Important Specification of Endoscopes
Endoscope specification, as for any product, should relate to the intended use. As such, some specifications that would generally be viewed as important in optical design become less important in a clinical setting. For example, distortion may be a critical specification in industrial applications such as machine vision. In endoscopy, distortion is generally well-accepted even though levels of distortion in wide-angle systems can approach 100%. In fact, it has been argued that barrel distortion can be a useful feature in endoscopes: the magnification is high in the center of the image to provide a detailed view of the surgical site and is low around the periphery of the image to orient the clinician to the nearby anatomy.
Other specifications, however, may be more critical in endoscopes than in other products. For example, system length tolerance may need to be specified to less than 0.1 mm in order to be compatible with other parts of a surgical system, a tolerance tighter than is usually expected. Different endoscope types exist for use in almost every portion of the body. As discussed previously, endoscopes that enter the body through existing openings are generally flexible devices. The required insertion diameter depends on where the endoscope is used, but of course, smaller is better. Rigid endoscopes used in neurosurgery present a unique combination of requirements, a sort of “perfect storm” requiring a combination of high resolution, small diameter, and durability. Unfortunately, these characteristics work against each other and present a difficult challenge to the designer. Orthopedic endoscopes (arthroscopes) are also a challenge as these delicate precision optical devices are also used by the orthopedic surgeon as “crowbars”.
In a way, video endoscopes are the simplest scopes to specify from an optics point-of-view. They are merely miniature cameras. When considered as medical devices the situation becomes much more complicated: we must consider sterilization, articulation angle, bending radius, and so on. The current trend in 3-D endoscopes further challenges the designer.
In these endoscopes, there are usually two endoscope channels placed side-by-side to create parallax for the surgeon. The design engineer, therefore, not only has to consider the optical quality of the camera systems alone, but also the matching of the two optical channels. The surgeon’s brain does a good job of fusing two images together, but in the presence of aberrations such as angular misalignment, magnification mismatch, and resolution mismatch, the resulting image can quickly cause eyestrain and headache. The following is a discussion of the most important specifications influencing quality in endoscopes.
1. Field of view
Because the depth of field of endoscopes is so large, specifying field of view by the maximum diameter image that can be seen is not useful. The lateral field of view is, of course, completely dependent on the object distance. Therefore, field of view is usually specified as the maximum angle that can be viewed.
If the vertex of the field of view angle is positioned at the entrance pupil, the field of view specification is independent of object distance. Unfortunately, the location of the entrance pupil is generally not known except by the manufacturer/designer. It is a simple matter, however, to measure the linear field of view at two object distances a known amount apart. Simple trigonometry leads one to the true angular field of view. A systematic error is introduced if the simpler method of measuring object distance from the tip of the endoscope is utilized.
The ISO standard 8600-3 specifies a procedure for measuring field of view using the endoscope tip instead of the entrance pupil. A uniform object distance (measured from the scope tip) of 50 mm somewhat mitigates errors in comparing different endoscope samples for field of view. A problem remains, however, if the image is out of focus at 50 mm.
2. Direction of view
Sometimes the endoscope user needs to visualize a portion of the body that even wide field-of-view objectives cannot encompass. In other instances, the area of most interest to the surgeon is not directly in front of the endoscope. In these cases, a reflecting prism at the distal tip of the endoscope can be added. This permits an offset of the center of the image at a fixed angle relative to the shaft of the endoscope.
For example, in knee endoscopy, known as arthroscopy, a need usually exists to look at the side of the anatomy that is directly in front of the endoscope (see Figure below).
Most arthroscopes have prisms that incline the direction of view to either 30 deg or 70 deg relative to the shaft of the endoscope. Generally, the field of view should be designed wide enough to ensure that the view directly in line with the shaft is visible even with an inclined direction of view. With sufficient field of view, the risk of the tip of the endoscope inadvertently colliding with anatomy is, therefore, reduced.
The other benefit of attaching these types of reflecting prisms to the tip of the scope is that the surgeon, by rotating the endoscope shaft about its length, can sweep a wider field of view. So, for example, a 30-deg endoscope effectively increases the field of view by 2 times 30 deg, or 60 deg. This is very significant because in many cases, the surgeon is unable, because of anatomical restrictions, to tilt the endoscope itself to look at different areas. He or she can only rotate the scope about its length.
Direction of view is the angle formed between a line connecting the center of the endoscope entrance pupil to the center of the field of view and a line following the shaft of the endoscope. Measurement of this specification usually shares the systematic error of measuring from the endoscope tip (as described in ISO 8600-3) instead of from the entrance pupil.
Recently developed “swing-arm” endoscopes are rigid endoscopes with a movable mirror-prism at the distal end. These endoscopes have the advantage that the surgeon can adjust the field of view during surgery. For example, the surgeon may set a 15-deg direction of view in a sinuscope for navigation during the beginning of the procedure and then shift to 70 deg or more to better visualize the surgical site. Variable direction-of-view endoscopes are manufactured by Storz Endoscopy (EndoCA Meleon), among others.
