- Email: [email protected]
Combining transrectal ultrasound and CT for image-guided adaptive brachytherapy of cervical cancer: Proof of concept
Brachytherapy
-
(2016)
-
Technical Note
Combining transrectal ultrasound and CT for image-guided adaptive brachytherapy of cervical cancer: Proof of concept Nicole Nesvacil1,2,*, Maximilian P. Schmid1, Richard P€ otter1,2, Gernot Kronreif3, Christian Kirisits1,2 2
1 Department of Radiation Oncology, Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Medical University of Vienna, Austria 3 Austrian Center for Medical Innovation and Technology, Wr. Neustadt, Austria
ABSTRACT
PURPOSE: To investigate the feasibility of a treatment planning workflow for three-dimensional image-guided cervix cancer brachytherapy, combining volumetric transrectal ultrasound (TRUS) for target definition with CT for dose optimization to organs at risk (OARs), for settings with no access to MRI. METHODS AND MATERIALS: A workflow for TRUS/CT-based volumetric treatment planning was developed, based on a customized system including ultrasound probe, stepper unit, and software for image volume acquisition. A full TRUS/CT-based workflow was simulated in a clinical case and compared with MR- or CT-only delineation. High-risk clinical target volume was delineated on TRUS, and OARs were delineated on CT. Manually defined tandem/ring applicator positions on TRUS and CT were used as a reference for rigid registration of the image volumes. Treatment plan optimization for TRUS target and CT organ volumes was performed and compared to MRI and CT target contours. RESULTS: TRUS/CT-based contouring, applicator reconstruction, image fusion, and treatment planning were feasible, and the full workflow could be successfully demonstrated. The TRUS/ CT plan fulfilled all clinical planning aims. Dose-volume histogram evaluation of the TRUS/CToptimized plan (high-risk clinical target volume D90, OARs D2cm3 for) on different image modalities showed good agreement between dose values reported for TRUS/CT and MRI-only reference contours and large deviations for CT-only target parameters. CONCLUSIONS: A TRUS/CT-based workflow for full three-dimensional image-guided cervix brachytherapy treatment planning seems feasible and may be clinically comparable to MRIbased treatment planning. Further development to solve challenges with applicator definition in the TRUS volume is required before systematic applicability of this workflow. Ó 2016 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.
Keywords:
Cervix cancer; Brachytherapy; Image fusion; Transrectal ultrasound
Introduction Received 27 May 2016; received in revised form 16 August 2016; accepted 25 August 2016. Financial disclosure: The Department of Radiotherapy at the Medical University of Vienna receives/received financial and/or equipment support for research and educational purposes from Nucletron, an Elekta company. This work has been supported by ACMITdAustrian Center for Medical Innovation and Technology, which is funded within the scope of the COMETdCompetence Centers for Excellent Technologies program of the Austrian Government. The project was funded in part via a research contract from Elekta. * Corresponding author. Department of Radiation Oncology, Comprehensive Cancer Center, Medical University of Vienna, General Hospital of Vienna, W€ahringer G€urtel 18-20, A-1090 Vienna, Austria. Tel.: þ431404002692; fax: þ431404002693. E-mail address: [email protected] (N. Nesvacil).
There is increasing evidence on the benefits of performing transrectal ultrasound (TRUS) for the assessment of cervical cancer (1e3). TRUS could also be a valuable imaging modality for image-guided adaptive brachytherapy (IGABT) of cervical cancer, not only for real-time image guidance of applicator insertion, but also for target volume identification and assessment of dosimetric coverage. In particular, TRUS could become an important additional imaging modality, if MRI with applicator in situ is not available for treatment planning. However, presently, there are no clear concepts how to integrate TRUS into IGABT. Although TRUS can be used as sole imaging technique for prostate brachytherapy
1538-4721/$ - see front matter Ó 2016 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.brachy.2016.08.009
2
N. Nesvacil et al. / Brachytherapy
(BT), there are major limitations for TRUS-based cervical cancer BT. The relatively small field of view of TRUS and the anatomical changes caused by the insertion of the TRUS probe might pose a substantial challenge for the full evaluation of surrounding organs at risk (OARs). However, the combination with another three-dimensional (3D) volumetric imaging method might overcome those limitations. CT is widely used for 3D image-guided BT as it provides excellent depiction of BT applicators and OARs. However, the lower soft tissue contrast in comparison with MRI poses a challenge for target definition on CT images (4, 5). Using a combination of TRUS and CT may benefit IGABT treatment planning as the advantages of each modality may compensate for their respective disadvantages. Such a method could be beneficial, especially in settings where access to MRI is limited at the time of BT. In this article, we describe a TRUS/CT-based workflow that we tested in our institution and compared to MRI onlye and CT onlyebased treatment planning.
Methods and materials Patient For this study, a TRUS/CT-based workflow was simulated for a patient treated at our institution with MRIbased IGABT. The patient was 52 years old and had biopsy-confirmed squamous cell carcinoma of the uterine cervix staged as International Federation of Gynaecology and Obstetrics (FIGO) IIB. High-dose-rate IGABT was performed after radiochemotherapy according to our previously described clinical protocol (6). For BT, a 60 mm/60 tandem, a 34-mm Vienna ring (Elekta, Sweden) and four interstitial titanium needles were used in both applications. The tandem ring applicator and the interstitial needles were inserted under transabdominal and TRUS guidance and were fixated by vaginal packing. Treatment planning was based on MR imaging. In addition, TRUS and CT were performed. TRUS, CT, and 1.5 T MR images of this patient were part of a previously performed ultrasound assessment study (2), and the study was approved by the institution’s ethics committee.
