with a new one is cheaper than cleaning the older models.
But just as importantly, ISG designed the processing system so that the doctor can rotate the image using a thumb-driven joystick on the endoscope, eliminating the need to twist the scope in the body.
Whether it is advances in sensor design or vision-guided microscopic automation, machine vision is drastically changing the Methuselahian microscope. And while machine vision won’t be the total answer to escalating medical costs for an aging population, every technology that can provide better care at lower cost can only help.