June 30, 2017
by Scott Parker, DDS, AASDA
There are many 3D intra-oral and laboratory scanners entering the market, creating a growing trend of digital scan use for impressions, and replacing traditional techniques. With so many systems available, it is time for the practitioner to begin considering the transition to digital 3D scanning.
There is much discussion on industry blogs, dental industry web sites, and journals on the topic of digital scanning. Developers and manufacturers extol their own virtues and make claims of unsurpassed accuracy and resolution. While there is room for interpretation in the areas of best accuracy and resolution, the truth is that the majority of scanners on the market today yield excellent results, and can be implemented into daily practice quite easily. This begs the question: which scanner should be selected for the dental practice?
In order to meaningfully answer that question, there must be a basic understanding of what “accuracy” really means in context of its relevance to the practice of dentistry. With that in mind, it is helpful to understand four often-used terms as they relate to digital imaging:
Resolution, Accuracy, Stitching, and Interpolation.
The resolution of any imaging system is defined as its ability to distinguish two points as separate in space. The higher the spatial resolution, the smaller the distances that can be distinguished. Spatial resolution is primarily affected by how many pixels are contained on the surface of the imaging sensor, and the size of the sensor in relation to the pixel size.
A Charge Coupled Device (CCD) array is commonly used in capturing visual images. A CCD is an integrated circuit etched onto a silicon surface that forms light sensitive elements called pixels. Photons hit these elements and generate a charge that can be interpreted by electronics and create (thus creating) a digital representation of the object being imaged.
Ultrasound traditionally works in a similar fashion to visual imaging but instead of a CCD, the sensor is composed of a crystal element that is sensitive to sound vibrations (rather than photons) that are used to map the object being scanned. Recent advances in ultrasound utilize a process similar to CCD; sound wave-sensitive elements are embedded onto the silicon surface of an integrated circuit.
To better understand how pixels impact resolution, picture a sandbox that is 3m X 3m where the size of each pixel is roughly equivalent to the size of a basketball (25 cm). There is room for roughly 100 basketballs within the boundaries of that sandbox. Twenty basketballs could be used to create an image like the letter “S” in the middle of the grid. From a perspective above the sandbox, the letter will be clear but appears chunky or boxy, and is considered low resolution (Fig. 1).
Larger sized pixels can make an image look blocky.
Using the same sized sandbox (3m X 3m) but substituting golf balls (4.2 cm) for basketballs, nearly 5000 golf balls will fit within the boundaries of the frame compared to only 100 basketballs. To create the same letter “S” in the center, 1000 golf balls in the “S” space yield a much higher resolution. The finer the capacity of the sensor, the higher the resolution (Fig. 2).
Smaller sized pixels can make finer resolution.
For the clinician, resolution (when scanning a prepared tooth) is relevant in that higher resolution provides finer detail in the preparation such that the restoration will fit the tooth more precisely.
For a 3D scanning system, resolution is very important in all three planes. Clinicians are accustomed to viewing 2-dimensional radiographs. These planes are often referred to as the X and Y, or horizontal and vertical planes. Three-dimensional imaging adds an additional plane in the Z direction (or depth), which maps the surface parallel to the imaging source.
To better understand the Z plane, consider a cut away cross-section of a perfect sphere after the sphere is scanned. The cross section view shows where those digital pixels fall on the surface of the sphere (Fig. 3).
Axial resolution can be above or below the actual plane of the surface.
Accuracy is often confused with resolution. While they are related, accuracy is not the same as resolution.
Accuracy is defined as the amount of certainty in a measurement with respect to an absolute standard, or the degree to which a measurement conforms to the correct value or a standard. Importantly, accuracy should include the range or margin of error that is inherent in the measurement.
It is important to understand that averaging is often used to improve accuracy. The more times you measure a specific distance, the more accurate the averaged measurement can be. However, the practice of averaging can be deceptive as it often makes a scan look better than it actually is, and can therefore compromise consistency by blurring the crisp and highly defined edges of a preparation margin.
When scanning with a handheld optical scanner, multiple images are taken every second. Each image is indexed to the previous images. Thus, the resulting 3D image is a collage of many stitched images.
I grew up hiking in the Sierra Nevada Mountains of Northern California. Cameras at that time could not take in a wide panoramic scene in a single frame; I took multiple pictures by moving the camera horizontally across the scene while snapping images. After developing the pictures, I taped the images together to create a larger collage that could never have been captured in a single exposure. I was stitching my images together to create a single whole image from multiple smaller pictures.)
Another example of stitching is measuring a 40 ft. room using a 1 ft. ruler. The resulting measurement is most likely much different from measuring with a 20 ft. tape. A higher capacity to measure small spaces does not imply that the system is superior for measuring a large span.
