Accuracy or Precision – More than a Matter of Semantics

“Consistency [precision] is about engineering, accuracy is about knowledge.”

– Anonymous

Authors discussing various methods for targeting in Deep Brain Stimulation (DBS) use the term “accuracy” when, in actuality, they mean precision. Precision refers to the repeatability or reproducibility of a method or intervention. The question becomes, how often is the target “hit” on repeated trials, given a specific target? Precision in DBS lead implantation surgery is critical.

Accuracy refers to how often the actual target is reached, as opposed to an arbitrarily selected point. Throwing darts is a useful analogy. How often the “bull’s eye” is hit is a measure of accuracy. “Grouping” of darts, when each dart is thrown the same way, is a measure of precision, even if the “bull’s eye” is not hit. Thus, one can be very precise but highly inaccurate. However, if one is not precise, then one is more likely to be inaccurate. In other words, darts scattered all over the dart board may result in one or a few darts in the “bull’s eye,” compared to when the darts are “grouped’ in the bull’s eye. Typically, precision is related to instrumental variability and thus, precision is a better term for neurosurgeons, specifically when demonstrating the reproducibility of their technology and methods.

If one throws the first dart and then subsequently uses the first dart as the target for the next dart thrown, then the closeness of the subsequent darts to the first dart could be considered a measure of accuracy. Therein may lie the confusion. In the case of DBS surgical methodology, the use of the term accuracy might be appropriate in the sense of picking a target, analogous to throwing the first dart, and then seeing how often the DBS lead can be accurately placed within the target. However, this would be, at the least, a different notion of targets and accuracy than commonly used. Using the term precision may be less confusing.

The issue of accuracy versus precision is not academic. A neurologist, patient, family member or caregiver, reading a paper describing the high accuracy rate of DBS lead implantation using method A; which is less expensive, less resource demanding, and less traumatic, may think that method A leads to better clinical outcomes. Patients and their caregivers may be influenced to request a particular neurosurgeon using method A, which may impact the neurologist’s surgical referral. The neurologist certainly can appreciate the need for precision, but it is likely that the neurologist will view accuracy as being able to consistently place the DBS lead in a location that will provide maximal benefit and minimal risk. This concern is not addressed by the discussion of accuracy if accuracy is confused for precision.

The notion of accuracy, which predicts clinical outcomes, is important. Determination of accuracy in the context of clinical outcomes presupposes that the target is known. One could argue that the debate or dilemma is resolved when the target for precision is the same as the target for accuracy; which is defined as the target most predictive of maximum effective clinical outcome. In this case, the notion of precision and accuracy coalesce into the same concept. If, however, the precision target is not the same as the outcomes target, then use of the term accuracy in place of precision would be confusing. If the target for precision is not exactly the same as the target for clinical outcomes, then the literature describing accuracy can be misleading. At the very least, those wishing to use the term accuracy should clearly state that the definition of the term in manner to avoid confusion.

Based on the association of precision and instrumental variability, which is considered to be at the heart of most discussions regarding DBS lead implantation methodology, the term precision target will be used; reserving the term accuracy target for that which predicts the maximal clinical efficacy. If the precision target is an anatomical structure, such as the subthalamic nucleus (STN), then whatever method is used to identify the STN precision target, must also be used to distinguish the accuracy target in the STN. Only then, can the precision target be used as a proxy or equivalent for the accuracy target. Therein lies the problem.

The accuracy target is a physiologically defined target, specifically, the sensori-motor region which is not co-extensive with the anatomical STN. While it is thought that the sensori-motor region lies in the dorsal and lateral STN, it is not clear that targeting the dorsi-lateral region is sufficient. If this were the case, then there would be a good correlation between the location of the physiologically defined optimal target and the anatomical targets, such as the midpoint of the line connecting the anterior and posterior commissures, adjusted by direct visualization. Unfortunately, this has not proven to be the case. Thus, CT or MRI scans that cannot directly distinguish the sensori-motor region of the STN will not be able to claim that the precision target is co-extensive with the accuracy target. Thus, inferring clinical accuracy from precision would be highly problematic.

The same considerations hold true when targeting the thalamus. MRI and CT scans cannot distinguish between the ventral intermediate nucleus of the thalamus (Vim) and the posterior ventral caudal thalamus, which must be avoided. The situation is more problematic in the case of Vim and the globus pallidus interna (GPi), where the homuncular representation appropriate to the patient’s symptoms closely approximates the accuracy target.

Clearly, one advantage of intra-operative CT or MRI is the ability to account for brain shift. Brain shift can be a consequence of head position. The brain has a specific gravity (1.0489) (Torack RM, Alcala H, Gado M, Burton R. Correlative assay of computerized cranial tomography CCT, water content and specific gravity in normal and pathological postmortem brain. J Neuropathol Exp Neurol. 1976 Jul;35(4):385-92. PubMed PMID: 932786), slightly greater than cerebrospinal fluid (from 1.0063 to 1.0075) (Levin E, Muravchick S, Gold MI. Density of normal human cerebrospinal fluid and tetracaine solutions. Anesth Analg. 1981 Nov;60(11):814-7. PubMed PMID: 7197494), and consequently will sink to the lowest position, which is dependent depends on the head position. Thus, if the position of the head taken while the MRI and CT scan is performed compared to the position of the head in the operating room, the value of the pre-operative MRI and CT scan will be compromised.

Brain shift can be exacerbated by cerebral atrophy and pneumocephaly. However, if the brain shift is extensive, intraoperative CT or MRI may allow the trajectories to be corrected. But there is considerable evidence that suggests that the DBS lead should not be implanted in the event of significant brain shift, as resolution of the brain shift is likely to pull the DBS lead out of position ( Sillay KA, Kumbier LM, Ross C, Brady M, Alexander A, Gupta A, Adluru N, Miranpuri GS, Williams JC. Perioperative brain shift and deep brain stimulating electrode deformation analysis: implications for rigid and non-rigid devices. Ann Biomed Eng. 2013 Feb;41(2):293-304. doi: 10.1007/s10439-012-0650-0. Epub 2012 Sep 26. PubMed PMID: 23010803).
Perhaps further imaging techniques that can identify the physiological target may evolve. This would help move the precision targets closer to the accuracy targets, which measures of precision directly translate to measure of accuracy. At that point, clinicians may become more confident that demonstrations of precision will translate to optimal clinical effect. However, that time has not come.