The purpose of this study was to determine whether changing the acquisition trajectory of the CBCT in relation to the pedicle screw position could reduce metal artifacts and consequently improve image quality and clinical assessability of the screw position on the artificial bone model.

In summary, it was shown that maximizing the angular change of the acquisition trajectory, both by swivel and angulation, can highly significant increase subjective image quality and clinical assessability (visual examples in Fig. 10).

Fig. 10

Visual examples of pedicle screws in Th12 and L4

Pedicle screw angulation

Over the entire spinal area examined, image quality is highly significant improved by increasing pedicle screw angulation in all anatomical regions evaluated.

The worst subjective image quality is achieved with a pedicle screw angulation of 5° in Th10 (1.78 ± 0.76). The best subjective image quality was achieved by L5 with a pedicle screw angulation of 30° (4.63 ± 0.55).

The effect of changed acquisition trajectory is strongest on the vertebral pedicles (d = 1.06), especially in the thoracolumbar transition (spine area 1) (d = 1.52).

A view at the vertebrae examined shows: Looking at the delta between the pedicle screw angulation angle with the worst and best subjective image quality score, L1 benefits the most (Δ = 2.28) and L4 the least (Δ = 0.77) from the changed acquisition trajectory.

Angulated acquisition trajectories, often referred to as scan planes, have already been described in the literature for CT. They are used to spare radiosensitive structures with a tilted scan plane [20] or to minimize hardening artifacts caused by bone in skull imaging [21].

There are also studies on the influence of angulated scan planes in CBCT. Zhao et al. [22] found that in cervical spine imaging, angulation of the scan axis by − 35° nearly halved the incidence of image noise and doubled the contrast-to-noise ratio in subsets. This examination also focuses on avoiding radiopaque bony structures.

Related to the avoidance of metal artifacts in CBCT, Wu et al. [23] were able to develop an algorithm that can reduce the occurrence of "blooming" artifacts by 46–70%. Again, angulation of the acquisition trajectory is a component contributing to this positive result. However, since this angulation is subject to dynamic change via the orbital rotation of the 3D C-arm, no fixed angulation angle for optimal artifact reduction can be drawn from the study of Wu et al.

The study situation shows that angulation of the acquisition trajectory can have a positive impact on image quality as well as metal artifact reduction in both CT and CBCT. Minimizing the amount of metal in the beam path by altering the acquisition trajectory also reduces the likelihood of photoelectric effect [24] and beam hardening [25], leading to a reduction in artifacts.

Previous studies, reporting angulation angles, always referred to the change from the 0° setting on the device, but not to the angle with respect to the metal implant examined. Only in the study by Wu et al. the position of the metal implants was considered, but without outputting a defined angulation angle.

Exact comparisons between previous work on angulation in CBCT and this study are not possible because in this study the angulation angle was set in relation to the examining pedicle screws. Also, for the first time the focus was on an evaluation specific to every vertebral body and performed by qualified observers from a clinical point of view.

In several studies it has been shown that 3D imaging in spine surgery reduces the incidence of revision surgery. A recent analysis by Zimmermann et al. [11] reports that malposition was detected in 7% of inserted pedicle screws because of intraoperative CBCT, which led to intraoperative revision. The rate of revision surgery decreases significantly with navigated instrumentation using 3D-CBCT compared with the freehand technique using 2D fluoroscopy, from 4.38% to 1.35% [26].

It is likely that the use of angulated acquisition trajectories and the resulting improved image quality will reduce the rate of revision surgery even further than it has already been possible with the use of intraoperative 3D imaging versus 2D imaging.

Pedicle screw swivel

The purpose of the test series with the artificial bone model was to investigate the influence of the pedicle screw swivel on the subjective image quality. Due to the design, there was no possibility of compensating for the anatomically induced angulation of the pedicle screws in prone position. Thus, in this series of experiments, the influence of pedicle screw angulation on image quality varied depending on the initial position.

Pedicle screw swivel highly significantly improves image quality in spine area 1 (Th10–L1) in all categories of the questionnaire. In spine area 2 (L2–L5), however, there is only a significant improvement in the category of “vertebral pedicles.” Considering the effect sizes it also becomes apparent that even the significant results in spine area 2 are very weak compared to spine area 1.

These different results are caused by the fact that in spine area 1 the influence of the pedicle screw angulation was still small, because they were relatively close to the 0° of the standard acquisition trajectory (1°, 0°, 9°, 1°) in spine region 2; on the other hand, significantly stronger pedicle screw angulation angles (17°, 12°, 20°, 48°) and correspondingly stronger effects on the subjective image quality were seen.

The fact that the effect in spine region 2 is so small is also evident in the comparison of the subjective image quality between 0° pedicle screw swivel and 30° pedicle screw swivel. While in spine region 1 the subjective image quality in the baseline situation at 0° pedicle screw swivel always reached values in the lower range of the score (< 3) on average, in spine region 2 it was permanently above the score of 3 on average. Corresponding to the different baseline situation, the increase in quality over the increasing pedicle screw swivel angle turns out to be smaller. The effect size remains small.

It can be assumed that the strong influence of pedicle screw angulation, as already seen in the experiments on pedicle screw angulation on the artificial bone model, has ensured the good image quality at 0° pedicle screw swivel in spine region 2 and thus reduces the possibilities for increasing the subjective image quality by pedicle screw swivel.

To investigate this further, future studies would need to examine the controlled combination of pedicle screw angulation and pedicle screw swivel.

A literature review did not reveal any relevant results on the influence of swivel on the image quality of 3D datasets. This is also due to the fact that common, stationary CT scanners do not offer direct adjustment options to change the swivel.

