Most machine vision system designers think about the cameras and lighting, but optics are often overlooked. There are hundreds of types of optics and optical assemblies available for use in the broad life sciences industry, like medical imaging, research, microscopy, slide reading, pill counts and quality control during manufacturing and packaging in pharmaceutical arenas.
Optics can be imaging, such as those that go on cameras, and non-imaging, like lenses that are put in front of a light to focus it, or in front of a pattern projector to project it when tracking the deformation of a surface when performing 3D imaging. While some applications aren't demanding from an optics perspective, others require very high accuracy.
One fast growing application in life sciences from an optics perspective is super resolution microscopy, used in life science research. Live cell imaging, DNA sequencing and other research applications need the most sophisticated custom optics, custom illumination and highly sensitive cameras to image at the diffraction limit and see below sub-micron details.
Another fast growing sector is the clinical diagnostic point of care application. There's a strong push now to develop instruments that allow doctors to draw and test blood and give instant results in an office or perform HIV testing in the field. The need for these types of instruments used in remote areas, in third world countries, or anywhere health care isn't available is growing quickly. The optics in these systems are not as sophisticated; they have an imaging lens and maybe a focusing lens for a small laser.
Unique Considerations for Life Science Industry
Although lenses go in unique environments in life science industries, inside scopes that are inserted into people or cameras that people swallow to image the inside of digestion tracks or into dental cameras, there is nothing unique from an optical point of view about the lenses that serve the market.
Machine and robotic vision applications don't need expensive lenses. Medical imaging needs lenses with high quality, tight tolerances, and robust mechanics, very low chromatic aberrations, and low distortion – properties also needed in other measurement industries.
It is important that the lens be able to image with the frequencies of light that are present. In medical imaging applications, illumination is typically done with LEDs and lasers. Small, portable vein-viewing systems used in doctor's offices and hospitals that draw blood from veins shine a near-IR light onto an arm revealing the vein below the skin. A lens looks at it, forms an image, and shows a real time pattern so the blood can be drawn on the first attempt, reducing the risk of injury for children or the elderly. Systems like these needed optics that could not only correct for white light in the 400-700nm range, but could image in near IR over 700nm-1000nm.
In some life science applications it is more likely that objects to be imaged will be human-sized, larger than objects in industrial vision applications. There is a wider spectrum of object sizes and fields of view in this market.
Users should start with the optics and specify optical quality and mechanical quality first. Quality and repeatability are the top considerations for optics going into life science inspection systems. In these types of applications, it is important to look for very high performing lenses that can work in low light conditions and that can be optically corrected (color corrected) from the visible to the Near IR spectrum. In the medical imaging industry, imaging occurs in the visible and near IR spectrum, with CCD sensors operating in the 400-1200nm range and LEDs and lasers illuminating in that range as well. Color correction is also important in the inspection arena, and some lens manufacturers implement color correction on all lenses.
How stable and robust a lens needs to be on a medical system is typically overlooked. In the growing field of medical surgery 3D imaging, the alignment between the multiple imaging elements must be maintained to create the 3D look. Maintaining alignment is important not only during the procedure, but for many years of use. It is almost as equally important as the quality of the optics.
Repeatability from lens to lens, lot to lot, and year to year is also a critical consideration for optics used in life sciences where tolerancing and traceability is important. Some lens manufacturers mark every lens with unique identification to maintain traceability throughout the lens's life. In systems with multiple cameras, like body scanners that have 24 cameras that image the skin used in dermatology, all of the cameras have to be calibrated and optically the same in terms of focal length and mechanical tolerances. Little to no variation is important if a camera and lens are replaced to avoid unmatched oddballs.
Optics are typically the last thing budgeted for or even considered during system design. Instrument OEMs usually have expectations about optical performance, but the size, cost, and other specifications of the optics are not always considered. Involve camera, optics, and illumination parties earlier, to make the process easier from an engineering perspective and prevent problems down the road for being over size or budget.
As LED lighting improves and camera pixels are becoming smaller and massively more capable, lenses are becoming more interesting. But, given the limitations of physics, lenses can only do so much at a reasonable price tag. It's important to have realistic expectations for the optics from the start and know what is actually achievable from a lens, which typically constrains the spatial resolution of a system. The more refined the measurement and the higher the accuracy needed, the more sacrifices have to be made. If a lens with sub-micron accuracy that equates each pixel to 0.01mm is displaced in the z-axis 0.5mm, the chances of being able to keep the object in focus is slim.
Researchers and others looking for machine vision systems rarely know the specifications for everything and don't often think through the entirety of the measurement, such as the amount of displacement of the object and its rate of motion. It's important to know the constraints of and resolution limited by the lens itself, not by the camera or available light. Everything is important since the systems are interdependent. Knowing the speed of the measurement, exposure times, depth of field of the measurement, and accuracy actually needed will help determine which lenses may be suitable at a reasonable price tag.