-
(2016)
-
during a rotation (longitudinal image acquisition) of the TRUS probe in the stepper unit, with the patient in lithotomy position. Volume acquisition was done with a fan angle of 90 , 7 cm depth, and 7 cm length for longitudinal scans. The applicator was fixated to the stepper during volume acquisition by means of a small holding arm (Fisso, Switzerland) and dedicated connectors. CT was performed with a Somatom Plus S scanner (Siemens, Germany) in 2-mm slice intervals, and T2weighted MRI was performed with a 1.5 T system (Avanto, Siemens, Germany) in axial, para-axial, sagittal, and coronal image orientation with 5-mm slice thickness. Based on the GEC-ESTRO recommendations for MRIbased contouring (7), an adapted contouring protocol was used for TRUS delineation. The CTVHR was defined on TRUS by gray-scale imaging as the complete hypoechogenic (dark) cervix surrounded by the bright parametrial tissue. Contouring was performed on TRUSBT, on reconstructed transversal planes acquired by the longitudinal array (Fig. 1a), taking into consideration also TRUSpreBT. CTVHR contouring was performed on MRI for the clinical planning, as well as on CT for interimage-modality comparison in this study, in accordance to respective guidelines and recommendations (5, 7). Three-dimensional measurements of CTVHR maximum width, thickness, and height were recorded, according to the protocol described in (2), on CT, TRUS, and MRI. Due to the previously reported difficult assessment of CTVHR height on TRUS and CT (6), an MRI at diagnosis was used to support the definition of the CTVHR height. In comparison to the MRI-based CTVHR concept, the target height on CT and TRUS at the time of BT would therefore be overestimated. Taking the height of the initial target into account for target definition on other modalities was chosen to avoid geometrical misses at the time of BT. OARs (rectum, colon sigmoideum, small bowel, and urinary bladder) were contoured on MRI and CT. Contouring of all data sets was done blinded, by one observer. All contouring, applicator reconstruction, and treatment planning presented in this study were done with the Oncentra GYN treatment planning system (v 1.2.3, Nucletron, the Netherlands). Applicator reconstruction on TRUS
Imaging and contouring on TRUS TRUS was performed before (TRUSpreBT) and after insertion of the applicator (TRUSBT) with a 5e10 MHz Integrated US transducer (BiopSee, MedCom, Germany), which consists of a biplane probe (BIPC6.5/10/128, BIPL7.5/70/128), specially designed and optimized for transrectal usage. The biplanar TRUS probe was inserted transrectally and fixated with a customized ultrasound stepper unit. Continuous 3D image acquisition was performed by a dedicated registration software (based on an adapted version of Oncentra Prostate, Elekta, Sweden) (1) during a manual pullback (transversal image acquisition) and (2)
The TRUS-based applicator reconstruction was done using a 3D library model of the CT/MR compatible plastic Vienna tandem/ring (Elekta, Sweden). On transversally and/or sagittally reconstructed views, the center of the ring and the position of the tandem near the cranial edge of the TRUS volume were defined as a starting point to place the 3D applicator model. Sagittal and coronal views were used to rotate the 60 tandem ring applicator until it was correctly aligned with the tandem and vagina axis, as seen in the images. On transversal images, the visible needle positions could also be used for rotation fine tuning.
N. Nesvacil et al. / Brachytherapy
-
(2016)
-
3
Fig. 1. Depiction and reconstruction of BT implant on TRUS images: tandem (T), ring (R), and Ti needles (N) in transversal (left), sagittal (middle), and coronal (right) views. (aec): applicator depiction in TRUS images. Dotted lines indicate slice positions for all images in the same row. (def): TRUS image and reconstructed 3D applicator model. (gei): Depiction of needles in TRUS images. Note that window levels have been adjusted in the TPS, to increase visibility of applicator and enable direct reconstruction. BT 5 brachytherapy; TPS 5 treatment planning system; TRUS 5 transrectal ultrasound; 3D, three dimensional.
TRUS/CT image fusion and treatment planning For treatment planning, a combination of the TRUS image volume and the CT image volume was obtained by rigid image registration based on the applicator position defined on the TPS. The applicator was reconstructed on CT images using the same 3D library model described in the previous section. As described in Nesvacil et al. (8), Oncentra GYN can perform an automatic image registration based on the defined applicator position as a common reference coordinate system. Because target and applicator are fixed to each other by the vaginal packing, no relative movement between them occurs between TRUS and CT image acquisition. The stability of these implants has been reported previously by Lang et al. (9). After TRUS/CT image registration, the delineated TRUS target contour was transferred to the CT image volume and the OARs (bladder, rectum, sigmoid) were delineated on CT. The interstitial needles were reconstructed
as depicted on CT. Treatment planning was simulated according to our clinical protocol for four HDR fractions using the following planning aims (EBRT þ IGABT): CTVHR D90 of $85 Gy EQD2 (a/b 5 10 Gy), as well as total OAR D2cm3 of #90 Gy for bladder, #70 Gy for rectum, and #70 Gy for sigmoid, in EQD2a/b53 Gy. To evaluate the potential for the TRUS/CT treatment planning method to obtain similar treatment plan quality as obtained by the use of standard delineation techniques, the TRUS/CT-based dose distribution was evaluated on the MRI and CT data sets.
