Please select the desired project time frame:

• July 2026
• January 2026
• July 2025
• January 2025
• July 2024
• January 2024
• July 2023
• January 2023
• July 2022
• January 2022
• July 2021
• January 2021
• July 2020
• January 2020
• July 2019
• January 2019
• July 2018
• January 2018
• July 2017
• January 2017
• July 2016
• January 2016
• July 2015
• January 2015
• July 2014
• January 2014
• July 2013
• January 2013
• July 2012
• January 2012
• July 2011
• January 2011
• July 2010
• January 2010
• July 2009
• January 2009
• July 2008
• January 2008
• July 2007
• January 2007
• July 2006
• January 2006
• July 2005
• January 2005
• July 2004
• January 2004
• July 2003
• January 2003
• July 2002
• January 2002
• July 2001
• January 2001

Start of funding 01.01.2015

Ultra-fast Image Registration for Ultra-fast Radiation Therapy

---

One fundamental remaining barrier to precise, accurate, highly conformal radiation therapy is patient, tumor, and organ motion that occurs during treatment delivery. Conventional radiation delivery times are long relative to the time scale for physiological motion, requiring between 2 minutes and up to 90 minutes for more complex dose painting. Stanford University is developing new hardware approaches (very high-energy electron beams) that permit delivery of a full dose of radiation therapy within 1 s. This fundamentally different approach would allow physiological motion to be ‚frozen‘ during treatment. The successful implementation of this technology presents a new and unique challenge to image guidance: very fast registration with the current physical position of the patient followed by modification of the treatment plan is required. To achieve this real-time medical image processing methods are required which are being investigated at University of Erlangen-Nuremberg. In the proposed project, we aim at using manifold learning techniques to quickly generate a treatment plan.

Final report:
This project discusses fast pose verification for radiation therapy on a new high-speed radiation therapy device. The PHASER system follows the idea of 4th generation CT imaging and allows fast 360 degree treatment using a steerable electron beam. Doing so, dose delivery is possible in few seconds. A major problem, however, is fast verification of the patient pose during treatment. In this project, we suggest to use a projection-based approach that can be evaluated quickly and allows an accuracy below 1 mm as shown by our simulation study based on planning data from six 4-D CT data sets.

1 Introduction

Patient motion is a major problem for imaging [1] and radiation therapy [2]. In radiation therapy, dose delivery is typically planned on a 3-D CT image [3] and high attention is paid to align the patient’s actual position at the treatment site with the planning scan [4]. For tumors in the head and neck region, this can be achieved by the use of an immobilization mask [5] which prevents head motion by fixating the patient to the couch. Due to the long duration of the radiation treatment, there is also motion that cannot be avoided completely. In particular, respiratory motion may cause the target area to move up to 2 cm [6]. Without compensation, this would cause the dose to be delivered to the wrong location, resulting in damaged healthy tissues and more importantly the poten- tial survival of the malignant tumor itself [7]. In order to compensate for this, many approaches have been suggested ranging from implanted gold markers [8] to the use of respiratory surrogate signals [9,10] and the prediction of dense deformation fields [11,12]. In summary, these methods are feasible, but come at significant additional efforts.

In this project, we focus on a different treatment device that has been suggested by Maxim and Loo [13]. The Pluridirectional High-Energy Agile Scanning Elec- tron Radiotherapy (PHASER) System is able to deliver the entire treatment dose within only a few seconds and, therefore, provides a treatment duration range that can effectively compensate for respiratory motion by a simple breath hold command. Additionally, in contrast to traditional radiation therapy, electrons instead of photons are used to deliver the radiation dose. This way, energy can be deposited much faster using fewer particles and with higher accuracy [14].

2 Materials and Methods

2.1 PHASER System

While delivering such high amounts of dose in such short time has many ad- vantages, it also poses special challenges to the imaging. In order to image the patient quickly, a special CT gantry was designed [15]. The associated detec- tor is curved on a circle segment with a diameter of 1300 mm, an arc length of 1024 mm and a detector height of 192 mm. The detector provides high resolu- tion in the center of the field-of-view (FOV) while offering larger pixels towards the outside of the FOV which typically is of lower interest for the purpose of radiation therapy.

2.2 Pose Verification

In contrast to typical radiation therapy, the PHASER system allows only very little time to verify the patient pose and deliver the dose. The whole process must be completed within a single breath-hold. As patients often suffer from impaired lung function, we assume this period to be within the range of 12 to 16 seconds [16]. Therefore, computationally expensive approaches that require reconstruction and motion compensation are not applicable [6]. Furthermore, we do not want to create an additional burden to the patients by using e.g. implanted gold markers [5].

The current work-flow on the PHASER system will involve a high-quality CT scan, registration to the planning CT, and an adaptation of the treatment plan [15] to accommodate the current patient position. Hence, we expect the motion during the treatment to emerge only from respiratory motion.

In order to compensate for the current motion state, we propose to use projection-based imaging only, based on projections of the 4-D planning CT. In addition, we reuse a lung segmentation that is created during the planing procedure for the pose verification process. Doing so, we compute a similarity measure in the lung area from the projection data.

3 Results

From the relative change in lung volume be observed that states with a similar lung volume also show similar values in our measure. We observed a similar relation in the other five patients.

