Hexapod-based ring fixators such as the TSF have simplified deformity correction due to their modularity together with web-based planning software. The high theoretical accuracies of 1/1,000,000 in. and 1/10,000 degrees are extreme for clinical practice, but with approximate correction accuracies of 1 mm and 1°, the TSF is a very powerful device for deformity correction [1]. However, the TSF can only provide accurate corrections if the MPs are measured and programmed precisely [1, 5]. Even with extensive surgical experience with the frame accurate MP measurements can be difficult, and slight errors result in residual deformity in angulation and/or translation requiring subsequent correction [1, 13, 14]. Although one of the major benefits of the TSF is the ability to conduct residual corrections without reassembling the frame, the consequences are treatment delay, longer healing indices and more follow-up X-rays.
The quality of radiographic MP determination is essentially associated with the quality of the ap and lateral X-ray images, which must be perfectly orthogonal to the plane of the reference ring to display it as a flat line on the image [7]. The clinical problem with the postoperative radiographic technique is that it relies on precise positioning of the central X-ray beam on the exact orthogonal plane of the reference ring. In our “in vitro” study, obtaining precisely adjusted X-rays of Sawbone® frame-mounting was not difficult. However, in clinical practice, accurate imaging can be problematic depending on factors such as patient positioning with frame assembly and the exact adjustment technique of the radiologist [5,6,7]. Therefore, Kanellopoulos et al. [5] proposed a custom-made radiolucent guide frame that can be attached to the TSF to guide the surgeon and radiologist in obtaining radiographs that are perfectly perpendicular to the reference ring [5]. The authors found that the guide frame was very useful in children and adolescents, but its application may be limited in obese adults and in patients who cannot tolerate the necessary positioning [5]. Recently, Wright et al. [7] proposed the “silhouette technique”: The authors used a shadow silhouette created by the reference ring projected as a single line to indicate an orthogonal view. For MP measurements, the entire reference ring must be depicted on the X-ray image, and the orientation and field view must be indicated to the radiologist by stickers applied on the reference ring before the X-ray. In all eight patients, the authors obtained adequate radiographs on which planning of the mounting parameters was possible [7]. In contrast to MP measurements on postoperative X-rays, Gantsoudes et al. [1] and Park and Bradish [4] used intraoperative fluoroscopic intensifier images. Therefore, image quality is the responsibility of the surgeon regardless of postoperative imaging. The frame is rotated as long as the reference ring is depicted as a flat line with the ap or lateral markers exactly superimposed.
In addition to the need for exact orthogonal images, all radiographic and intraoperative fluoroscopy techniques must consider magnification effects. Gantsoudes et al. [1] used an image intensifier to obtain exact orthogonal images but measured MPs with a sterile ruler directly off the frame and rancho cube instead of intensifier images. The authors recognized the magnification problem due to the offset of the rancho cube from the bone, although they reported that this is negligible as long as the origin is close to the reference ring [1]. However, a greater distance between the reference ring and the origin results in less accurate MP measurements (Fig. 3). Therefore, the ring mounted closest to the origin is usually defined as the reference ring [1]. To account for the magnification effect, Park and Bradish [4] proposed the use of a calibration ball on the intensifier images. With their technique, MPs are measured on a printed image, and the surgeon adjusts the values for the magnification factor by calculation. Kucukkaya et al. [3] described a CT-based technique for precise calculation of the MPs and proposed that exact calculation is impossible with radiographic techniques. However, our results demonstrate that this only applies to non-calibrated images. With calibrated X-ray and fluoroscopic images, magnification problems can be minimized as MP measurements, even at significant axial distances, are comparable to reference measurements in our study. The calibration tool used in our study simplifies the technique by automatically calculating the magnification effect. The image can be standardized to a known dimension and the magnification is calibrated to the size of the calibration marker. The measurement scale on the X-ray images is then based on the calibration tool of the planning software [15]. On non-calibrated images, the unknown extent of magnification results in higher measured values. While the magnification effect is relatively small at short distances, it increases proportionally with longer distances. This is the reason for the large differences at the furthermost distances of the axial MPs measured on non-calibrated images in our study (method A), whereas the ap and ml MPs showed significant but smaller absolute