Results Contouring, applicator reconstruction, image fusion, and TRUS/CT-based treatment planning were performed for this patient. The feasibility of the full described workflow
4
N. Nesvacil et al. / Brachytherapy
-
(2016)
-
Fig. 2. Illustration of the first full TRUS/CT planning workflow by applicator-based registration with a customized TPS. (1) TRUS volume acquisition and target delineation (red line) in relation to applicator position. (2) US and CT image registration via applicator and US target transfer to CT. (3) OAR (bladder, rectum, and sigmoid) contouring on CT and dose planning with the full US/CT structure set. For comparison, three targets are shown: US based (red), MR based (green), and CT based (blue). The dashed orange isodose line corresponds to total dose of 85 Gy EQD2a/b510Gy of the US/CT-optimized treatment plan, assuming a full treatment of 45 Gy EBRT þ 4 HDR BT. Transversal view (left column) and sagittal view (right column) are shown. The correlation of transversal and sagittal views in (1) and (3) is indicated by a dashed line. BT 5 brachytherapy; EBRT 5 external beam radiotherapy; HDR 5 high-doserate; OAR 5 organs at risk; TRUS 5 transrectal ultrasound; US 5 ultrasound. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
N. Nesvacil et al. / Brachytherapy
was simulated, and individual steps of the procedure, performed with our prototype system, were evaluated. The results are illustrated in Figs. 1 and 2. Image acquisition During volume acquisition by pullback of the TRUS probe, a distortion of the visualized 3D volume was observed, which could be attributed to small movements of the ring applicator. Therefore, the tandem and ring were no longer aligned perpendicularly in the transversal US image series, which rendered it unsuitable for treatment planning. When images were acquired with a rotation of the TRUS probe using the longitudinal array, no applicator movement was observed by the operators or in the acquired image volume. Therefore, the image volume acquired with the longitudinal transducer was used for treatment planning. Applicator reconstruction Other than on CT or MRI only parts of the intracavitary applicator could be identified on TRUS. The posterior part of the ring was well depicted in the reconstructed transversal, sagittal, and particularly in coronal views (Figs. 1ae1c). The tandem was very well visible on reconstructed transversal and coronal views, and excellent depiction was given in the sagittal plane (Fig. 1b). Therefore, these parts of the applicator were simultaneously observed in all three reconstructed image views during fine tuning of the reconstruction (Figs. 1de1f). By scrolling through the images in all available orientations, placement of the applicator was verified and fine tuned until a plausible applicator position was established for all views. The anterior part of the ring was hardly distinguishable from the posterior bladder wall, due to artifacts from the applicator. Similar to the tandem, the interstitial titanium needles were identified in all reconstructed image orientations (Figs. 1ge1i). Despite the overall limitations of applicator depiction in TRUS images, for this patient, the achieved applicator reconstruction was found acceptable for applicator-based TRUS/CT registration and further investigation of the treatment planning workflow. Contouring The CTVHR volume was 30 cm3 on TRUS, 31 cm3 on MRI, and 59 cm3 on CT. The CTVHR width was 53 mm, 49 mm, and 71 mm, and the thickness was 34 mm, 33 mm, and 52 mm, respectively, whereas the CTVHR height was within 10% for all target volumes. All CTVs were transferred to the CT data set and are shown in Fig. 2, panel 3. The TRUS and MRI contours differed mainly in the ventral part near the posterior bladder wall, which corresponds to the area where it is most difficult to distinguish the target volume from the bladder on TRUS images.
-
(2016)
-
5
Differences in bladder filling might also contribute to this finding. On CT, the extent of parametrial involvement indicated by the contrast enhancement of the tumor was overestimated on both sides and posteriorly in direction of the mesorectal fascia, whereas with TRUS, the involvement could be defined similar to MRI. Due to limits of overall probe length, rectum length, and the location of the transversal array distal to the longitudinal array, the cranial border of the target was not clearly depicted on longitudinal TRUS images. Therefore, on TRUS as well as on CT, the height of the target on pretreatment MRI defined the cranial border of the delineated contour. Treatment planning An overview of the treatment planning workflow is shown in Fig. 2. The TRUS/CT plan fulfilled all planning aims (target: 92 Gy, bladder: 85 Gy, rectum: 64 Gy, sigmoid: 66 Gy [EQD2a/b53Gy]). The evaluation of the TRUS/CT treatment plan on the CT-only, the TRUS/CT, and the MRIref contour set resulted in a CTVHR D90 of 69 Gy, 92 Gy, and 89 Gy (EQD2a/b510Gy), respectively. For OARs, the total D2cm3 of all structure sets were within 3% for rectum and bladder and 5% for sigmoid (Table 1). Based on the evaluation of the TRUS/CT dose distribution on MRI structures, the plan would have fulfilled clinical acceptance criteria.
Discussion In this study, we analyzed the possibility of a combined TRUS/CT treatment planning approach for cervix cancer BT. The complete workflow, from image acquisition, contouring, and applicator reconstruction to image fusion and treatment planning, could be demonstrated. The huge potential of ultrasound (1, 10) and especially TRUS (2, 3) has been previously demonstrated. A 3D volumetric image acquisition including the target volume, the relevant OARs, and the applicator has not been described so far. Due to the limited field of view of 3D volumetric TRUS, full evaluation of OARs and the whole applicator is not possible; therefore, combination with an additional CT image volume seems meaningful. Applicator reconstruction was Table 1 Evaluation of the treatment plan optimized for TRUS CTVHR and CT OARs, for three contour sets: TRUS/CT, MRI only, and CT only Evaluated parameter
TRUS/CT contours
MRI contours
CT contours
CTVHR D90 (Gy) Bladder D2cm3 (Gy) Rectum D2cm3 (Gy) Sigmoid D2cm3 (Gy)
92.3 85.2 63.5 66.1
88.8 84.0 63.7 62.9
69.0 85.2 63.5 66.1
CTVHR 5 high-risk clinical target volume; OAR 5 organs at risk; TRUS 5 transrectal ultrasound.
