Machine Vision Helps Connect Modern Medical Research with Tomorrow’s Miracles
| By: Winn Hardin, Contributing Editor
LEDs Improve Real-Time Diagnostics
LEDs are making their way into surgeries and diagnostic labs thanks to improving spectral balance and their inherent strengths over traditional incandescent and gas-discharge lamps.
According to Steve Kinney, business development manager for CCS America, Inc., his company’s LED research laboratory in Kyoto, Japan, is working to improve the spectral balance of its white light sources to give a more balanced, natural spectral profile, which is important for endoscopy and microscopy applications.
“Many suppliers will say they have a white LED at a given color temperature, but often, those statements are based on a simple RGB test, rather than a full spectrum,” says Kinney. “They’ll claim a color temperature based upon the ratio of total energy in the R, G, and B spectrums, but there are spikes and dips at certain wavelengths within each channel, deviating significantly from the ideal spectrum. CCS Natural Light LEDs have a color index rating of RA 98%, meaning that 98% of the points along the full spectrum compare with natural light. Medical diagnostic OEMs who are developing non-contact methods for measuring pulse and blood flow through the skin rely on this technology. The accurate spectrum of the Natural Light products combined with a NIR source is key to distinguishing difficult skin tones from the veins below the surface for precision measurements."
Combining white and NIR LED lights with multispectral cameras in new techniques allows surgeons to see fluorescent tags on cancer cells on monitors during surgery to make sure all the diseased tissue is removed. In other cases, CCS PFBR series light boxes are replacing traditional metal-halide lights used in endoscopic instruments.
“With up to 5 million lux peak, these LED-based products’ advantages aren't only lower power and reduced heat,” Kinney says. “In endoscopy where intensity counts, they are actually brighter than a 250 W metal-halide bulb, regardless of coupling distance.”
growth by an average of 2%. If GDP went up 3%, healthcare spending went up 5%. Unfortunately for the world at large, runaway healthcare spending isn’t exclusive to the U.S. The average age of people living in all developed countries is steadily increasing thanks to improving living conditions and medical advances.
“The world has a huge population, period. And globally, it’s an aging population,” says Greg Hollows, director of Machine Vision Solutions at Edmund Optics (Barrington, New Jersey). “And people don't like to be sick, or worse. So there’s
a lot of research money floating around [life sciences]. The world is realizing that it’s more expensive to put people in hospital beds than to develop equipment that helps them stay out of the hospital.”
Overcoming Optical Limitations
Just a few years ago, machine vision cameras, as well as scientific cameras, were focused on the commercialization of large sensor arrays and all the applications you could accomplish with cost-effective, high-quality 20-megapixel-plus cameras. The new challenge for industrial applications is how to build cost-effective optics large enough to accommodate 1-inch and larger sensors. And in the case of microscopy and analytical instrument applications, system designers want both a wider field of view and higher resolution in a smaller package if possible. “What does that mean?” asks Hollows. “It means they are interested in more than just off-the-shelf objectives.”
Today, medical diagnostic imaging systems are going through some revolutionary changes. In the past, the line separating R&D systems at universities, medical-device manufacturers, and pharmaceutical companies from production-level systems was clear. Imaging systems for R&D were very large, complex, flexible systems with high price tags, and production models were optimized for just one of a handful of tests and fit on a desktop. That part is still the same, but the rate of change is accelerating so quickly that start-up companies are bringing optimized production models to market almost as fast as the leading objective manufacturers of the world can develop “high-end” R&D systems. Where there were years between R&D and production system development in the past, now only months may separate the two systems,
which means hardware hasn’t had time to develop and ease the pain, time, and cost of miniaturization.
Edmund Optics regularly fields calls from start-ups who have very specific requests when it comes to objectives and optics for medical diagnostics and research instruments. “They’ll want field of view, and high resolution, and want it all in a fraction of the size of today’s optics,” explains Hollows. “But there are two costs that these start-up companies have to face: design costs and time costs. They want really fast development, and that’s the good part about Edmund serving both industrial and life-science applications, where the applications may vary but the terminology and products are essentially the same, with the difference being how you couple it and package it. We can get them an optics that is ‘good enough’ by leveraging designs and materials we have on hand.”
“But when it comes to the high-end space,” adds Edmund’s biomedical engineer Stephan Briggs, “good enough isn’t good enough. They need to maintain the value that differentiates them.”
And this is where it gets really interesting, adds Briggs, because large OEMs and start-ups are using a number of techniques to achieve super resolution, enhanced 3D, less expensive confocal microscopes, and other microscopy-based diagnostic systems that don’t use objectives at all. These new systems depend on automation and computer processing as well as new ways to control light to generate data that the human eye either can’t see, in the case of R&D systems that defy standard physical limitations, or doesn’t need to, because the result of the test is all that matters, such as cancer screening.
One example of a system that overcomes the physical limitations of optics in microscopy systems is Fourier Ptychographic Microscopy (FPM). FPM delivers high-resolution gigapixel images across very large fields of view. In a traditional microscope configuration, acquiring high-resolution images across large fields of view would require a very large and complex optic, with many different optical elements to overcome color and chromatic distortion. FPM uses phase information in addition to intensity information to computationally correct for optical aberrations, creating a virtually perfect optic by using a data fusion algorithm to combine phase data with high-resolution images through a technique known as aperture synthesis, which uses multiple images from different locations to create super-resolution images that exceed the diffraction limit of the optics and illumination source.
According to both Hollows and Briggs, it's the ability to control and process light that is making these new techniques possible, not necessarily revolutionary leaps forward in new technologies. “We’re seeing new techniques being made possible using existing technology,” such as liquid lenses that are resulting in ultracompact systems, Hollows says. “This is going to lead to the types of systems that futurist Michio Kaku spoke about at AIA’s business conference a couple of years ago, [like] non-invasive systems that use daily body waste to see if you’re healthy or not.”
Optical specialists at Coherent Inc. (Santa Clara, California) are helping to reduce the size of medical instrumentation further by developing cost-effective laser illumination sources in very small form factors that provide the spectral precision and right amount of optical intensity necessary for medical R&D and diagnostic systems.
“Our BioRay lasers were developed specifically to help bring PCR, flow cytometry, and similar instrumentation to desktop form factors,” says Wallace Latimer, product line manager for machine vision at Coherent. “These systems struggle to achieve a small form factor while handling the thermal load and spectral brightness they need. LEDs can be cheaper, but as broadband light sources they generate cross excitation and overwhelm the return signal for fluorescence applications, for example. LED output varies considerably over time compared to the spectral brightness requirements of FACS [fluorescence-activated cell sorting] and similar techniques. And with proper thermal control, and depending on wavelength, lasers are stable light sources for up to 20,000 or 30,000 hours of operation. LEDs last longer, but thermal degradation means their output will vary over time, as well as batch to batch.”
Adds Edmund’s Hollows: “All these components come into play for new medical instruments, filters, sensors, light sources, and controllability. When you’re talking about a single-point light source that you can control in a highly repeatable way, software comes into play that can analyze lots of minute details and build up highly sophisticated images—much better than just a collimator or laser pounding away at the sample where the control is limited to ‘on’ or ‘off’ or zero 100% intensity. We can thank the manufacturers of micromechanical displays and LED signs, and their control electronics for these advances.”
“Different industries are influencing each other, and that’s where the new technology is coming from,” Hollows continues. “AIA sits at a very interesting point, helping to bridge the technology of industrial imaging with life sciences. We need to educate our customers and let them know that we have an incredible breadth of technology and expertise from machine vision on the industrial side, but it’s also completely valid for life-sciences and other medical imaging applications as well.”