Puncture and Dose Accuracies of Navigation-Assisted 3D-Printing Templates Combined with Computed Tomography Guided Radioactive Iodine-125 Seed Implantation in the Treatment of Malignant Tumors

Objective: To preliminarily verify the accuracy of navigation-assisted seed implantation by comparing preoperative and actual differences in puncture characteristics and dosimetry in computed tomography-guided, navigation-assisted radioactive iodine-125 seed implantation using 3D-printed templates for the treatment of malignant tumors. Methods: A total of 27 tumor patients who were treated with seed implantation under combination guidance in our hospital between December 2018 to December 2019 were enrolled in this study. Navigation needles (n=1–3) were placed in each patient to obtain preoperative and intraoperative puncture information, including angle, depth, insertion point, and tip position; we also investigated the dosimetry parameters in the preoperative and postoperative plans, including D 90 , V 100 , V 150 , V 200 , minimum peripheral dose (MPD), conformal index, external index, and homogeneity index of the target area. The t-tests and nonparametric correlation tests were used for analysis (P<0.05 was considered signicant). Results: The means errors of the angle, depth, insertion point, and tip position were 0.47 ± 0.521°, 0.35 ± 0.238 cm, 1.7 ± 0.99 mm, and 3.1 ± 1.75 mm, respectively. There were no signicant differences between the intraoperative and preoperative angles (P = 0.271), but there was a signicant difference in depth (P = 0.002). Errors of the angle, depth, and insertion point were larger for the pelvic/retroperitoneal area than for the head and neck/chest wall (P < 0.05). With the exception of MPD, there was no signicant difference in dosimetry indices between the postoperative and preoperative plans (P > 0.05). The MPD in the postoperative plan was higher than that in the preoperative plan (mean: 72.1 Gy and 63.8 Gy, respectively; P < 0.05). Conclusion: Seed implantation under combination guidance showed good accuracy, and the actual intraoperative puncture information and postoperative doses were in good agreement with those in the preoperative plan, thereby demonstrating promising


Introduction
Radioactive iodine-125 seed implantation (RISI) is being increasing used for the local treatment of tumors [1]. However, as it involves a puncture procedure, which largely depends on the physician's personal clinical experience and level of expertise, the quality of the procedure and treatment effect vary, limiting the further promotion of RISI for clinical use. Three-dimensional-printing template (3DPT) technology is a milestone in the advancement of RISI therapy. With 3DPT guidance, accurate control of the seed needle has been realized, greatly improving adherence to the preoperative plan [2]. Recently, some studies have used an image navigation system (INS) to assist in the treatment of head and neck tumors using seed implantation, achieving good puncture accuracy [3].
This study applied INS and 3DPT to RISI treatment for tumors in various parts of the body to evaluate the puncture and dose accuracies under combination guidance and provide a theoretical basis and data support for the rational selection and optimization of implantation protocols.

Baseline clinical data
A total of 27 patients with malignant tumors who were treated with INS-assisted, 3DPT-guided, computed tomography (CT)-guided RISI in our department between December 2018 and December 2019 were enrolled in this study. All patients had complete imaging data, preoperative/postoperative plans, and intraoperative data. All treated patients met the indication criteria for radioactive seed implantation [1]: (1) patients who showed relapse after surgery or external radiotherapy or refused surgery or external radiotherapy with a tumor diameter ≤ 7 cm; (2) the pathological diagnosis was clear; (3) there was a suitable puncture path; (4) there was no bleeding tendency or hypercoagulable state; (5) the general physical condition was good (Karnofsky performance status score > 70), radioactive seed implantation could be tolerated; and (6) the expected survival was > 3 months. All patients signed an informed consent form prior to treatment. Table 1 shows the baseline information of the patients, target lesions, and the preoperative plan.
The research protocol was approved by the Ethics Committee.