6
N. Nesvacil et al. / Brachytherapy
performed directly on TRUS images, by adapting existing guidelines (11). First results on the potential of TRUS for target depiction in cervix cancer BT have been published previously by Schmid et al. (2, 3). In comparison, TRUS image examples shown in the present study were optimized for simultaneously depicting target and applicator. The latter is necessary to perform a direct applicator reconstruction, as is done routinely for CT- and MRI-based treatment planning. However, even then applicator visibility is limited by artifacts in the anterior part of the ring or the TRUS volume being too short to contain the tandem tip. Alternative solutions for applicator placement in the TRUS volume might help to overcome some of these limitations and allow to improve target depiction using an optimized imaging protocol. For the investigated clinical case, the combination of CT for OAR depiction and TRUS for target definition resulted in a plan and dose reporting almost similar to a full MRI approach. In this specific case, the location and orientation of the applicator could be directly reconstructed with a similar accuracy as typically achieved on MRI. For other cases observed during the system development (unpublished results), substantial artifacts showed that this direct applicator reconstruction would not always be possible on the ultrasound images. This study shows that a workflow for TRUS/CT-based treatment planning is, in principle, feasible in a clinical setting, if customized software and hardware for volumetric image acquisition, DICOM export to TPS, and applicatorbased image registration are available. Further developments of hardware, software, and operation procedures, that is, optimization of machine settings, training of users, and systematic clinical testing, are required to increase image quality and precision of target delineation and applicator depiction. Future studies applying a systematic treatment planning workflow are needed to standardize individual steps in the procedure. Based on the proposed workflow, followup monocentric and multicentric studies are needed to study its performance for different stages of disease and the interobserver and intraobserver variability, to develop and validate a reproducible target contouring concept. Important aspects such as precision of applicator reconstruction and image registration will have to be further investigated. Similar to the successful development of CT- and MRI-based gynecological BT, joint efforts are needed to fully exploit the potential of TRUS/CT-based treatment planning. The most critical technological element for a systematically applicable procedure remains the accurate reconstruction
-
(2016)
-
of the applicator in the ultrasound images, which is the basis for appropriate image registration. If it will be possible to establish applicator tracking during the ultrasound acquisition, this TRUS/CT-based workflow may become available as a full 3D volumetric treatment planning approach. BT applicator tracking using optical, electromagnetic, or other approaches is being investigated by several research groups. The combination of these methods with the proposed workflow may result in a feasible technical solution. References [1] Epstein E, Testa A, Gaurilcikas A, et al. Early-stage cervical cancer: tumor delineation by magnetic resonance imaging and ultrasoundda European multicenter trial. Gynecol Oncol 2013;128:449e453. [2] Schmid MP, Nesvacil N, P€otter R, et al. Transrectal ultrasound for image-guided adaptive brachytherapy in cervix cancerdan alternative to MRI for target definition? Radiother Oncol 2016;8140: 00052e00059. [3] Schmid MP, P€otter R, Brader P, et al. Feasibility of transrectal ultrasonography for assessment of cervical cancer. Strahlenther Onkol 2013;189:123e128. [4] P€otter R, Federico M, Sturdza A, et al. Value of magnetic resonance imaging without or with applicator in place for target definition in cervix cancer brachytherapy. Int J Radiat Oncol Biol Phys 2016;94:588e597. [5] Viswanathan AN, Dimopoulos J, Kirisits C, et al. Computed tomography versus magnetic resonance imaging-based contouring in cervical cancer brachytherapy: results of a prospective trial and preliminary guidelines for standardized contours. Int J Radiat Oncol Biol Phys 2007;68:491e498. [6] P€otter R, Georg P, Dimopoulos JC, et al. Clinical outcome of protocol based Image (MRI) guided adaptive brachytherapy combined with 3D conformal radiotherapy with or without chemotherapy in patients with locally advanced cervical cancer. Radiother Oncol 2011;100: 116e123. [7] Haie-Meder C, Potter R, Van Limbergen E, et al. Recommendations from Gynaecological (GYN) GEC-ESTRO Working Group (I): concepts and terms in 3D image based 3D treatment planning in cervix cancer brachytherapy with emphasis on MRI assessment of GTV and CTV. Radiother Oncol 2005;74:235e245. [8] Nesvacil N, Potter R, Sturdza A, et al. Adaptive image guided brachytherapy for cervical cancer: a combined MRI-/CT-planning technique with MRI only at first fraction. Radiother Oncol 2013; 107:75e81. [9] Lang S, Nesvacil N, Kirisits C, et al. Uncertainty analysis for 3D image-based cervix cancer brachytherapy by repetitive MR imaging: assessment of DVH- variations between two HDR fractions within one applicator insertion and their clinical relevance. Radiother Oncol 2013;107:26e31. [10] van Dyk S, Schneider M, Kondalsamy-Chennakesavan S, et al. Ultrasound use in gynecologic brachytherapy: time to focus the beam. Brachytherapy 2015;14:390e400. [11] Hellebust TP, Kirisits C, Berger D, et al. Recommendations from Gynaecological (GYN) GEC-ESTRO Working Group (III): considerations and pitfalls in commissioning and applicator reconstruction in 3D image-based treatment planning of cervix cancer brachytherapy. Radiother Oncol 2010;96:153e160.
