Elsevier

Physica Medica

Volume 39, July 2017, Pages 121-131
Physica Medica

Original paper
3D catheter reconstruction in HDR prostate brachytherapy for pre-treatment verification using a flat panel detector

https://doi.org/10.1016/j.ejmp.2017.06.008Get rights and content

Highlights

  • Demonstration of a method to reconstruct (3D) HDR prostate catheters, for position verification.

  • Performed in the treatment bunker immediately prior to treatment delivery.

  • Provides a method to directly compare and validate catheter positions against the treatment plan.

  • A clinically relevant catheter displacement detection threshold of 2.2 mm was demonstrated.

  • Can be combined with in vivo dosimetry improving confidence in existing error detection regimes.

Abstract

Purpose

High dose rate prostate brachytherapy is a widely-practiced treatment, delivering large conformal doses in relatively few treatment fractions. Inter- and intra-fraction catheter displacements have been reported. Unrecognized displacement can have a significant impact on dosimetry. Knowledge of the implant geometry at the time of treatment is important for ensuring safe and effective treatment. In this work we demonstrate a method to reconstruct the catheter positions pre-treatment, using a ‘shift’ imaging technique, and perform registration with the treatment plan for verification relative to the prostate.

Methods

Two oblique ‘shift’ images were acquired of a phantom containing brachytherapy catheters, representing the patient immediately pre-treatment. Using a back projection approach, the catheter paths were reconstructed in 3D and registered with the planned catheter paths. The robustness of the reconstruction and registration process was investigated as a function of phantom rotation. Catheter displacement detection was performed and compared to known applied displacements.

Results

Reconstruction of the implant geometry in 3D immediately prior to treatment was achieved. A mean reconstruction uncertainty of 0.8 mm was determined for all catheters with a mean registration uncertainty of 0.5 mm. A catheter displacement detection threshold of 2.2 mm was demonstrated. Catheter displacements were all detected to within 0.5 mm of the applied displacements.

Conclusion

This technique is robust and sensitive to assess catheter displacements throughout the implant volume. This approach provides a method to detect, in 3D, changes in catheter positions relative to the prostate. The method has sufficient sensitivity to enable clinically significant decisions immediately prior to treatment delivery.

Introduction

High dose rate (HDR) prostate brachytherapy performed with computed tomography (CT) imaging for treatment planning is a widely practiced approach [1]. The CT image data represents a ‘snap shot’ in time of the implant geometry, but changes may occur by the time of treatment. Oedema [2] and other influences may impact the position of the catheters relative to the prostate anatomy, and so pre-treatment implant verification is recommended [3], [4]. Inter- [5], [6], [7], [8], [9] and Intra- [10], [11] fraction catheter displacements have been reported and occur largely in the cranial-caudal direction.

For CT based treatment planning, an imaging approach at the time of treatment, ideally with the patient in the treatment position, is required to identify any catheter displacement relative to the surrounding anatomy. External visual inspection of the catheters and template is important but may not truly capture the internal catheter-to-anatomy relationship. Unrecognized catheter displacement can have a significant impact on dosimetry, as illustrated by Holly et al. [10], showing the volume of the prostate receiving 100% of the prescription dose (V100) decreases by approximately 20% per centimetre of catheter displacement, highlighting the importance of verification prior to delivery of large fractional doses.

Pre-treatment imaging, immediately prior to treatment delivery, is essential in order for in vivo dosimetry (IVD) systems to yield an output that can be confidently interpreted for error detection because knowledge of the geometry of the implant at the time of treatment is imperative [12], [13]. IVD approaches depend upon accurate knowledge of the location of the detector and the source (catheter positions) for interpretation of the outputs. Whether an IVD discrepancy should be interpreted as a potential treatment delivery error depends upon knowledge of any implant geometry changes between planning and treatment delivery [14]. If the IVD measurement discrepancy is consistent with a known change of catheter position, then the difference may not be due to delivery error.

Generally 3D pre-treatment verification imaging for comparison with the CT or MRI data is challenging to achieve inside the HDR brachytherapy treatment bunker. Approaches to pre-treatment verification imaging are often performed with CT [8], [15], or with a C-arm system [11] in the treatment bunker. While CT imaging provides information to compare with previous fraction image data, it may not reflect the position of the patient at the time of treatment. The implant position may be compromised by moving the patient from CT to the treatment bunker. Catheter displacement evaluation with CT data when performed (between treatment fractions) is usually only determined as a 1D (Superior-Inferior) assessment [7], [8], such as distance between catheter tip and fiducial markers. In-treatment-room imaging is desirable as it truly represents the catheter to anatomy relationship immediately prior to treatment. Acquisition of an anterior-posterior (AP) pelvic X-ray can provide 2D information but it can be difficult to identify all catheters and to account for (implant) catheter tilt and rotation. One study implemented kV-CBCT with a C-arm device, but the 3D catheter evaluation was performed offline [10] after the treatment.

For the purposes of HDR prostate treatment verification, we have previously integrated a flat panel detector (FPD) into the treatment couch to perform treatment delivery source tracking [16]. In an earlier phantom study [17], we characterised the same FPD for 2D pre-treatment imaging (AP image) to confirm catheter positions and to aid in the interpretation of in vivo source tracking outputs relative to the implanted catheters at the time of delivery.