Preoperative plan
We performed 3DPT-guided RISI in accordance with published profession standards [6]. INS assistance was applied based on the following aspects: (1) Positioning and preoperative plan design: CT, with a slice thickness of 2.5 mm, was performed 2 days before operation. The posture (supine/prone/lateral) was selected according to the location of the lesion; the vacuum pad was xed, and the body surface was marked with a pendulum line. CT data were transmitted to the BTPS for developing a preoperative plan design. The tumor in the target area (gross tumor volume, GTV) and the organs at risk (OAR) within 2 cm around the target area were outlined to set the prescribed dose and seed activity; identify the direction, distribution, and depth of the seed needle; and set up 1-3 navigation needles. The number of seeds was calculated, and the spatial distribution of the seeds was simulated. Through BTPS optimization, GTV D 90 (when GTV receives 90% of the dose) was adjusted to meet the prescribed dose setting requirements.
(2) Design and production of 3DPT: Personalized digital modeling of the treatment area template in the BTPS and additional information on the alignment axis and puncture characteristics were used to set the printing range of the template. A 3DPT was produced using a 3D photocuring rapid-prototyping machine and medical photocuring resin material. The 3DPT included body surface information, alignment marks, and puncture information with respect to the treatment area of the patients. After insertion of the navigation needle was completed, CT was performed to con rm that the position of the needle was accurate, and ne-tuning was conducted when there was an error. After the navigation needles were correctly calibrated, the 3DPT position was considered accurate.
(4) Puncture and seed implantation: After the 3DPT was accurately positioned, the rest of the seed needles were percutaneously punctured to the predetermined depth through a template guide hole. During the puncture process, CT was performed to monitor the needle insertion path, and the needle was ne-tuned, if necessary, to avoid damage to the OAR. After the needle was in place, the seeds were implanted according to the preoperative plan, and CT was performed to con rm the seed distribution. As there may be differences between the actual intraoperative situation and the preoperative plan, intraoperative optimization was carried out if necessary, and the needle insertion path and seed distribution were adjusted in real time to ensure that the dose received by the GTV met the requirements of the preoperative plan.
(5) Postoperative dose evaluation: CT was performed postoperatively, the image was transmitted to the BTPS for dose veri cation, and the actual doses delivered to the GTV and OAR were evaluated. The technical process is shown in Fig. 2. The postoperative dose quality evaluation complied with the quality evaluation standard de ned by the BC (British Columbia) Cancer Research Center [7].
According to the immediate veri cation of the target area D 90 and V 100 (volume when the GTV receives 100% of the prescribed dose), the evaluation results were characterized as excellent, good, fair, and poor.

Collection and comparison of treatment parameters
(1) Puncture information: In the BTPS, the images taken after the intraoperative navigation needle was in place were fused with the preoperative plan images, and rigid registration for the bone was performed. In the fusion image, both the virtual and actual puncture characteristics with respect to the preoperative plan were displayed. The angle and depth of the puncture before and during the operation were compared, and the absolute value of the difference was taken as the error value. Meanwhile, the straight-line distances between the two puncture points on the body surface and between the two tip positions were recorded. the target area, the volume of the target area receiving the prescribed dose, and the total volume (cm 3 ) contained in the prescribed dose, respectively. The optimal CI was 1, which indicated that the prescribed dose covered the target area but the received dose outside the target area was lower than the prescribed dose. A larger CI indicated that the volume inside the target area receiving the prescribed dose was larger, whereas the volume outside the target area receiving the prescribed dose was smaller. In terms of the external index (EI) [9], EI = (V ref -V T,ref )/V T × 100%. The optimal EI was 0, which suggested that the dose tissues outside the target area received less than the prescribed dose. A higher EI implied that the volume of the prescribed dose received outside the target area was larger. Regarding the homogeneity index (HI) [9], the target area receiving 150% of the prescribed dose. The ideal optimal HI was 100%. A higher HI suggested a more uniform dose distribution in the target area. As the location of the lesions was scattered and the adjacent OAR varied, this study was not designed to compare OAR doses.
(3) Other treatment information: In the BTPS, other preoperative and postoperative treatment parameters were collected and compared, including GTV and the numbers of needles and seeds.

Statistical methods
For comparisons between groups, a Shapiro-Wilk test was rst used to verify whether the data in each group were normally distributed. A P value > 0.05 indicated that the data conformed to a normal distribution. For normally distributed data, a t-test (including paired and independent sample t-tests) was used for comparison, and the t-value was used to describe the test value. For non-normally distributed data, a nonparametric test was adopted for comparison (Wilcoxon test for correlated samples and Mann-Whitney U test for independent samples), and the z-value was used to describe the test value. P-values ≤ 0.05 were considered as statistically signi cant, and the statistical software used was SPSS 25 (IBM Corporation).