-
(2016)
-
Technical Note
Combining transrectal ultrasound and CT for image-guided adaptive brachytherapy of cervical cancer: Proof of concept Nicole Nesvacil1,2,*, Maximilian P. Schmid1, Richard P€ otter1,2, Gernot Kronreif3, Christian Kirisits1,2 2
1 Department of Radiation Oncology, Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Medical University of Vienna, Austria 3 Austrian Center for Medical Innovation and Technology, Wr. Neustadt, Austria
ABSTRACT
PURPOSE: To investigate the feasibility of a treatment planning workflow for three-dimensional image-guided cervix cancer brachytherapy, combining volumetric transrectal ultrasound (TRUS) for target definition with CT for dose optimization to organs at risk (OARs), for settings with no access to MRI. METHODS AND MATERIALS: A workflow for TRUS/CT-based volumetric treatment planning was developed, based on a customized system including ultrasound probe, stepper unit, and software for image volume acquisition. A full TRUS/CT-based workflow was simulated in a clinical case and compared with MR- or CT-only delineation. High-risk clinical target volume was delineated on TRUS, and OARs were delineated on CT. Manually defined tandem/ring applicator positions on TRUS and CT were used as a reference for rigid registration of the image volumes. Treatment plan optimization for TRUS target and CT organ volumes was performed and compared to MRI and CT target contours. RESULTS: TRUS/CT-based contouring, applicator reconstruction, image fusion, and treatment planning were feasible, and the full workflow could be successfully demonstrated. The TRUS/ CT plan fulfilled all clinical planning aims. Dose-volume histogram evaluation of the TRUS/CToptimized plan (high-risk clinical target volume D90, OARs D2cm3 for) on different image modalities showed good agreement between dose values reported for TRUS/CT and MRI-only reference contours and large deviations for CT-only target parameters. CONCLUSIONS: A TRUS/CT-based workflow for full three-dimensional image-guided cervix brachytherapy treatment planning seems feasible and may be clinically comparable to MRIbased treatment planning. Further development to solve challenges with applicator definition in the TRUS volume is required before systematic applicability of this workflow. Ó 2016 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.
Keywords:
Cervix cancer; Brachytherapy; Image fusion; Transrectal ultrasound
Introduction Received 27 May 2016; received in revised form 16 August 2016; accepted 25 August 2016. Financial disclosure: The Department of Radiotherapy at the Medical University of Vienna receives/received financial and/or equipment support for research and educational purposes from Nucletron, an Elekta company. This work has been supported by ACMITdAustrian Center for Medical Innovation and Technology, which is funded within the scope of the COMETdCompetence Centers for Excellent Technologies program of the Austrian Government. The project was funded in part via a research contract from Elekta. * Corresponding author. Department of Radiation Oncology, Comprehensive Cancer Center, Medical University of Vienna, General Hospital of Vienna, W€ahringer G€urtel 18-20, A-1090 Vienna, Austria. Tel.: þ431404002692; fax: þ431404002693. E-mail address: [email protected] (N. Nesvacil).
There is increasing evidence on the benefits of performing transrectal ultrasound (TRUS) for the assessment of cervical cancer (1e3). TRUS could also be a valuable imaging modality for image-guided adaptive brachytherapy (IGABT) of cervical cancer, not only for real-time image guidance of applicator insertion, but also for target volume identification and assessment of dosimetric coverage. In particular, TRUS could become an important additional imaging modality, if MRI with applicator in situ is not available for treatment planning. However, presently, there are no clear concepts how to integrate TRUS into IGABT. Although TRUS can be used as sole imaging technique for prostate brachytherapy
1538-4721/$ - see front matter Ó 2016 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.brachy.2016.08.009
2
N. Nesvacil et al. / Brachytherapy
(BT), there are major limitations for TRUS-based cervical cancer BT. The relatively small field of view of TRUS and the anatomical changes caused by the insertion of the TRUS probe might pose a substantial challenge for the full evaluation of surrounding organs at risk (OARs). However, the combination with another three-dimensional (3D) volumetric imaging method might overcome those limitations. CT is widely used for 3D image-guided BT as it provides excellent depiction of BT applicators and OARs. However, the lower soft tissue contrast in comparison with MRI poses a challenge for target definition on CT images (4, 5). Using a combination of TRUS and CT may benefit IGABT treatment planning as the advantages of each modality may compensate for their respective disadvantages. Such a method could be beneficial, especially in settings where access to MRI is limited at the time of BT. In this article, we describe a TRUS/CT-based workflow that we tested in our institution and compared to MRI onlye and CT onlyebased treatment planning.
Methods and materials Patient For this study, a TRUS/CT-based workflow was simulated for a patient treated at our institution with MRIbased IGABT. The patient was 52 years old and had biopsy-confirmed squamous cell carcinoma of the uterine cervix staged as International Federation of Gynaecology and Obstetrics (FIGO) IIB. High-dose-rate IGABT was performed after radiochemotherapy according to our previously described clinical protocol (6). For BT, a 60 mm/60 tandem, a 34-mm Vienna ring (Elekta, Sweden) and four interstitial titanium needles were used in both applications. The tandem ring applicator and the interstitial needles were inserted under transabdominal and TRUS guidance and were fixated by vaginal packing. Treatment planning was based on MR imaging. In addition, TRUS and CT were performed. TRUS, CT, and 1.5 T MR images of this patient were part of a previously performed ultrasound assessment study (2), and the study was approved by the institution’s ethics committee.