Direction of view should be contrasted with articulation angle. Articulation angle is a mechanical curving of the last few centimeters of an endoscope through the use of internal wires under physician control. Many flexible endoscopes have two pairs of internal wires to enable a zero direction-of-view endoscope to be articulated in two orthogonal directions, in many cases by more than ±90 deg.
A cartoon shows a salesman demonstrating an endoscope with a 4-in. diameter shaft. “Yes,” the salesman admits, “but in every other respect it’s excellent.” Were it not for the requirement of very small diameter, endoscope design would be quite straightforward.
The author has seen groups of engineers meeting for an hour with the sole purpose of reducing the endoscope diameter by 50 μm. And it is not just the lens component diameter that needs to be minimized. As discussed previously, a typical knee arthroscope has a shaft diameter of 4 mm. Contained in that diameter is the outer tube with wall thickness of at least 100 μm, a tube of similar wall thickness holding the lens components, and sometimes a third tube separating the illumination optical fibers from the lens train to allow a more modular manufacturing procedure. Added to this is the needed space for the illumination fibers themselves. And then there is the wasted space created because an inclined direction of view arthroscope needs room to bend the fibers to the same angle as the direction of view.
Typically, the space left restricts the diameter of the lenses to around 2.8 mm.
Diameter specification drives almost everything about an endoscope design. Lens diameter limits the f-number of the optical system and thus the image brightness. This increases the need for more illumination optical fiber, but again there is no room for an arbitrary number. Even if there is sufficient light to reduce the lens diameter, the designer is now faced with an ultrafragile system that may break during normal clinical usage. An attempt can be made to reduce the stainless steel tubing wall thicknesses, but then the designer runs up against material strength issues.
There really isn’t any one fix. All one can do is optimize the optical design in concert with the mechanical design according to the clinical requirements of the device.
4. Chief ray angle
Modern CMOS image sensors often have an array of microlenses aligned to each pixel of the camera.
If the microlenses are placed directly in front of each pixel, the required chief ray angle is zero to maximize the collection of light on the sensor.
If the microlenses are displaced as a function of image size, then the optimal chief ray angle (in image space) should also be displaced as a function of image size. The required chief ray angle as a function of image height is given in the manufacturer’s specification of the image sensor.
The extreme ray angle in image space determines the minimum diameter of the lens system that is possible without introducing vignetting at the periphery of the image. The optical designer must consider the specified chief ray angle in the lens design as a parameter of equal importance to the image size.
5. Image Resolution
The specification of image resolution is critical to achieving a sharp image. Overspecification and the addition of unnecessary lens components may lead to an uneconomical cost of goods. Image resolution is generally specified by a modulation transfer function (MTF).
Because these lens systems are optimized to create an image that looks “best” to the clinician, low frequency MTF is generally the most important parameter to maximize. Much empirical research into image evaluation has been conducted by Edward Granger et al. at Kodak, creating the concept of subjective quality factor (SQF).Optimizing SQF generally achieves an image that is qualitatively seen as “sharp” with black blacks and white whites. Maximizing high-frequency MTF, on the other hand, may allow the visualization of pairs of lines that are closer together but that is usually not what the surgeon is looking for.
f-number specification has the distinction of being largely responsible for the three most important characteristics of an endoscope.
These characteristics are image brightness, depth of field, and resolution. As with normal cameras, it is well known that decreasing the f-number increases image brightness but decreases depth of field. Increasing f-number generally reduces optical aberrations and thus increases resolution. However at large f-numbers diffraction effects begin to dominate and the limiting resolution due to physical optics can be reached. It is the responsibility of the lens designer to understand the medical device and choose an f-number that represents the best trade-offs for the application.
Of course, it is dark inside the body. Therefore, some form of external illumination is required for visible systems. The notable exception is the mid-infrared endoscope, which is sensitive at the patient’s blackbody wavelength and does not require external illumination. Because of heat exchange issues within the LED circuitry when confined at the endoscope distal tip, illumination is almost always provided via an optical fiber bundle cable connected to a remotely mounted light source. As discussed previously, high radiance LED sources are replacing traditional xenon and tungsten halide sources used previously. Note that assumptions about the light source usually affect the other specifications.
8. Sterilization requirements
Repeated sterilization of the finished endoscope is a challenge not only to the mechanical structure but also to the optics. The most common sterilization regimens used in hospitals include autoclave, proprietary hydrogen peroxide system (Steris), proprietary gas plasma technology (Steris), and chemical disinfection (Cidex). Some sterilization processes can degrade antireflection coatings on lenses. Therefore, exterior optical surfaces are generally left uncoated. Other sterilization regimens can degrade the epoxies used to seal exterior optical components against leakage. Therefore, it is preferable to braze or solder exterior components.