differences. The clinical impact of the smaller MP differences may be negligible as Gantsoudes et al. [1] proposed, but the extent to which an MP miscalculation or a residual or new deformity becomes clinically evident is unknown. In a clinical trial, Sokucu et al. [16] compared intraoperative versus postoperative MP measurements in 17 cases. The authors measured the parameters under fluoroscopy during surgery as described by Gantsoudes et al. [1] and compared the values to measurements taken from digital postoperative radiographs. In their report, whether the X-ray images were calibrated is not mentioned, and the measured MPs that were used for the actual correction plan are not indicated. Although the authors reported that they found no significant differences between the intraoperative and postoperative MPs, the compared values showed substantial variance, particularly the axial MP values: 95.3 (range: 25 to 155) mm intraoperatively versus 109.5 (range: 28 to 195) mm postoperatively [16], indicating that the greatest difference of a single MP value measured with the two techniques was 40 mm (155 versus 195). This obvious inaccuracy must result in new deformities during correction and lengthening and was very likely caused by magnification effects. In a bone model Kucukkaya et al. [3] showed for example that already a 10 mm-error in the MP measurements caused a residual translation of 7 mm during the correction of a 10° rotational deformity.
The difficulty of determining the correct magnification factor of radiographs for digital planning is a major problem and still of ongoing debate in orthopaedic surgery [17]. One of the major challenges in clinical practice is the correct positioning of the calibration markers [17, 18]. A structural measurement error may occur if the position of the calibration marker deviates too much from the plane of interest [17, 19]. This is a major limitation of our study as the markers (calibration ball or rancho cube) were positioned at the bone level and not at the exact level of the different virtual hinges (the respective steel balls). The extent of a potentially structural error in calibration cannot be discussed in our study as various MPs were determined on X-ray pictures without changing the position of the calibration marker (Fig. 2). In digital planning of hip arthroplasties, for example, a mean error of 6% and range from − 5 to 15% has been described despite the use of calibration markers [20]. Other studies found absolute differences between the actual and calculated size ranged from 0.16 to 1.40 mm [17, 21,22,23]. The clinical difficulties of positioning radiological markers are related to patient specific factors (e.g., obesity and/or bony deformities) [17]. In patients with external TSF-frames it is found to be difficult or sometimes might even be impossible to install a calibration marker at the correct plane as it may interfere with the circular rings, soft tissues, half pins and thin wires.
Another limitation is the requirement of positioning the reference ring in zero degree rotation, only controlled by eyeballing and not verified using a CT scan in our study. Therefore, the effects on MP measurements of any possible malrotation of the reference ring cannot be discussed. However, the TSF software allows for a non-orthogonal mounting of the reference ring only in the axial plane: this is the reference ring being mounted in internal or external rotation [24]. In clinical practice a rotatory frame offset is usually determined by approximation and therefore avoided as far as possible [3, 9, 10].
In our clinical practice, we have been using the intraoperative fluoroscopic technique, referencing images with a rancho cube at the bone level (method C). With calibrated digital images, a surgeon can define the origin postoperatively at any time and double-check plans and calculations with colleagues, enabling recalculations that may be required in rare cases during treatment. One disadvantage of this technique is the increased fluoroscopic use required to obtain adequate intraoperative images perpendicular to the reference ring and to the rancho cube. On the other hand, the surgeon controls the image quality independently, without requiring radiologists and subsequent postoperative X-rays to obtain orthogonal images of the reference ring. The slight deviation of the measured MP values relative to the direct measurements may be due to incomplete orthogonal rendering of the reference ring as it is only depicted partially on the intensifier images. Another reason for the measurement deviation might be explained by the problem of image intensifier distortion that occurs with intraoperative imaging [25]. The obtained images in this study were not corrected for distortion. However, even with large axial distances between the reference ring and the origin, the maximum deviation was no more than 4 mm (or a maximum percentage difference of 3%). We have been using this technique since 2012, and the number of correction plans due to residual deformities has been reduced from up to 5 per case to a maximum of 2 per case. However, this result may be affected by many different factors, such as increased surgical experience with the TSF system over the years.