-
(2016)
-
during a rotation (longitudinal image acquisition) of the TRUS probe in the stepper unit, with the patient in lithotomy position. Volume acquisition was done with a fan angle of 90 , 7 cm depth, and 7 cm length for longitudinal scans. The applicator was fixated to the stepper during volume acquisition by means of a small holding arm (Fisso, Switzerland) and dedicated connectors. CT was performed with a Somatom Plus S scanner (Siemens, Germany) in 2-mm slice intervals, and T2weighted MRI was performed with a 1.5 T system (Avanto, Siemens, Germany) in axial, para-axial, sagittal, and coronal image orientation with 5-mm slice thickness. Based on the GEC-ESTRO recommendations for MRIbased contouring (7), an adapted contouring protocol was used for TRUS delineation. The CTVHR was defined on TRUS by gray-scale imaging as the complete hypoechogenic (dark) cervix surrounded by the bright parametrial tissue. Contouring was performed on TRUSBT, on reconstructed transversal planes acquired by the longitudinal array (Fig. 1a), taking into consideration also TRUSpreBT. CTVHR contouring was performed on MRI for the clinical planning, as well as on CT for interimage-modality comparison in this study, in accordance to respective guidelines and recommendations (5, 7). Three-dimensional measurements of CTVHR maximum width, thickness, and height were recorded, according to the protocol described in (2), on CT, TRUS, and MRI. Due to the previously reported difficult assessment of CTVHR height on TRUS and CT (6), an MRI at diagnosis was used to support the definition of the CTVHR height. In comparison to the MRI-based CTVHR concept, the target height on CT and TRUS at the time of BT would therefore be overestimated. Taking the height of the initial target into account for target definition on other modalities was chosen to avoid geometrical misses at the time of BT. OARs (rectum, colon sigmoideum, small bowel, and urinary bladder) were contoured on MRI and CT. Contouring of all data sets was done blinded, by one observer. All contouring, applicator reconstruction, and treatment planning presented in this study were done with the Oncentra GYN treatment planning system (v 1.2.3, Nucletron, the Netherlands). Applicator reconstruction on TRUS
Imaging and contouring on TRUS TRUS was performed before (TRUSpreBT) and after insertion of the applicator (TRUSBT) with a 5e10 MHz Integrated US transducer (BiopSee, MedCom, Germany), which consists of a biplane probe (BIPC6.5/10/128, BIPL7.5/70/128), specially designed and optimized for transrectal usage. The biplanar TRUS probe was inserted transrectally and fixated with a customized ultrasound stepper unit. Continuous 3D image acquisition was performed by a dedicated registration software (based on an adapted version of Oncentra Prostate, Elekta, Sweden) (1) during a manual pullback (transversal image acquisition) and (2)
The TRUS-based applicator reconstruction was done using a 3D library model of the CT/MR compatible plastic Vienna tandem/ring (Elekta, Sweden). On transversally and/or sagittally reconstructed views, the center of the ring and the position of the tandem near the cranial edge of the TRUS volume were defined as a starting point to place the 3D applicator model. Sagittal and coronal views were used to rotate the 60 tandem ring applicator until it was correctly aligned with the tandem and vagina axis, as seen in the images. On transversal images, the visible needle positions could also be used for rotation fine tuning.
N. Nesvacil et al. / Brachytherapy
-
(2016)
-
3
Fig. 1. Depiction and reconstruction of BT implant on TRUS images: tandem (T), ring (R), and Ti needles (N) in transversal (left), sagittal (middle), and coronal (right) views. (aec): applicator depiction in TRUS images. Dotted lines indicate slice positions for all images in the same row. (def): TRUS image and reconstructed 3D applicator model. (gei): Depiction of needles in TRUS images. Note that window levels have been adjusted in the TPS, to increase visibility of applicator and enable direct reconstruction. BT 5 brachytherapy; TPS 5 treatment planning system; TRUS 5 transrectal ultrasound; 3D, three dimensional.
TRUS/CT image fusion and treatment planning For treatment planning, a combination of the TRUS image volume and the CT image volume was obtained by rigid image registration based on the applicator position defined on the TPS. The applicator was reconstructed on CT images using the same 3D library model described in the previous section. As described in Nesvacil et al. (8), Oncentra GYN can perform an automatic image registration based on the defined applicator position as a common reference coordinate system. Because target and applicator are fixed to each other by the vaginal packing, no relative movement between them occurs between TRUS and CT image acquisition. The stability of these implants has been reported previously by Lang et al. (9). After TRUS/CT image registration, the delineated TRUS target contour was transferred to the CT image volume and the OARs (bladder, rectum, sigmoid) were delineated on CT. The interstitial needles were reconstructed
as depicted on CT. Treatment planning was simulated according to our clinical protocol for four HDR fractions using the following planning aims (EBRT þ IGABT): CTVHR D90 of $85 Gy EQD2 (a/b 5 10 Gy), as well as total OAR D2cm3 of #90 Gy for bladder, #70 Gy for rectum, and #70 Gy for sigmoid, in EQD2a/b53 Gy. To evaluate the potential for the TRUS/CT treatment planning method to obtain similar treatment plan quality as obtained by the use of standard delineation techniques, the TRUS/CT-based dose distribution was evaluated on the MRI and CT data sets.