Results
Navigation-assisted 3DPT combined with CT-guided RISI can be successfully completed according to the established technical procedures. A total of 52 navigation needles were used in 27 patients (mean: 2 ± 1). The preoperative and intraoperative needle angles ranged between 72.5-111.1° and 73.9-109.9°, respectively, and the absolute value of the difference between the two groups was 0-1.7°. The preoperative and intraoperative needle depths ranged between 2-11 cm and 2-11.8 cm, respectively, and the absolute value of the difference between the two groups was 0-0.89 cm. Table 2 lists the results of the comparison between the two groups. In addition, the range, median, and mean value of error of the needle insertion point for the two groups were 0-3.9 mm, 1.3 mm, and 1.7 ± 0.99 mm, respectively, and those of the needle tip were 0-0.88 mm, 2.9 mm, and 3.1 ± 1.75 mm, respectively. The rang and mean value of the needle depth in the head & neck and chest wall were 2-6.6 cm and 3.4 ± 1.08 cm, respectively, and those in the retroperitoneal & paravertebral and pelvic part were 4.3-11.8 cm and 5.8 ± 1.71 cm, respectively. The results of a Mann-Whitney U test showed that there was signi cant difference in needle depth between the two groups (P = 0.000). If divided into two groups by position, the error of the retroperitoneal & paravertebral and pelvic cavity were signi cantly larger than those of the head & neck and chest wall (Table 3), and the differences in angle, depth, and insertion point were statistically signi cant (P < 0.05).    another solution for puncture intervention technology. This technology fuses and registers CT, magnetic resonance, or positron emission tomography images obtained before the operation with intraoperative real-time images. During the operation, the position of the puncture needle is tracked in real time by a tracer (optical or magnetic positioning) to locate the lesion on the fusion image, thereby guiding the operator to perform the puncture [11]. Owing to its real-time, dynamic, and visible characteristics, INS is widely used in biopsy, ablation, and other puncture-related operations [12][13][14]. However, the combination of the two techniques for RISI and its application in the treatment of tumors in many parts of the body have rarely been reported in this country and abroad. The current data showed that the accuracy of INS-assisted 3DPT-guided RISI treatment was good.
According to different positioning principles, INS mainly includes magnetic and optical positioning and navigation [15]: (1) Magnetic positioning and navigation are chie y to x the sensor coil on the tracked device (such as the puncture device). When the sensor coil moves relative to the magnetic eld emitter, it produces different intensity currents, and then locates the tracked device through the current signal. (2) Optical positioning and navigation are primarily used to x the tracer on the CT machine and tracked device, and the infrared camera is used to directly detect the position of the tracer. Compared with magnetic positioning, optical positioning is faster in data transmission and has higher accuracy. The disadvantage is that there can be no obstruction between the optical camera and tracked device, and the position of the needle tip cannot be tracked. Optical positioning and navigation were adopted in this study, but the optical positioning and navigation tracer was large and occupied a large space after being xed to the seed needle. Furthermore, RISI is a multi-needle operation, so it is impossible to equip all seed needles with tracers. Therefore, 1-3 navigation needles were set for each case in this study. By tracking, guiding, and controlling the navigation needles, we accurately set the 3DPT position, reduced the number of times of adjustment of 3DPT position and CT con rmation, and preliminarily obtained the accuracy data of INS with 3DPT-guided puncture, which could be used as a reference for follow-up multi-needle guidance.
Based on the data in this study, the needle accuracy under INS with 3DPT was good, and the mean errors of the angle, depth, insertion point, and tip were 0.47 ± 0.521°, 0.35 ± 0.238 cm, 1.7 ± 0.99 mm, and 3.1 ± 1.75 mm, respectively. Thus, the mean error of the angle was < 1°, and the mean error of distance was approximately 3 mm. Compared with the previous studies on 3DPT and/or INS assisted brachytherapy (the average tip error was 0.86-7 mm, and the average angle error was 1.9-5.6°), the accuracy of our study was similar or even better [3,[16][17][18]. Good accuracy contributes to reducing the times of CT scan con rmation and shortening the operation time. For angle comparison, the preoperative and intraoperative angles were not statistically different (P = 0.271), but the intraoperative depth was slightly greater than the preoperative depth (4.4 ± 1.81 cm and 4.3 ± 1.78 cm, respectively; P = 0.002).
