Detectability of fiducials’ positions for real‐time target tracking system equipping with a standard linac for multiple fiducial markers

Abstract Purpose To investigate the detectability of fiducial markers’ positions for real‐time target tracking system equipping with a standard linac. The hypothesis is that the detectability depends on the type of fiducial marker and the gantry angle of acquired triggered images. Methods Three types of ball fiducials and four slim fiducials with lengths of 3 and 5 mm were prepared for this study. Triggered images with three similar fiducials were acquired at every 10° during the conformal arc irradiation to detect the target position. Although only one type of arrangement was prepared for the ball fiducials, a three‐type arrangement was prepared for the slim fiducials, such as parallel, orthogonal, and oblique with 45° to the gantry‐couch direction. To measure the detectability of the real‐time target tracking system for each fiducial and arrangement, detected marker positions were compared with expected marker positions at every angle of acquired triggered images. Results For the ball‐type fiducial, the maximum difference between the detected marker positions and expected marker positions was 0.3 mm in all directions. For the slim fiducial arranged parallel and oblique with 45°, the maximum difference was 0.4 mm in all directions. When each slim fiducial was arranged orthogonal to the gantry‐couch direction, the maximum difference was 1.5 mm for the length of 3 mm, and 3.2 mm for the length of 5 mm. Conclusions The detectability of fiducial markers’ positions for the real‐time target tracking system equipping with a standard linac depends on the form and insertion angles of the fiducials.


| INTRODUCTION
To improve clinical results for pancreatic and prostate cancer, hypofractionation, stereotactic body radiation therapy, and microboosting to high-risk areas are often performed. 1 Hence, there is a growing significance for the motion management of tumors during irradiation. [2][3][4][5] In the chest, abdominal, and pelvis regions that have respiratory-induced motion and physiological position errors, the bony anatomy matching results in clinically significant errors in the target localization and verification. To deliver radiation without geometric misses in these regions, implanted fiducials are useful because they help to predict the tumor's position during irradiation. 6,7 To aid respiratory management, several systems for real-time tumor tracking, such as Calypso (Varian Medical Systems, Palo Alto, CA) and Micropos (Micropos Medical AB, Goteborg, Sweden), are now available in clinics. [8][9][10] More so, real-time tumor-tracking systems with oblique kV images for CyberKnife, O-arm, and a standard linac were developed and applied in clinics. [11][12][13][14] In these systems, both kVsources and detectors were fixed on the floor and ceiling. Therefore, the angles of acquired kV images were constant during the treatments. been several reports about the clinical effectiveness of ABH. 15,16 ABH has a limitation in that the visibility of slim fiducials among triggered images varies due to a change in the kV source and detector by rotating the gantry. There have been few discussions about the detectability of ABH in various types of fiducials. The geometric miss occurs due to a wrong detection of targets' positions, and this decreases dose coverage for tumors. 17 Moreover, a wrong detection of tumor positions in ABH may significantly increase treatment time as the ABH system does not allow to deliver irradiation without the images, which indicate that markers' positions were correct. [14][15][16] The aim of this study is to investigate the detectability of fiducial markers' positions for real-time target tracking systems, which are equipping with a standard linac. The study's hypothesis is that the detectability depends on the type of fiducial markers, orientation of marker placement, and the gantry angle of acquired triggered images.

2.A | Fiducials and display settings
Nine types of fiducials were embedded on the surface of a cuboidal water-equivalent phantom of size of 1.0 × 1.0 × 4.0 cm 3 (Taisei Medical, Osaka, Japan) (Fig. 1). The center of the fiducials and cuboids coincided. In each fiducial, three similar cuboids were prepared. Table 1 summarizes the name, form, measured diameter, and measured length of each fiducial. As presented in Figs. 2(a) and 2(b), the plates with sizes of 16 × 16 × 1.0 cm 3 , including three similar fiducials, were created by displaying several sizes of cuboids with and without fiducials, and they were inserted in the I'mRT phantom (IBA Dosimetry, Bartlett, TN). The surface of the plate including fiducials coincided with the center of the I'mRT phantom.
To simulate insertion angles of fiducials in a body, three settings were prepared. In the first setting (Plate_A), shown as Fig. 2(a), each slim fiducial turned to the gantry side. Here, the visibility of the fidu- Plate_B represents fiducials which were inserted diagonally to the body's axis. Finally, to simulate that when fiducials were inserted at 45°for a body axis, Plate_A 45 was prepared as shown in Fig. 2  The I'mRT phantom with the plate including fiducials, and the QUASAR platform (Modus Medical Devices, Ontario, Canada) with infrared reflective marker were placed on the treatment couch as shown in Fig. 2(b). The QUASAR platform was used to produce respiratory motion and acquire the gate signal for the ABH system. Therefore, the I'mRT phantom was not placed on the platform.