Results Contouring, applicator reconstruction, image fusion, and TRUS/CT-based treatment planning were performed for this patient. The feasibility of the full described workflow
4
N. Nesvacil et al. / Brachytherapy
-
(2016)
-
Fig. 2. Illustration of the first full TRUS/CT planning workflow by applicator-based registration with a customized TPS. (1) TRUS volume acquisition and target delineation (red line) in relation to applicator position. (2) US and CT image registration via applicator and US target transfer to CT. (3) OAR (bladder, rectum, and sigmoid) contouring on CT and dose planning with the full US/CT structure set. For comparison, three targets are shown: US based (red), MR based (green), and CT based (blue). The dashed orange isodose line corresponds to total dose of 85 Gy EQD2a/b510Gy of the US/CT-optimized treatment plan, assuming a full treatment of 45 Gy EBRT þ 4 HDR BT. Transversal view (left column) and sagittal view (right column) are shown. The correlation of transversal and sagittal views in (1) and (3) is indicated by a dashed line. BT 5 brachytherapy; EBRT 5 external beam radiotherapy; HDR 5 high-doserate; OAR 5 organs at risk; TRUS 5 transrectal ultrasound; US 5 ultrasound. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
N. Nesvacil et al. / Brachytherapy
was simulated, and individual steps of the procedure, performed with our prototype system, were evaluated. The results are illustrated in Figs. 1 and 2. Image acquisition During volume acquisition by pullback of the TRUS probe, a distortion of the visualized 3D volume was observed, which could be attributed to small movements of the ring applicator. Therefore, the tandem and ring were no longer aligned perpendicularly in the transversal US image series, which rendered it unsuitable for treatment planning. When images were acquired with a rotation of the TRUS probe using the longitudinal array, no applicator movement was observed by the operators or in the acquired image volume. Therefore, the image volume acquired with the longitudinal transducer was used for treatment planning. Applicator reconstruction Other than on CT or MRI only parts of the intracavitary applicator could be identified on TRUS. The posterior part of the ring was well depicted in the reconstructed transversal, sagittal, and particularly in coronal views (Figs. 1ae1c). The tandem was very well visible on reconstructed transversal and coronal views, and excellent depiction was given in the sagittal plane (Fig. 1b). Therefore, these parts of the applicator were simultaneously observed in all three reconstructed image views during fine tuning of the reconstruction (Figs. 1de1f). By scrolling through the images in all available orientations, placement of the applicator was verified and fine tuned until a plausible applicator position was established for all views. The anterior part of the ring was hardly distinguishable from the posterior bladder wall, due to artifacts from the applicator. Similar to the tandem, the interstitial titanium needles were identified in all reconstructed image orientations (Figs. 1ge1i). Despite the overall limitations of applicator depiction in TRUS images, for this patient, the achieved applicator reconstruction was found acceptable for applicator-based TRUS/CT registration and further investigation of the treatment planning workflow. Contouring The CTVHR volume was 30 cm3 on TRUS, 31 cm3 on MRI, and 59 cm3 on CT. The CTVHR width was 53 mm, 49 mm, and 71 mm, and the thickness was 34 mm, 33 mm, and 52 mm, respectively, whereas the CTVHR height was within 10% for all target volumes. All CTVs were transferred to the CT data set and are shown in Fig. 2, panel 3. The TRUS and MRI contours differed mainly in the ventral part near the posterior bladder wall, which corresponds to the area where it is most difficult to distinguish the target volume from the bladder on TRUS images.
-
(2016)
-
5
Differences in bladder filling might also contribute to this finding. On CT, the extent of parametrial involvement indicated by the contrast enhancement of the tumor was overestimated on both sides and posteriorly in direction of the mesorectal fascia, whereas with TRUS, the involvement could be defined similar to MRI. Due to limits of overall probe length, rectum length, and the location of the transversal array distal to the longitudinal array, the cranial border of the target was not clearly depicted on longitudinal TRUS images. Therefore, on TRUS as well as on CT, the height of the target on pretreatment MRI defined the cranial border of the delineated contour. Treatment planning An overview of the treatment planning workflow is shown in Fig. 2. The TRUS/CT plan fulfilled all planning aims (target: 92 Gy, bladder: 85 Gy, rectum: 64 Gy, sigmoid: 66 Gy [EQD2a/b53Gy]). The evaluation of the TRUS/CT treatment plan on the CT-only, the TRUS/CT, and the MRIref contour set resulted in a CTVHR D90 of 69 Gy, 92 Gy, and 89 Gy (EQD2a/b510Gy), respectively. For OARs, the total D2cm3 of all structure sets were within 3% for rectum and bladder and 5% for sigmoid (Table 1). Based on the evaluation of the TRUS/CT dose distribution on MRI structures, the plan would have fulfilled clinical acceptance criteria.
Discussion In this study, we analyzed the possibility of a combined TRUS/CT treatment planning approach for cervix cancer BT. The complete workflow, from image acquisition, contouring, and applicator reconstruction to image fusion and treatment planning, could be demonstrated. The huge potential of ultrasound (1, 10) and especially TRUS (2, 3) has been previously demonstrated. A 3D volumetric image acquisition including the target volume, the relevant OARs, and the applicator has not been described so far. Due to the limited field of view of 3D volumetric TRUS, full evaluation of OARs and the whole applicator is not possible; therefore, combination with an additional CT image volume seems meaningful. Applicator reconstruction was Table 1 Evaluation of the treatment plan optimized for TRUS CTVHR and CT OARs, for three contour sets: TRUS/CT, MRI only, and CT only Evaluated parameter
TRUS/CT contours
MRI contours
CT contours
CTVHR D90 (Gy) Bladder D2cm3 (Gy) Rectum D2cm3 (Gy) Sigmoid D2cm3 (Gy)
92.3 85.2 63.5 66.1
88.8 84.0 63.7 62.9
69.0 85.2 63.5 66.1
CTVHR 5 high-risk clinical target volume; OAR 5 organs at risk; TRUS 5 transrectal ultrasound.