The challenge with the stitching process is that every stitch introduces a minuscule error. By the time many images are merged together, accuracy in the larger image may be lost. While this error is not significant for small spans of stitched images, such as single teeth or quadrants, the inaccuracy is greatly magnified for images that use thousands of images stitched together.
Stitching is often problematic with full arch scans. The error may be substantial in both the occlusal plane and the arch width dimension. Many 3D imaging equipment providers correctly warn clinicians that their intraoral systems are not recommended for more extensive full mouth cases due to this possible accumulation of small errors. The distortion is too great for applications such as full mouth implant cases where there is no room for error; implants are inflexibly anchored in bone and cannot tolerate unpredictable placement or loading.
When an object is scanned, depending on the resolution and accuracy of the system, data is often “filled” into missing areas. Software can use complex mathematical algorithms to determine where data is missing, and exactly how it can be improved. This process is called interpolation.
In the example of the sandbox mentioned previously, consider the blocky letter made from basketballs. Software can be used to define and smooth the borders of the letter and then to fill in all excess space with golf balls. The image will appear much smoother, with clearly defined borders, making it look much more crisp. In essence, intelligent software has improved the scanned resolution, making the image look much more defined (Fig 4).
Interpolation can use smart algorithms (math) to smooth and fill in missing data.
3D ultrasound scan of full arch plaster model (green) superimposed on optical scan (gray) of same model for metrology analysis showing an average margin of error under 100 microns.
3D ultrasound scan of full arch plaster model (gray) superimposed on optical scan (blue) of same model.
Given the above parameters, what are the questions that are relevant to clinical practice, and how do they impact the purchase of a 3D scanning system?
1. What is the native scan resolution before averaging and interpolation?
When the scan generates higher native resolution or raw images, there is decreased need to use software to enhance the capture. The acquired image is a good representation of the scanned object. The restoration or appliance is more likely to fit well with a more accurate system. When a whole arch scan is accurate in all three planes, full mouth cases will fit correctly (orthodontics, implants, full mouth restorations).
2. How accurately does the scanner scan?
The selected scanner should offer more than current needs require; new procedures and new technologies are continually expanding the scope and breadth of clinical practice and the scanner should be able to grow into these opportunities.
3. What is the margin of error (accuracy) of the system
The selected scanner should have a low margin of error in order to yield the best possible results. While digital scanners show excellent results for most single tooth restorations, and even three tooth fixed partial dentures, it is generally recognized that scanners with a larger margin of error still fall short in complex full mouth cases.
4. How accurate is the scan around a whole arch
Deviations in the horizontal planes when taking full mouth scans are a known challenge. In order for a full arch scanner to be effective in full arch dentistry, it would need to reproduce the arch form in all three dimensions without distortion.
With this understanding of accuracy and resolution, and its importance to the growth of digital dentistry, there are good clinical reasons reason to pursue ultrasound as a technology platform. This effort is being led by S-Ray Incorporated (Portland, OR). www.srayinc.com.
Key reasons for this effort are:
• Sound does not require line of sight to image a target and is capable of imaging through liquids such as saliva and blood.
• Innovative subgingival imaging techniques that do not require retraction to obtain an ideal digital impression are currently being evaluated.
• Current designs for ultrasound imaging technology scan and register all three planes of the teeth simultaneously (buccal, occlusal, and lingual). By scanning the three planes concurrently, there is a significant reduction in the back and forth movement of the scan head, and thus a reduction in the potentially arbitrary stitching of images. This reduction in stitching leads to a more accurate image with improved cross-arch accuracy.
• The signal-to-noise ratio is typically better with ultrasound compared to traditional light-based imaging. This improved signal-to-noise ratio requires less filtering and manipulation of the raw data to create a good 3D image. Decreased manipulation of the data offers better reliability in the scan, with an improvement in the accuracy and reliability of the raw scan data.
While ultrasound will provide advantages over optical systems, 3D scan data in an open file format will enable a practitioner or dental lab to utilize multiple scan technologies to enhance treatment presentation, planning, and evaluation. S-Ray’s current software development direction includes “image fusion”, a path that accommodates other system scans into ultrasound scan data. OH
Oral Health welcomes this original article.
Dr. Parker is the Clinical Director of S-Ray Incorporated.
About the Author
Dr. Parker graduated from Loma Linda University School of Dentistry in 1996. Since then, he has been involved in teaching clinical techniques and technology to dentists across the US and around the world. His course reviews are consistently outstanding because of his commitment to compassionate care and higher education while still being grounded in the reality of running an active dental business. Dr. Parker serves on multiple clinical advisory boards for dental companies offering his insight into product development. As a member of the Catapult Group since its inception he has been designated “elite” membership status for his dedication to continuing education and clinical excellence.