2.B.1 | ABH Settings for acquiring triggered images
During the experiments, the QUASAR platform moved sinusoidally to use the respiratory management system of TrueBeam STx (Varian Medical systems). In the amplitude gating, the gated region was set to cover the whole range of the trajectory to prevent the wave form from stopping the irradiation during the experiments. The treatment console was set to acquire the triggered images at every 10°. Therefore, they were acquired with 10°step from 90°of OBI kV source angle. The image acquisition parameters, such as tube voltage, tube current, and time, were 102-116 kVp, 100-172 mA, and 0.05-0.058 s, respectively. In the ABH system, marker settings (diameter and length) and error tolerance should be determined before irradiation. Error tolerance value indicates the distance of detected marker position from planned marker position, which determine whether beams continue or stop. A straight type of GoldAnchors (Naslund Medical AB, Huddinge, Sweden) with lengths of 10 and 20 mm were excluded in this study as there was no setting for marker lengths >5 mm in the ABH system. In this study, to acquire all triggered images continuously during the arc even in the detection of setup errors, the error tolerance was deliberately made large, for example, making the diameter 3.0 cm. These parameters could also be changed during irradiation.

2.B.2 | Form of fiducials on triggered image
In Plate_A, a one-marker setting was selected in the ABH system, and it is presented in Table 1 for each fiducial. In Plate_B, a twomarker setting was selected in the ABH system for slim fiducials of length 5 mm. The marker length in the triggered images changed from diameter to length. For example, in VISICOIL, the marker length in the triggered images changed from 0.5 to 5 mm during the arc.
When the marker length in the triggered images was <2 mm, a marker setting of 5 mm length in the ABH was inadequate. Hence, the ABH system could not detect the marker's position. When the marker's length in the triggered images was approximately 5 mm, the marker setting in ABH was 3 mm in length, and the ABH system could not also detect the marker's position. Therefore, two types of marker settings were prepared for Plate_B with VISICOIL and Lumi-Coil (Boston Scientific, Boston, MA) of length 5 mm. In Acculoc (Civco Medical Solutions, Iowa, USA) with a length of 3 mm, the marker length in the triggered images changed from 1 to 3 mm.
Hence, a one-type marker setting was adequate for Acculoc. In Pla-  When the setup error (d mm) occurred in the anterior direction, the following formula was used: When the setup error (d mm) occurred in the left direction, the following formula was used: where θ is the kV source angle of the triggered images.When the setup error (d mm) occurred in the superior direction, following formula was used: The The paired student's t-test was used to examine the statistical significance differences in the detectability between following five settings.

| RESULTS
In the plate for Plate_A and Plate_B, Figure 4 Table 2 shows the absolute mean errors with each fiducial in the plate for Plate_A. In Plate_B with VISICOIL, the expected and detected values were mutually inconsistent at some angles. Table 3 Table 3 shows the absolute mean errors with each fiducial in the plate for Plate_ A 45 . In each fiducial and direction, the absolute difference was <0.5 mm. There was no significant difference between Plate_A and Plate_A 45 for VISICOIL. There was significant difference between appropriate setting and inappropriate setting.