The reasons for the errors were considered to be changes in skin contour and thickness caused by anesthesia, template alignment issues, and an unsatisfactory t between the template and body surface. In terms of the treatment site, angle, depth, and insertion point errors in the pelvic/retroperitoneal region were larger than those in the head and neck/chest wall (P < 0.05). In addition to the abovementioned reasons, these errors were possibly attributed to the fact that longer insertion paths led to greater errors (the difference in insertion depth between the two groups was statistically signi cant, P < 0.05). Moreover, the stability of tissues and organs in the head and neck/chest wall treatment area was better than that in the pelvic/retroperitoneal region. This prompted us to focus on the accuracy of needle insertion and perform multiple CT scans to con rm if it was necessary in areas where the treatment location was deep and the tissue was soft and prone to change.
By comparing the preoperative and postoperative dosimetry parameters in this study, we revealed that most of them were not signi cantly different, suggesting that the postoperative dose could better meet the requirements of the preoperative plan. The only dosimetry index with a signi cant difference was MPD, which was alternatively expressed as D 100 (when GTV receives 100% of the dose). MPD in the postoperative plan was higher (P < 0.05), indicating that the error in this study did not result in a reduction in dose to the target area. The reason may be that we replanted seeds in the patients whose immediate postoperative CT scans showed unsatisfactory seed distribution. This also explains why the number of seeds in the postoperative plan was higher than that in the preoperative plan, and the difference was statistically signi cant (P = 0.031). Moreover, the GTV in the postoperative plan was larger than that in the preoperative plan (P = 0.005), which was possibly related to intraoperative and postoperative bleeding, edema, and in ammatory reactions. Immediate postoperative CT allows for real-time observation of seed distribution. For patients with poor seed distribution, the timely replanting of seeds could effectively prevent a cold spot of the dose in the target area [1]. The postoperative D 90 of the three cases with fair and poor evaluation quality did not reach 90% and 80% of the prescribed dose, respectively, suggesting that RISI is an operation-dependent treatment, and there remain a few cases that have di culty in meeting expectations in the actual operation. It is believed that the accuracy of treatment will be improved with further pro ciency and experience with the procedure. Many studies have analyzed the dose accuracy of RISI guided by 3DPT, the results indicate that the dose accuracy was good [2,10,19,20], however, there was no analysis of needle path error in each study. Whether INS combined with 3DPT guidance is really better than 3DPT guidance alone still needs to be further studied.
The limitations of this study are as follows: (1) the technical development time was slightly short, and the sample size was small; (2) it was limited to comparisons between treatment plans, and further observation and follow-up is needed for subsequent clinical effects; (3) due to the small number of cases and scattered treatment sites, effective comparison of the dose distribution to endangered organs in different parts was not available; and (4) the combination and process of INS with 3DPT remain in the initial stages and need to be further explored and improved. In a follow-up study, we will further accumulate cases and conduct in-depth and detailed research.

Conclusion
The combination of INS and 3DPT technologies for RISI treatment showed good accuracy and feasibility and exhibited good quality with regard to completion of the implantation plan. Errors in the actual needle insertion during the operation were smaller than that before the operation. The actual D 90 in the target area, CI, and dose uniformity all met the design requirements of the preoperative plan. In a follow-up study, we will expand the number of cases, re ne the study, and clarify the e cacy and safety of this approach based on clinical data.   Flowchart of INS-assisted 3DPT-guided RISI. a. Preoperative plan (navigation needle setup); b. Navigation-guided insertion of the navigation needle; c. Seed needles inserted when the navigation needle was in place; d. Preoperative images fused with the intraoperative images to measure the errors (angle, insertion point, and tip) of the navigation needle; e. Preoperative plan; f. 3DPT

List Of Abbreviations
and CT guided seed needles insertion; g. Seed implantation; e. Dose veri cation.