Firstly, a throughout literature retrieve was carried out with respect to the keyword “calypso & radiotherapy” via the Web of Science. It was found that only three articles were reported by Chinese institutions. Two of them were talking about the accuracy evaluation(27, 28), and the third was focusing on algorithms comparison for localizing prostate tumor(29). All of them were medical physical researches rather than clinical experiences. Therefore, based on the aforementioned literature search, it can be concluded that we are the first institution to utilize internal tracing system for precise liver radiotherapy in China. The first implement of internal motion monitoring for liver treatment in the world was carried out in March 2015, and the first paper was published in April 2015 by two different groups.
FigureS1 (in Supplementary Information) is the flowchart of implementing the internal tracing system for liver radiotherapy. It can be seen that the main processes are the same as the regular radiotherapy except for beacon implantation, beacon marking, etc. The main differences to regular radiotherapy are that beacons for tracking must be implantation before CT positioning, and then, the beacons are tracked during the treatment. Apart from that, QA and array setup must be done prior to tracking and treatment to make sure the setup offsets distributing in an acceptable range. The QA fixture is a white cube block containing 3 beacon transponders for performing the QA procedure, and the array is a plate planted above the fixture or the patient for tracking the beacons movements inside the patient’s body.
Figure 1 shows the comparison of CT images of the same patient by a free-breath scan and a breath-hold end-exhale scan. It can be seen that the liver has a movement of up to 2 cm which is a big challenge for non-motion managed radiotherapy. Although respiration management techniques such as ABC (Active Breathing Coordinator) and RPM (real-time position management) were utilized for assisting precise radiotherapy, these were all performed indirectly. The internal electromagnetic motion management is working in a direct way by monitoring the real-time signals transmitted by the beacons which are elliptic cylinders with an 8.7 mm length and a 1.3 mm diameter, and were implanted into the patient’s body by 17G needles as shown in Fig. 1C and D.
CT positioning, cancer and organs at risk contouring were the same as regular processes. The slice thickness of CT images should be in the range of 1.0 mm to 1.5 mm to ensure that the beacons would not be lost in adjacent image slices. Prior to planning, beacons were found and marked by the “Calypso Beacon Detection” function in the Contouring Module (Eclipse 13.6, Varian Medical Systems, Inc. USA).
Figure 2 shows the planning and DVH results. It can be seen that the 100% isodose line has a good conformity with regard to the PTV. This can be attributed to that the PTV approximate spherical shape. Table S1 (in Supplementary Information) is a list of the dose statistical results for CTV, PTV, Liver and Spinal Cord. It can be noticed that mean doses for CTV and PTV have reached the prescription. Mean dose for liver and max dose for spinal cord were in a safe range. Besides, it can also be seen that max dose for CTV is lower than that of PTV, indicating that the max dose point located in PTV rather than CTV. As the dose was delivered in a breath-hold end-exhale phase, it can be assumed that it is still safe for the patient.
Before the plan delivery, several procedures should be done in advance. To ensure proper operation of the system, a QA procedure must be performed daily. The QA procedures in this article were done by the QA fixture, as shown in Fig. 3A. Subsequently, distance offsets of the measured position of the QA fixture isocenter from the calibrated isocenter position in the lateral, longitudinal and vertical directions were recorded. Finally, the isocenter offsets were calculated by the following equation:
Where in the equation, x, y and z mean the offsets in the lateral, longitudinal and vertical directions, respectively. The isocenter offsets must be less than or equal to 0.2 cm to ensure the QA procedure to pass. By summarizing the 12 times of QA test results, it was shown that the tracking system has 0.108 ± 0.026 cm, -0.054 ± 0.037 cm and 0.08 ± 0.027 cm offsets in the lateral, longitudinal and vertical directions, respectively, as shown in Table S2 (in Supplementary Information). The isocenter offset, which was determined by the above three offsets, was 0.152 ± 0.024 cm. The QA results indicated that the tracking system was considerably stable during the 12 fractions, which ensured the isocenter stability for the subsequent patient and array setup.
After the patient and array setup were done, the system was detecting the intertransponder distances inside the patient’s body prior to dose delivery. Table S3 (in Supplementary Information) shows the intertransponder distances offsets of the measured positions from the distances in the plan. Comparing with the plan data, distance variances between each two transponders were − 0.056 ± 0.032 cm, 0.017 ± 0.033 cm and − 0.082 ± 0.068 cm, indicating that the transponders variances were very small in the liver and guaranteed the accuracy of target tracking.
Positions of the three transponders were in further determining the geometric residual of the target, the pitch, roll and yaw angles, which were shown in Table S4 (in Supplementary Information), with the variances of 0.048 ± 0.021 cm (threshold 0.2 cm), 2.17°±1.85°(threshold 10°), -2.42°±1.51° (threshold 10°) and 1.67°±1.07° (threshold 10°), respectively. It was indicating that the target was located in the correct positions which was beneficial to the following treatments.
During the tracing procedure, the actual dose delivery time is very short, as shown in the light gray areas of Fig. 4A. The reason is that only the end-exhale phases were used for beam on. Although there was a gating interlock to control the beam on and off, the appropriate range in a respiratory cycle is so short, therefore, the delivery efficiency was reduced. In addition, therapists could see the liver tumor motion in real-time. When the patient was exhaling, curves of three directions were moving towards 0. Meanwhile, therapists told the patient to hold the breath using the talkback system. Therefore, by using the end-exhale phases and observing the tracking curves, doses could be precisely delivered.
Although breath-hold end-exhale could favor high dose delivery and reduce the overall treatment time, it is still a challenge for elder patients. Figure 4B shows the third patient being treated in our hospital by the tracking system on 15th April, 2020, who was male with the age of 83 years old. It is clearly seen that it is very hard for the patient to hold-breath during the beam on cycle, and a big base-line drift could be seen in Fig. 4C. In sake of a more stable motion during breath-hold period, a tracking video signal synchronizing device was designed to transmit the video signals of the internal electromagnetic motion, OSMS and RPM from the control room to the treatment room via a three-channel switch. As is shown in Fig. 4B, the patient could observe his own target motion curves by a mini screen which is connected to a VGA signal cable, and the screen displayed exactly the same motion curves as that of the screens in the control room. Thus, the patient could make the tracking signal more stable during the beam-on phase by adjusting his own breath motion. By using the video signal synchronizing device, it is clearly seen from Fig. 4D that the motion curves are more stable than that in Fig. 4C. In short, the video signal synchronizing device could give to more stable breath-hold phases, accurate treatment positions of the target and less treatment durations for each fraction.
For the sake of investigating how much the tracking system could affect the treatment time, delivery time for each field and treatment duration for each fraction were recorded. The delivery time of the five fields were 13.8 s, 13.1 s, 11.18 s, 11.57 s, 11.62 s with the average value of 12.25 ± 1.13 s (Fig. S2A, Supplementary Information). Therefore, the patient was keeping breath-hold for just five short time and was easily adapt to the treatment style in each fraction. It is also noted that base lines in the short delivery time almost have no drifts (Fig. 4A). Treatment durations of each fraction were also recorded from the beginning to the end of the tracking (Fig. S2B, Supplementary Information). The durations were ranging from 6.22 minutes to 21.43 minutes, with the average value of 11.25 ± 5.03 minutes. Due to CBCT scan, image registration and couch shifts, durations of the last two fraction were reach up to about 20 minutes. Even though, the tracking system did not notably lengthen the treatment time compared to the one without the system in our hospital. After the radiotherapy completed, the patient only showed I°myelotoxicity and became better with symptomatic treatment before discharging from the hospital.