| DISCUSSIONS
The detection of fiducials' positions in a body using a linac equipped with an imaging system is novel without previous investigations. The Abbreviations: SE, setup errors; GA, GoldAnchor; LC, LumiCoil; Dir., Directions; Ant., anterior; Lt., left; Sup., superior. In this study, it was verified that changing the visibility of the fiducials affected the detection of the positions in the ABH system. Further, this study suggests optimal fiducials and insertion angles of fiducials in a body for the ABH system.
The insertion angle of fiducials in Plate_A was parallel to the body's axis. As presented in Table 2, the maximum error is 0.4 mm for Plate_A in all fiducials. When the insertion angle of the fiducials was easily parallel to the body's axis, it was found that the detectability of the fiducials' position with the ABH system was highly accurate for each fiducial. In previous studies involving prostate cancers, the insertion angle of the fiducials implanted from the perineum was found to be easily parallel with the body's axis, 6,18 and the angle of the fiducials implanted from the rectum was easily oblique with the body's axis. Rosario et al. 15 also used the ABH system for VMAT in prostate cancers, precisely the 1.2 mm × 5 mm gold marker (QLRAD, Zwolle, The Netherlands). The conditions used by these researches are likely similar to those of Plate_A and Pla-te_A 45 using VISICOIL. In prostate cancer, the detectability of the ABH will be accurate for any fiducials.
Furthermore, it was believed that the insertion angle of fiducials was perpendicular to the body's axis in a transcutaneous placement of fiducials in the liver, because short path for the insertion is better for patients. In patients placed with VISICOIL on liver at our institution, the insertion angles for fiducials were 90°AE 10°in 10%. The insertion angles are similar to those of Plate_B. In the settings, the changing lengths of the fiducials in the triggered images were more pronounced than in Plate_A and Plate_A 45 . Therefore, in Plate_B of the slim fiducials, there were angles at which the length of fiducials in the triggered images significantly differed from the setting length in the ABH system. At these angles, the detectability of ABH reduced swiftly. When the slim fiducials in Plate_B were used, users needed to change the settings of fiducials in ABH during the arc.
However, in Plate_B, short slim fiducials, such as Acculoc, had smaller errors than the other slim fiducials, such as VISICOIL. This was due to the smaller changing lengths of the short slim fiducials in the triggered images compared to the slim fiducials. To prevent a detection of positions in errors due to the changing lengths in triggered images, we recommend using short slim fiducials for the ABH system.
The ABH system can simultaneously detect three fiducials' positions. However, only one fiducial setting is determined in the system.
In the clinics, the length of three fiducials in the triggered images was different in each case when slim fiducials were used, as in F I G . 6. Detected vertical positions in the appropriate (a) and inappropriate settings (b) when VISICOIL was used in the plate for Plate_A. The appropriate setting was 5 mm in length in the ABH system. The inappropriate setting was 3 mm in length in the ABH system. There were no setup errors in any direction. The blue dots were of the vertical position in the triggered images detected in the ABH system. Fig. 1, in a report by Vinogradskiy et al. 16 Hence, it is possible that some fiducials were detected normally while others were detected with inappropriate settings. In the latter case, users may remove the fiducials with lengths that do not match with the marker setting in ABH.
Slim fiducials, such as VISICOIL and LumiCoil, may bend in the body. Bended coils for VISICOIL have been observed in clinics. In this study, there was no bended fiducials, except LumiCoil with form 8. Therefore, it was not evaluated whether the ABH system moved normally to the bended fiducials. As detailed in Plate_B in Table 2, the difference between the expected and detected positions in the vertical direction was <0.5 mm in all fiducials except in LumiCoil with form 8 when the setup error occurred in superior direction.
The reason is that slim coils were described as lines or points in the triggered images. The length of these lines and points was vertically <1 mm. However, LumiCoil with form 8 was described as "∞" or "▮" and the length of "∞" and "▮" was vertically 2 mm at maximum.
Hence, LumiCoil with form 8 in Plate_B has larger errors than any fiducials in vertical directions. When fiducials bend in a body, a change in the thickness occurs in the triggered images. Hence, the detected errors become larger. Therefore, it is recommended that fiducials do not bend in a body.
Regarding limitations, this study examined the detectability of fiducial position using only phantom images rather than clinical images. Three fiducials did not overlap with one other during irradiation, and the inside of the phantom did not include inhomogeneous materials in this measurement. In clinics, fiducials inserted in a body may overlap with bones and other organs by the angle of acquiring the triggered images. Additionally, the image acquisition parameters vary depending on the body's thickness, and the optimal parameters for each patient are not clear. 19

| CONCLUSIONS
The detectability of fiducial markers' positions with the ABH system depends on the visibility of the fiducials in the triggered images. To ensure high detectability of the ABH system, the following are recommended. First, fiducial markers should be inserted in the body such that the angles that change the visibility of the fiducials are small in triggered images. Second, short slim or ball fiducial markers should be used.

CONF LICT OF I NTEREST
The authors declare that they have no conflict of interest.