In a second experiment, we excluded one of the breathing phases from the data and performed a leave-one-out evaluation to estimate the accuracy of unknown breathing motion. We observe that the average error is much below the maximal tumor motion. This results in an average error of 0.80 ± 0.25 mm compared to 2.50 ± 2.10 mm maximal tumor motion. In the case with the largest tumor motion of 7 mm the error is reduced down to 0.83 mm.

4 Discussion

From the experimental results, we observed that the localization accuracy is on average below 1 mm. Compared to current clinical safety margins of about 7 mm even for motion compensated treatment, this is a great reduction. However, one has to be careful with the interpretation of the results, as the average motion in our patients was only 2.50 ± 2.10 mm. Thus, our error is well below the maximal motion, but also higher than one would expect given that the motion was sampled ten times. This is related to the resolution of our 4-D CT scan that had a voxel size of 0.98 × 0.98 × 2.0 mm^3 . In case of our patients, the main magnitude of motion occurs along the z axis, i.e. our results lie below the accuracy of one voxel. In future studies, we will have to verify whether this low amount of motion occurs in more patients with lung cancer. As a result, we would have to increase the resolution of the planning CT in the affected directions to alleviate the problem. A possible solution for this problem would be adaptive detector binning [17]. Nonetheless even with the current setup, patients with large tumor motion benefit greatly from the method. In a patient with a large tumor motion of 7 mm, the error could be reduced down to 0.8 mm thereby preventing incorrect deposition of the radiation dose.

Another challenge that we plan to investigate in future work is continuous treatment using precomputed 4-D treatment plans for patients such as young children who cannot follow breathing commands. With the current system setup, we would be able to select the correct treatment plan for the current motion state in real-time.

References

[1] Nehmeh SA, Erdi YE. Respiratory Motion in Positron Emission Tomogra- phy/Computed Tomography: A Review. Seminars in Nuclear Medicine: Devel- opments in Instrumentation. 2008;38(3):167 – 176.
[2] Korreman SS. Motion in radiotherapy: photon therapy. Physics in Medicine and Biology. 2012;57(23):R161.
[3] Censor Y, Altschuler MD, Powlis WD. A computational solution of the inverse problem in radiation-therapy treatment planning. Applied Mathematics and Com- putation. 1988;25(1):57 – 87.
[4] Bauer S, Wasza J, Haase S, Marosi N, Hornegger J. Multi-modal Surface Registra- tion for Markerless Initial Patient Setup in Radiation Therapy using Microsoft’s Kinect Sensor. In: IEEE International Conference on Computer Vision (ICCV) Workshops; 2011. p. 1175–1181.
[5] Chen S, Lu Y, Hopfgartner C, Su ?hling M, Steidl S, Hornegger J, et al. 3-D Print- ing Based Production of Head and Neck Masks for Radiation Therapy Using Ct Volume Data: A Fully Automatic Framework. In: IEEE International Symposium on Biomedical Imaging: From Nano to Macro. Prague, Czech Republic; 2016. p. 403–406.
[6] Bögel M, Hofmann HG, Hornegger J, Fahrig R, Britzen S, Maier A. Respira- tory motion compensation using diaphragm tracking for cone-beam c-arm CT: a simulation and a phantom study. Journal of Biomedical Imaging. 2013;2013:6.
[7] Shepard DM, Ferris MC, Olivera GH, Mackie TR. Optimizing the Delivery of Radiation Therapy to Cancer Patients. SIAM Review. 1999;41(4):721–744.
[8] Huang CY, Tehrani JN, Ng JA, Booth J, Keall P. Six degrees-of-freedom prostate and lung tumor motion measurements using kilovoltage intrafraction monitoring. International Journal of Radiation Oncology*Biology*Physics. 2015;91(2):368–375.
[9] Bögel M, Maier A, Hofmann HG, Hornegger J, Fahrig R. Diaphragm tracking in cardiac C- Arm projection data. In: Bildverarbeitung fu ?r die Medizin 2012. Springer Berlin Heidelberg; 2012. p. 33–38.
[10] Wasza J,Fischer P,Leutheuser H,Oefner T,Bert C,Maier A, et al. Real-TimeRespiratory Motion Analysis Using 4-D Shape Priors. IEEE Transactions on Biomed- ical Engineering. 2016;63(3):485–495.
[11] Taubmann O, Wasza J, Forman C, Fischer P, Wetzl J, Maier A, et al. Prediction of Respiration-Induced Internal 3-D Deformation Fields From Dense External 3- D Surface Motion. In: Computer Assisted Radiology and Surgery (CARS), 28th International Congress and Exhibition. International Journal of Computer Assisted Radiology and Surgery. Heidelberg; 2014. p. 33–34.
[12] Geimer T, Unberath M, Taubmann O,Bert C,Maier A.CombinationofMarkerless Surrogates for Motion Estimation in Radiation Therapy. In: Computer Assisted Radiology and Surgery (CARS), 30th International Congress and Exhibition; 2016. p. 59–60.
[13] Maxim PG, Loo BW. Pluridirectional High-Energy Agile Scanning Electron Ra- diotherapy (PHASER): Extremely Rapid Treatment for Early Lung Cancer. DTIC Document; 2015.
[14] Bazalova-Carter M, Qu B, Palma B, Hardemark B, Hynning E, Jensen C, et al. Treatment planning for radiotherapy with very high