Accuracy and Technical Predictability of Computer Guided Bone Harvesting from the Mandible: A Cone-Beam CT Analysis in 22 Consecutive Patients

Abstract

This study assesses the accuracy and technical predictability of a computer-guided procedure for harvesting bone from the external oblique ridge using a patient-specific cutting guide. Twenty-two patients needing bone augmentation for implant placement were subjected to mandibular osteotomy employing a case-specific stereolithographic surgical guide generated through computer aided design. Differences between planned and real cut planes were measured comparing pre- and post-operative Cone Beam Computed Tomography images of the donor site according to six validated angular and displacement indexes. Accuracy and technical predictability were assessed for 119 osteotomy planes over the study population. Three different guide fitting approaches were compared. An average root-mean-square discrepancy of 0.52 (0.30-0.97) mm was detected. The accuracy of apical and medial planes was higher than the mesial and distal planes due to occasional antero-posterior guide shift. Fitting the guide with an extra reference point on the closest tooth performed better than using only the bone surface, with two indexes significantly lower and less disperse. The study showed that the surgical plan was actualized with a 1 mm safety margin, allowing effective nerve preservation and reducing technical variability. When possible, surgical guide design should allow fitting on the closest tooth based on both radiological and/or intra-oral scan data.

Keywords: accuracy; bone harvesting; computer assisted surgery; cone beam computed tomography; predictability.

Figure 1

Representative pictures of the surgical fields (top row) and corresponding 3D renderings of the surgical guides (bottom row) for each study group. The surgical guide was designed and positioned according to one of the three following strategies: (a,d), group (A) using CBCT bone surface data; (b,e), group (B) using bone surface data and a built-in reference point to the closest tooth obtained from CBCT data; (c,f), group (C) using bone surface data and a built-in reference point to the closest tooth obtained from both CBCT and intraoral scan data. Before bone incision, the surgical guide was securely stabilized to the bone by placing one 1.3 mm-diameter screw in the built-in hole of the surgical guide (green arrowhead).

Figure 2

Definition of the six indexes for the quantification of accuracy between the real cut plane (red area) and the planned cut plane (grey area). The three displacement errors, root mean square displacement error (δ_RMS), signed average displacement (δ_mean), residual standard deviation (δ_b), were computed from the distances $d_i$ of points $s_i$ on the incision surface from the ideal cut plane. The three angular discrepancies were obtained from the direction of the normal to the ideal cut plane ${\widehat{b}}^{*}$ in relation to three orthogonal axes defined on the real cut plane: $\widehat{b}$, normal to the real cut plane; $\widehat{n}$, normal to the cut direction; and $\widehat{t}$, tangent to the cut direction. The angle between ${\widehat{b}}^{*}$ and $\widehat{b}$ is the overall angular error (θ_tot), which is divided in two components: the signed around-tangent angular error (θ_t), and the around-normal angular error (θ_n). Further details are reported in [28].

Figure 3

Representative case from group A. The guide model (cyan) and its planes (thick cyan lines) are viewed from the parallel direction. The cut samples (black points, red when corresponding to a specific guide plane) are placed on the post CT soft bone (grey). Due to the lack of a position reference point, the surgical guide was fixed, during the surgery, in a more mesial position compared to the planned position. Red numbers indicate osteotomy cuts: mesial (plane 2) and distal (plane 1) osteotomies were more negatively affected in terms of accuracy compared to the medial (plane 4) and apical ones (planes 3 and 5).

Figure 4

Representative case from group B. The guide model (cyan) and its planes (thick cyan lines) are viewed from a parallel direction. Red numbers indicate osteotomy cuts: plane 1: distal; plane 2: mesial; plane 3–4: apical; plane 5–6: medial. The cut samples (black points, red when corresponding to a specific guide plane) are placed on the post CT soft bone (grey). The surgical guide was minimally shifted mesially compared to the planned position.

Figure 5

Representative case from group C. The guide model (cyan) and its planes (thick cyan lines) are viewed from a parallel direction. Red numbers indicate osteotomy cuts: plane 1: distal; plane 2: mesial; plane 3–4: apical; plane 5–6: medial. The cut samples (black points, red when corresponding to a specific guide plane) are placed on the post CT soft bone (grey). Note the extreme accuracy between planning and execution of the osteotomy lines.

Figure 6

Box and whiskers plots of the six accuracy indicators computed on the four osteotomy cut directions: mesial, distal, medial, apical. From top to bottom and from left to right: root mean square displacement error (δ_RMS), overall angular error (θ_tot), signed average displacement (δ_mean), signed around-tangent angular error (θ_t), residual standard deviation (δ_b), and around-normal angular error (θ_n). Red crosses indicate outliers. Horizontal bars connect groups of data having significantly different distributions: p-values in bold represent significance to Mann-Whitney test (different medians), while p-values in italics represent significance to Ansari-Bradley test (different dispersions). Reported p-values include a Bonferroni multiple comparison post-hoc correction.

Figure 7

Box and whiskers plots of the six accuracy indicator values grouped according to patient group. From top to bottom and from left to right: root mean square displacement error (δ_RMS), overall angular error (θ_tot), signed average displacement (δ_mean), signed around-tangent angular error (θ_t), residual standard deviation (δ_b), and around-normal angular error (θ_n). Red crosses indicate outliers. Horizontal bars connect groups of data having significantly different distributions: p-values in bold represent significance to the Mann-Whitney test (different medians), while p-values in italics represent significance to the Ansari-Bradley test (different dispersions). Reported p-values include a Bonferroni multiple comparison post-hoc correction.