How many times was the same implant measured?
A radiation-free way to locate an implant is attractive when imaging needs repeating. A new peer-reviewed method comparison tested agreement between intraoral optical scanning and cone-beam computed tomography in mandibular models. Its useful strength is the paired design: both methods assess the same physical implants, avoiding anatomical differences between patients. But the 180 paired assessments do not represent 180 independent implants. The experiment contains 60 implants in 12 models.[1]
Repeated assessments reveal how much a measurement varies; they do not substitute for fresh independent specimens. Assessments of one implant share its geometry and acquisition conditions. The authors acknowledge that subgroup comparisons treat repeated assessments as individual observations and label those p values exploratory. That disclosure matters. It prevents the apparent precision of a comparison from resting entirely on an inflated reading of the sample count. Differences between tooth-bounded spaces and distal-extension sites remain a methodological question for further testing.[1]
Two different sources of repetition
The repeatability comparison contains a subtler asymmetry. Each model receives one cone-beam acquisition, and that image is evaluated 3 times. Optical scanning uses 3 separate acquisitions, each evaluated repeatedly. The first arrangement measures variation in evaluating an existing image; the second also includes reacquisition. Within-implant angular standard deviation was 0.55 degrees for cone-beam imaging and 0.20 degrees for optical scanning. Lower optical variability under these conditions is useful because it includes the burden of fresh acquisition. The two values nevertheless do not come from a design that repeats both complete acquisition chains symmetrically.[1]
My inference is that this distinction matters when deciding which stage needs improvement. Stable optical measurements could reflect effective scanning and surface registration in these models. Alternatively, evaluation of cone-beam images or their resolution may contribute more of the observed variation. The experiment does not separate those contributions through symmetric acquisition repeats. The useful interpretation therefore attaches the lower dispersion to the stage actually measured, rather than extending it to every imaging condition.[1]
The reference behind close agreement
The average difference between methods is small: minus 0.05 millimetres at the implant base. Yet the 95 per cent limits of agreement extend from minus 0.62 to 0.53 millimetres. An average does not describe the complete spread of differences between individual comparisons. More importantly, the study has no independent ground-truth position. Closely agreeing methods can share a directional error. That possibility was not demonstrated here; the comparison lacks the reference needed to exclude it. The authors appropriately limit their conclusion about absolute accuracy.[1]
The fair conclusion is that radiation-free measurement has complementary value in specified posterior mandibular sites. Polymer models do not reproduce saliva, bleeding, movement or variable soft tissue. For clinical generalization, an independent reference to the implant’s true position and fresh acquisitions in both modalities provide a more informative test. The present achievement is a concrete measurement of agreement and variability. Naming the quantity on which a decision depends makes this bounded, clearly defined result usable.[1]