In CBCT, intraoperatively, the 3D C-arm is probably already frequently run with a swivel angle over the pair of pedicle screws to be examined due to the limited space available. However, there are no measurements or experiments on the extent to which this angle influences image quality and the occurrence of metal artifacts.

A possible explanation for the improved image quality due to swivel could be that with precise positioning of the 3D C-arm, the central beam no longer runs centrally through both pedicle screws but radiates between them. As a result, there is less metal in the beam path and the image quality increases. This space between the two pedicle screws depends on the size of the respective vertebral body, which means that it is not always easy to target.

Compared to changing the angulation, changing the swivel of the 3D C-arm has the advantage that there is no need to intervene in the calibration of the 3D C-arm. In addition, the adjustment can be done faster and without significantly more effort than with normal intraoperative 3D imaging.

The advantage of the swivel is that there is no change at the C-arm settings compared to the normal CBCT workflow. The C-arm simply has to be moved "diagonally" over the ROI.

The disadvantage, however, is that the influence of the swivel on the reduction of metal artifacts is less effective than using the more complex angulation.

Different image quality of the spinal regions

In the "pedicle screw angulation" test series, as in the "pedicle screw swivel" test series, there is a dichotomy of subjective image quality in the starting position (standard acquisition trajectory). Consequently, there is a different potential for increase in image quality, which is the reason for a separate evaluation of two spine areas (Th10–L1 and L2–L5) with regard to significance and effect strength.

In the "pedicle screw swivel" test series, the different increase potentials in the spinal regions can be explained by the influence of the present pedicle screw angulation as described above.

In the “pedicle screw angulation” test series, there is no obvious covariate between the two areas. At this point it is helpful to take a look at the datasets created. In spine area 2, the occurrence of metal artifacts was also subjectively significantly minimized by pedicle screw angulation. However, due to the increasing size of the vertebral pedicles in area 2 compared to area 1, the metal artifacts in area 2 only slightly mask the pedicle cortical bone, which is why the assessability of this clinically very relevant structure remains good even without angulation adjustment, even in the presence of severe metal artifacts.

Clinical relevance

The remarkable influence of pedicle screw angulation as well as pedicle screw swivel on subjective image quality and consequently on the score for "assessability from a clinical point of view" shows that changes in acquisition trajectory are of clinical relevance.

Improved intraoperative image quality allows more precise detection of pedicle cortical injuries and better assessment of screw perforations. Especially in the thoracolumbar transition, which is characterized by relatively narrow vertebral pedicles compared to the lumbar spine, the increase in image quality can make a decisive contribution to the reliable assessment of screw position. The fact that most spinal injuries also occur in this region [3] further highlights the relevance.

The intraoperative use of CBCT can significantly lower the postoperative revision rate, reduce the radiation exposure for the OR staff, and minimize potential postoperative complications. At the same time, however, there are individual cases in which the intraoperative image quality of CBCT is insufficient to reliably assess the position of the screw due to artifacts [14]. This has been observed especially in combination with small vertebral pedicles, where the pedicle cortex has a small distance to the metallic implant.

Due to the significant increase in subjective image quality because of pedicle screw angulation and pedicle screw swivel, it can be assumed that the reported positive aspects of intraoperative CBCT can be further extended by optimizing image quality.

Clinical implementation

Angulation as well as swivel of the acquisition trajectory leads to significantly improved image quality in intraoperative 3D imaging (CBCT) with pedicle screws in situ. Implementation of this knowledge into daily surgical practice requires little additional effort. Sufficient angulation can be achieved by tilting the 3D C-arm or the operating table and by combining with the anatomically determined pedicle screw angulation.

With the Swivel, even less effort is required, as the 3D C-arm can simply be moved at an angle over the pedicle screws.

Limitations of pedicle screw angulation

For more precise adjustment of the pedicle screw angulation angles and to extend the range to 30°, a tilt table was used, which rotated the pedicle screws in relation to the 3D C-arm.

In the intraoperative setting, this is not possible to this extent with the aid of the operating table; here, the pedicle screw angulation may have to be adjusted or extended via the 3D C-arm. However, this C-arm angulation is currently only possible to an extent of 15° in the case of the Siemens "Cios Spin" device used.

Although the 15° 3D C-arm angulation initially appears to be a major limitation in relation to the possible quality improvement with 30° pedicle screw angulation, this problem appears less relevant on closer examination: On the one hand, many pedicle screws are already angulated in the beam path, which means that an addition of 15° should generally result in sufficient pedicle screw angulation in the beam path. Alternatively, the angle can be further increased by tilting the operating table additionally. On the other hand, C-arm angulation of up to 30° is also technically possible with the “Cios Spin” 3D C-arm. However, this is currently limited to 15° due to the lack of dynamic calibration.

Limitations of pedicle screw swivel

Due to the experimental setting, the angulated position of the pedicle screws could not be compensated for in the pedicle screw swivel experiments with the artificial bone model. Consequently, pedicle screw angulation also contributes to the image quality, but this was considered in the statistical evaluation.

Limitations of the artificial bone model

The results of experiments on the artificial bone model may differ from results obtained in a study on a human specimen or on a patient. Although the artificial bone model has a soft tissue sheath and was developed specifically for the medical imaging field, the transferability of results to patient use may be limited.

Limitations of the raters

Rater agreement was in the "fair" range for both swivel and angulation. Several factors have an influence here. Due to the relatively narrow scale with 5 points per question, even small deviations are associated with a strong decrease in the ICC.

While the ICC is also frequently used for metric measurements and can reach correspondingly high values here, the present data collection is a subjective assessment of the data. Therefore, values cannot be expected to be as high as with a metric measurement.