6
N. Nesvacil et al. / Brachytherapy
performed directly on TRUS images, by adapting existing guidelines (11). First results on the potential of TRUS for target depiction in cervix cancer BT have been published previously by Schmid et al. (2, 3). In comparison, TRUS image examples shown in the present study were optimized for simultaneously depicting target and applicator. The latter is necessary to perform a direct applicator reconstruction, as is done routinely for CT- and MRI-based treatment planning. However, even then applicator visibility is limited by artifacts in the anterior part of the ring or the TRUS volume being too short to contain the tandem tip. Alternative solutions for applicator placement in the TRUS volume might help to overcome some of these limitations and allow to improve target depiction using an optimized imaging protocol. For the investigated clinical case, the combination of CT for OAR depiction and TRUS for target definition resulted in a plan and dose reporting almost similar to a full MRI approach. In this specific case, the location and orientation of the applicator could be directly reconstructed with a similar accuracy as typically achieved on MRI. For other cases observed during the system development (unpublished results), substantial artifacts showed that this direct applicator reconstruction would not always be possible on the ultrasound images. This study shows that a workflow for TRUS/CT-based treatment planning is, in principle, feasible in a clinical setting, if customized software and hardware for volumetric image acquisition, DICOM export to TPS, and applicatorbased image registration are available. Further developments of hardware, software, and operation procedures, that is, optimization of machine settings, training of users, and systematic clinical testing, are required to increase image quality and precision of target delineation and applicator depiction. Future studies applying a systematic treatment planning workflow are needed to standardize individual steps in the procedure. Based on the proposed workflow, followup monocentric and multicentric studies are needed to study its performance for different stages of disease and the interobserver and intraobserver variability, to develop and validate a reproducible target contouring concept. Important aspects such as precision of applicator reconstruction and image registration will have to be further investigated. Similar to the successful development of CT- and MRI-based gynecological BT, joint efforts are needed to fully exploit the potential of TRUS/CT-based treatment planning. The most critical technological element for a systematically applicable procedure remains the accurate reconstruction
-
(2016)
-
of the applicator in the ultrasound images, which is the basis for appropriate image registration. If it will be possible to establish applicator tracking during the ultrasound acquisition, this TRUS/CT-based workflow may become available as a full 3D volumetric treatment planning approach. BT applicator tracking using optical, electromagnetic, or other approaches is being investigated by several research groups. The combination of these methods with the proposed workflow may result in a feasible technical solution. References [1] Epstein E, Testa A, Gaurilcikas A, et al. Early-stage cervical cancer: tumor delineation by magnetic resonance imaging and ultrasoundda European multicenter trial. Gynecol Oncol 2013;128:449e453. [2] Schmid MP, Nesvacil N, P€otter R, et al. Transrectal ultrasound for image-guided adaptive brachytherapy in cervix cancerdan alternative to MRI for target definition? Radiother Oncol 2016;8140: 00052e00059. [3] Schmid MP, P€otter R, Brader P, et al. Feasibility of transrectal ultrasonography for assessment of cervical cancer. Strahlenther Onkol 2013;189:123e128. [4] P€otter R, Federico M, Sturdza A, et al. Value of magnetic resonance imaging without or with applicator in place for target definition in cervix cancer brachytherapy. Int J Radiat Oncol Biol Phys 2016;94:588e597. [5] Viswanathan AN, Dimopoulos J, Kirisits C, et al. Computed tomography versus magnetic resonance imaging-based contouring in cervical cancer brachytherapy: results of a prospective trial and preliminary guidelines for standardized contours. Int J Radiat Oncol Biol Phys 2007;68:491e498. [6] P€otter R, Georg P, Dimopoulos JC, et al. Clinical outcome of protocol based Image (MRI) guided adaptive brachytherapy combined with 3D conformal radiotherapy with or without chemotherapy in patients with locally advanced cervical cancer. Radiother Oncol 2011;100: 116e123. [7] Haie-Meder C, Potter R, Van Limbergen E, et al. Recommendations from Gynaecological (GYN) GEC-ESTRO Working Group (I): concepts and terms in 3D image based 3D treatment planning in cervix cancer brachytherapy with emphasis on MRI assessment of GTV and CTV. Radiother Oncol 2005;74:235e245. [8] Nesvacil N, Potter R, Sturdza A, et al. Adaptive image guided brachytherapy for cervical cancer: a combined MRI-/CT-planning technique with MRI only at first fraction. Radiother Oncol 2013; 107:75e81. [9] Lang S, Nesvacil N, Kirisits C, et al. Uncertainty analysis for 3D image-based cervix cancer brachytherapy by repetitive MR imaging: assessment of DVH- variations between two HDR fractions within one applicator insertion and their clinical relevance. Radiother Oncol 2013;107:26e31. [10] van Dyk S, Schneider M, Kondalsamy-Chennakesavan S, et al. Ultrasound use in gynecologic brachytherapy: time to focus the beam. Brachytherapy 2015;14:390e400. [11] Hellebust TP, Kirisits C, Berger D, et al. Recommendations from Gynaecological (GYN) GEC-ESTRO Working Group (III): considerations and pitfalls in commissioning and applicator reconstruction in 3D image-based treatment planning of cervix cancer brachytherapy. Radiother Oncol 2010;96:153e160.