Fundamental Packing Constraints for Optics
Lens sizes are driven by the end user. When developing optics for use in a black box, dimensions, weights, thermal and vibration environments are specified by the customer who needs to be aware of the size and weight of the lens. Usually they get this information from the system producer or end user.
Time lapse images have to occur in controlled environments and can't sit out on a desk. If a lens will be used in a controlled environment, such as an inside an incubator that images cell morphology, moisture and humidity need to be considered. Sealing and managing the optics properly so no moisture enters the optical assembly prevents fogging from happening inside between the elements.
Some microscope objectives used in life science inspections inspect objects that are immersed in water or oil for days at a time. For water and oil immersion life experiments, the objective packaging has to be engineered properly so fluids don't leak into the objective, contaminate the lenses, and corrode.
Lens Advancements Are In the Customized Solutions
While lens fabrication and tolerance processes have gotten better, higher speed machines polish the glass quicker and more accurately, the basic architecture of a lens today is not that different from twenty years ago. Of the three components in machine vision, camera/illumination/optics, advancement in optics lags behind the other two. More creative illumination and powerful cameras have been commercialized in the last five years than new optics technologies or basic lens designs. The advancements in optics in this industry have been in the increase in requests for lenses that serve new applications and a trend in moving away from using off-the-shelf optics towards designing custom solutions.
Users who are pushing the envelope of performance stop using off-the-shelf optics. Designers who want to improve their optics/lens assembly in ways that make a difference, verify performance with off-the-shelf lenses to see how well they work and then bring their specs to create custom solutions to optics manufacturers. With so many solutions available from many optics companies, many times off-the-shelf optics are a great solution. When customers improve on something that already exists and create high performing optical assemblies, advancements happen, but it is in the customization, which doesn't always translate into a commercial success for the optics manufacturer or optics industry.
Cost of Optics
Costs for machine vision optics in life science applications are the same as in other industries; the same lenses are used and they solve the same problems. Lenses range from a few hundred dollars for simple lenses to several thousand dollars. At one end of the spectrum, for machine vision applications, fairly inexpensive optics that look at gross features can be used.
At the high end, in super resolution microscopy where features are smaller and more accurate lenses are required, the optics become more complex and expensive. The most accurate lenses, fast for low light levels with low f-numbers are used in medical systems, are difficult to make and control. Faster lenses are harder to produce with tight tolerances. Faster lenses also require special focusing mounts, adaptors and accessory mechanics to go with it. Tight tolerance is paramount in 3D imaging, and it is a price driver.
Main Challenges Life Science Applications Pose for Optics Manufacturers
The perpetual challenge for optics is that the desired resolution keeps increasing but lenses are limited in performance. In life sciences, the need for optics assemblies that have high resolution and are small in size and also have a good working distance from what is being imaging is a major challenge. For instruments that go into labs or on desktops to take up as little space as possible and still be powerful, sometimes the optics have to take up little space inside too, and cost as little as possible. Manufacturers engineer techniques to fight against the physics and use creativity. To save space they use mirrors, creative folding techniques, prisms, beam splitters, annular setups, and try to manipulate the packages as best they can.
The biggest challenge for a lens manufacturer for new life science applications is the fact that almost every project is a custom optic development and many users aren't used to long development cycles. Even though off-the-shelf lenses may meet most of what a customer wants, the lens is usually the last thing considered. If a customer comes with constraints after settling on specs for the camera, lighting, computer, drivers, algorithms, and system housing and then expects to find an off-the-shelf lens to fit the mostly designed system, it is often impossible to find something suitable. Customers don't realize that developing, quoting, settling on the final specifications for a lens, developing a prototype, and starting production can take a year for a new lens. Designing concurrently, at the beginning of the design process, and starting with a standard or stock lens and making a simple variation can cut the lead time down substantially, sometimes to as little as four months.
Product quality and repeatability are the most important things to look for in optics for life science applications. It is important to carefully evaluate test data with actual measured optical performances when choosing a lens to guarantee that it will be repeatable. It can be a battle to obtain manufacturing data about a lens, but lens manufacturers will sometimes provide it or allow a developer to evaluate or borrow a lens to put on the camera and see how it performs and compare. Users should understand the data, and know how it was measured.
It is also important to know lenses. It is straightforward to know what lens to put on a camera, but not so easy to determine which lens delivers the needed resolution. In wellplate imaging, optics need sufficient depth of focus to image an entire slide and focus into each well or several wells. In microscopy applications, the numerical aperture and ability to resolve very small features is important. One lens doesn't solve all problems. Each is highly dependent on what the imaging system is asked to do.