![Figure 1: Top and side view of simple construction for multi camera system (based on 2009 patent [5]).](https://www.insightnews.com.au/wp-content/uploads/2020/08/Dispensing.png)
Ranging from rulers through to tower-based systems, Grant Hannaford examines the various measuring techniques and systems used in practice and how they stack up.
Part one of this column discussed the ways that accuracy and precision may influence the results of measurements in practice. So how does this relate to our day-today practice? It is after all, a somewhat abstract concept that seems more at home in a lab than in an optical practice.
In reality, we should all consider ourselves to be performing scientific measurements on a daily basis. With this mindset we will be able to approach each patient as a unique case rather than another face in the crowd and ensure their lenses are fit as close to the ideal fitting point as possible.
Now that we have a feeling for precision in measurements, we can look at how these techniques compare. This can be a fairly emotive issue as practitioners may mistake a discussion of precision as a critique of their performance. This is absolutely not the case. Looking at the accuracy of the various methods from the previous article provides a glimpse of the differences in precision and therefore, the potential sources of error that must be overcome in order to produce accurate data.
We are all familiar with manual techniques such as rulers and measurement jigs that are mounted on the spectacle frame. Essentially they are the same, simply providing a scale for measurement and have come in many forms over the last 50 years. The primary difference between these two methods is the ruler is hand-held by the operator while measurement jigs are mounted on the patient’s frame. Mounting the ruler or scale on the frame we are able to remove a potential source of error due to hand movement from the operator, though there is still an issue with parallax error. These manual methods have been found to have significant dependence on the training of the operator, with experienced operators having the best results. Nonetheless the range of values generated, even by skilled operators, can vary by over 4mm ([1-3]) for the same patient which makes this style of measurement unreliable for lenses requiring higher precision.
App-based measurement systems tend to be developed by established lens laboratories or independent sources such as developers on the app store. These types of apps often use a reference device to obtain their measurements like a frame mounted jig or, for the apps aimed at the home user, a credit card.
By placing a reference of known dimensions into an image, the relationship between a known element and the unknown facial parameters provides a ratio for determining lengths etc. Devices from lens labs that are built for purpose deliver far more consistent results than an ad-hoc system like credit cards. Angular measurements and CRR data are not usually included in the lab-based systems and are not included in the ad-hoc (credit card) systems. ‘Tru vision’ and similar features found on modern phones can use the depth detection feature to develop a model through simple trigonometric relationships as well as ratios as used in the other methods. These tend to be very susceptible to angular issues like head rotation, cant and tilt. We have observed errors of over 8mm in measurements induced by head movements of less than 10 degrees, while these features present interesting potential they are perhaps not mature enough to deliver reasonable accuracy at this point.
Tower based systems use a similar theory regarding trigonometric relationships to develop data [4-6]. In single camera systems a known element like a frame jig provides the ‘base’ of a triangle from which data is drawn. Multiple camera systems reverse this concept by placing the ‘known’ part of the system on the measuring unit itself in the form of camera angles and distance between cameras (see Fig 1). In essence, these systems aim to provide greater accuracy by reducing the number of variables, degrees of motion/freedom and hence overall uncertainty.
Ultimately the purpose of these attempts to increase accuracy is so the benefits of more precise lens design and generation may be realised while minimising the effects of small changes in frame placement as the patient takes their glasses on and off.
The development of these systems is covered more comprehensively in a webinar on our website.
REFERENCES:
1. McMahon, T.T., E.L. Irving, and C. Lee, Accuracy and repeatability of self-measurement of interpupillary distance. Optom Vis Sci, 2012. 89(6): p. 901-7.
2. Brooks, C.W. and I.M. Borish, System for ophthalmic dispensing. 3rd ed. 2007, St. Louis, Mo.: Butterworth Heinemann., 665 p.
3. Holland, B.J. and J. Siderov, Repeatability of measurements of interpupillary distance. Ophthalmic Physiol Opt, 1999. 19(1): p. 74-8.
4. Altheimer, H., Method and system for optimizing a spectacle lens based on individual parameters of a wearer, U.P. Office, Editor. 2014, Rodenstock GmbH, Munich (DE): USA.
5. Rodenstock, Device and method for deterimining optical parameters US20090021693A1.pdf. 2009.
6. GMBH, C.Z.V.I., BILDAUFNAHMESYSTEM UND ANPASSSYSTEM. 2018.
ABOUT THE AUTHOR: Grant Hannaford is the co-founder and director of the Academy of Advanced Ophthalmic Optics. He has been practicing in optics for more than two decades and works with optometry and dispensing students, as well as industry professionals.