In this work, we extend the imaging capability of the FPD from 2D to 3D demonstrating the use of the FPD, and ceiling mounted X-ray system in the treatment bunker, to capture two oblique AP images of the phantom/patient. These images are then reconstructed to produce a 3D representation of the implanted catheter paths which can be directly compared to the 3D CT treatment planning system (TPS) catheter paths. The use of multiple image projections has been applied to the reconstruction of brachytherapy catheters, applicators [18], [19], [20], [21] and implanted low dose rate seeds [22], [23], [24], [25] for treatment planning purposes, but not for HDR brachytherapy pre-treatment verification, as performed in this proof-of-principle study.

Using a phantom of known geometry, we demonstrate a method to reconstruct the pre-treatment catheter positions in 3 dimensions and perform a registration with the TPS to verify against planned catheter positions. We determine the reconstruction and registration uncertainties and determine a catheter displacement detection threshold relative to the surrogate prostate.

Section snippets

Overview

The pre-treatment verification process (detailed below) consists of two main parts: (i) 3D catheter reconstruction from projection images, followed by (ii) registration between the reconstructed (measured) space and the TPS space, for direct comparison. In this work, we establish the utility of our pre-treatment verification process by demonstrating that the process is robust to set-up changes such as phantom tilt to mimic patient pelvic rotation (Section 2.1.3), and calculating the

3D catheter reconstruction

The geometric quality of the 3-D reconstruction was quantified by the agreement between the known and measured catheter marker positions. The differences for the initial and repeat datasets showed a mean absolute error of 0.8 mm (s.d. 0.4 mm), and featured a 3-D root mean square error (RMSE3-D) of 0.9 mm. This is the radius of the sphere centred at the true position, containing the reconstructed point with a probability of 68%. This was calculated using 13 catheter marker positions, in each of 7

Discussion

We have demonstrated a 3D catheter reconstruction method, which can be applied in the HDR brachytherapy treatment bunker, to perform pre-treatment implant position verification. This novel approach provides a 3D reconstruction of the implant geometry for direct comparison with the TPS data.

Assessment of catheter displacement can be performed in all 3 dimensions throughout the implant volume, and not just at the catheter tip as typically performed with 2D verification approaches. Integration

Conclusion

In this work we have demonstrated a method to reconstruct, in 3 dimensions, the catheter geometry at the treatment couch immediately prior to treatment delivery, and perform a registration for direct comparison with the treatment planning data. The reconstruction, registration and TPS imaging uncertainties were evaluated and the imaging geometry was optimised. A catheter displacement detection threshold (Dt) of 2.2 mm was demonstrated. This approach provides a method to confirm that catheter

Conflict of Interest disclosure

The authors have no COI to report.

Acknowledgements

This research received funding from the Radiation Oncology section of the Australian Government Department of Health and Ageing.

References (29)

Cited by (10)

  • An innovative gynecological HDR brachytherapy applicator system for treatment delivery and real-time verification

    2019, Physica Medica
    Citation Excerpt :

    Moreover, specific corrections for the possible energy and angular dependences of the diodes were not necessary, because the overall diode response was taken as reference and intrinsically modeled with the multi-term Gaussian functions. This is different to systems that are not built-in with the applicator, such as 2D arrays [8,9] or EPIDs [12], where more complex assumptions, pre-calibrations and calculations have to be performed to accurately reconstruct source positions inside the patient. For a correct calibration of the MVC system, it is however very important that it is calibrated in the same conditions that are present during treatment.

  • A Monte Carlo study on the feasibility of real-time in vivo source tracking during ultrasound based HDR prostate brachytherapy treatments

    2019, Physica Medica
    Citation Excerpt :

    Correct identification of catheter positions on these images however can be difficult, and there remains a need for comprehensive pre-treatment and/or in-vivo quality assurance (QA) [10–12]. Source tracking of dwell positions during HDR pBT treatments based on electronic portal imaging devices (EPIDs) have been performed previously [13–16]. However, when used for HDR brachytherapy source tracking, these EPID based devices are required to operate using low frame rates due to poor signal to noise ratios, and when compounded with their inherently slow read out electronics, loss of data for short dwell times can occur [14].

  • An integrated system for clinical treatment verification of HDR prostate brachytherapy combining source tracking with pretreatment imaging

    2018, Brachytherapy
    Citation Excerpt :

    An uncertainty of 2.2 mm represents an upper bound given the geometric dependence on the height of the marker from the imaging plane (hm). Here, we assume a maximum hm of 160 mm based on a study of 33 patients, where the maximum observed distance between a catheter position and the imaging plane was 157 mm (24). ( see Supplementary material for detailed calculations).

  • In vivo dose verification method in catheter based high dose rate brachytherapy

    2017, Physica Medica
    Citation Excerpt :

    Estimated tip-to-tip length of the dosimetry catheter was 122.5 mm; its length filled with dosemeters and spacers during the 6th treatment fraction was 118.0 mm, during the 7th treatment fraction – 118.0 mm. The precise evaluation of catheter lengths and identification of possible catheter movement plays an important role in dosimetry [35]. Comparison of in vivo measured (experimental) and TPS calculated (theoretical) doses at the same positions in the catheter are provided in Figs. 3(a) (6th fraction) and 4(a) (7th fraction).

View all citing articles on Scopus
View full text