---

Discussion

A linear measurement assigns a numerical value for the length of an object or between objects. Human craniofacial dimensions were first analysed by anthropologists and anatomists who were based on osteolytic landmarks. With the invention of the x-ray machine, measurements were made on 2-Dimensional (2-D) diagnostic image rendered from a 3-Dimensional (3-D) structure [7,8].

All conventional two-dimensional (2-D) imaging procedures suffer from inherent limitations such as magnification, distortion, and superimposition leading to the misrepresentation of anatomic structures [9]. For these reasons, 3-D digital imaging systems have almost superseded 2-D analog imaging. Computed Tomography and CBCT represents a paradigm shift from two-dimensional (2-D) to three-dimensional (3-D) data acquisition and visualization.

Identification of anatomical structures, landmarks and accuracy of linear measurements is found to be superior in CBCT when compared to 2-D imaging [10,11]. Several studies have been undertaken to calibrate and test the accuracy of the quantitative measurements of the CT images. In various studies the accuracy of CT scans were found to be 1%, 0.28% with a range of 0.1% - 3.0 [12-14]. Linear measurements made on human mandible by axial CT demonstrated the accuracy which was within acceptable limits (0.4-0.9 mm) [15]. Hence, there is enough evidence to suggest that CT provides diagnostically accurate images.

Last decade has seen a significant upsurge in the manufacture, utility and availability of CBCT. Low radiation dose, sub millimeter spatial resolution, reduced scanning time and increasing access, have made CBCT a preferred imaging modality for the maxillofacial region. Hence, assessment of accuracy and reliability of CBCT imagesin terms of linear measurements gains significance.

Two hundred and forty linear measurements were made on human dry mandible, out of which 226 measurements [95%] showed no significant difference between digital caliper measurements and CBCT measurements. The observed minimum discrepancy of 5% being considered clinically acceptable. Average discrepancy demonstrated by our study was 0.2-0.7 mm which is similar to previous studies done by Brown et al. Waltrick et al. and Torres et al. with values 0.45 mm ± 0.17 mm, 0.23 mm ± 0.2 mm, 0.68 mm to 0.72 mm [16-18]. Studies done by Stratemann et al. Moreira et al. showed lesser average discrepancy than our study [19,20]. Kamburoglu et al. concluded that CBCT provides highly accurate data with less than 1% error when compared to physical measurements [21]. Most of the studies which compared CBCT to physical measurements found a tendency of CBCT to underestimate the anatomic truth [6,16-24]. Lascala et al. who showed that CBCT measurements for internal and external cranial anatomic sites tend to be slightly smaller than the same measurements made with caliper [5]. Underestimation wasrecorded on CBCT when synthetic mandible were used [22,23]. Hilgers et al. reported good accuracy and reliability of CBCT but Brown et al. concluded that 3-D CBCT data set using commercial analysis software have variable accuracy [16,24]. Baumgaertel et al. reported a systematic overestimation when analysis was performed [25].

Cone Beam Computed Tomography is known to produce 1:1 images as the magnification of these images are auto corrected [26]. The reasons for underestimation of linear measurements on CBCT images can be due to measurement software or partial volume effect [25]. The software defined the center of the volume from which to measure. In other words, the software might actually have measured the distance between the midpoints of the voxels. The caliper measures the distance between the contours of the skull, while the measurement software could have measured the distance between the centers of the mesial and distal voxel. For the field of view used in this study, the voxel size was 0.25 mm. If measurements were made from the center of the voxel, half of the voxel would not have been included in the measurement on either side (Figure 2). This would lead to CBCT measurements that are 0.25 mm smaller than caliper measurements. But all measurements were not necessarily 0.25 mm less in our study. Hence, this may not be the only reason for underestimation [27].

A different theory termed “partial volume effect” could also explain our findings. According to this theory, if a voxel lies completely within an object, it would reflect that object’s density. However, if a voxel is at the junction of 2 objects of different densitieseg, bone and surrounding air-the voxel reflects an average value somewhere between the true values for bone and air (Figure 3) [27]. The ratio of bone to air in the voxel determines the voxel value. Such a “hybrid voxel” can be interpreted as part of the tooth or part of the surrounding air depending on the threshold value. High threshold values therefore create smaller than actual objects and vice versa.

The present study was designed to assess the reliability and accuracy of dental measurements made on CBCT images. Very few studies assessing the accuracy of CBCT measurements using dry skull have been documented in India. On reviewing the literature thoroughly, probably this is the first study made on Scanora 3D CBCT unit in comparing the linear measurements. One limitation of the present study is that no soft tissue or soft tissue equivalent was utilized so as to simulate a clinical situation. But using dry mandible could lead to better contrast in turn providing a higher level of accuracy.