Sonja Dieterich

Comparative performance of linear and nonlinear neural networks to predict irregular breathing

Breathing adaptation during external-beam radiotherapy is a matter of great concern because uncompensated tumour motion requires extended treatment margins that endanger sensitive tissue. Compensation strategies include beam gating, collimator tracking and robotic beam re-alignment. All of these schemes have a system latency of up to several hundred milliseconds, which calls in turn for predictive control loops. Irregularities in breathing make prediction difficult. We have evaluated the performance of two classes of control loop algorithms-the linear adaptive filter and the adaptive nonlinear neural network-for highly irregular patient breathing behaviours. The neural network demonstrated robust adaptability to all of the observed breathing patterns while the linear filter failed in a significant percentage of cases. For those cases where the linear filter could function, it made less accurate predictions than the neural network. Because the neural network presents no additional computational burden in the control loop we conclude that it is the preferred choice among heuristic predictive algorithms.

Radical stereotactic radiosurgery with real-time tumor motion tracking in the treatment of small peripheral lung tumors

BackgroundRecent developments in radiotherapeutic technology have resulted in a new approach to treating patients with localized lung cancer. We report preliminary clinical outcomes using stereotactic radiosurgery with real-time tumor motion tracking to treat small peripheral lung tumors.MethodsEligible patients were treated over a 24-month period and followed for a minimum of 6 months. Fiducials (3–5) were placed in or near tumors under CT-guidance. Non-isocentric treatment plans with 5-mm margins were generated. Patients received 45–60 Gy in 3 equal fractions delivered in less than 2 weeks. CT imaging and routine pulmonary function tests were completed at 3, 6, 12, 18, 24 and 30 months.ResultsTwenty-four consecutive patients were treated, 15 with stage I lung cancer and 9 with single lung metastases. Pneumothorax was a complication of fiducial placement in 7 patients, requiring tube thoracostomy in 4. All patients completed radiation treatment with minimal discomfort, few acute side effects and no procedure-related mortalities. Following treatment transient chest wall discomfort, typically lasting several weeks, developed in 7 of 11 patients with lesions within 5 mm of the pleura. Grade III pneumonitis was seen in 2 patients, one with prior conventional thoracic irradiation and the other treated with concurrent Gefitinib. A small statistically significant decline in the mean % predicted DLCO was observed at 6 and 12 months. All tumors responded to treatment at 3 months and local failure was seen in only 2 single metastases. There have been no regional lymph node recurrences. At a median follow-up of 12 months, the crude survival rate is 83%, with 3 deaths due to co-morbidities and 1 secondary to metastatic disease.ConclusionRadical stereotactic radiosurgery with real-time tumor motion tracking is a promising well-tolerated treatment option for small peripheral lung tumors.

Geometric uncertainty of 2D projection imaging in monitoring 3D tumor motion

The purpose of this study was to investigate the accuracy of two-dimensional (2D) projection imaging methods in three-dimensional (3D) tumor motion monitoring. Many commercial linear accelerator types have projection imaging capabilities, and tumor motion monitoring is useful for motion inclusive, respiratory gated or tumor tracking strategies. Since 2D projection imaging is limited in its ability to resolve the motion along the imaging beam axis, there is unresolved motion when monitoring 3D tumor motion. From the 3D tumor motion data of 160 treatment fractions for 46 thoracic and abdominal cancer patients, the unresolved motion due to the geometric limitation of 2D projection imaging was calculated as displacement in the imaging beam axis for different beam angles and time intervals. The geometric uncertainty to monitor 3D motion caused by the unresolved motion of 2D imaging was quantified using the root-mean-square (rms) metric. Geometric uncertainty showed interfractional and intrafractional variation. Patient-to-patient variation was much more significant than variation for different time intervals. For the patient cohort studied, as the time intervals increase, the rms, minimum and maximum values of the rms uncertainty show decreasing tendencies for the lung patients but increasing for the liver and retroperitoneal patients, which could be attributed to patient relaxation. Geometric uncertainty was smaller for coplanar treatments than non-coplanar treatments, as superior-inferior (SI) tumor motion, the predominant motion from patient respiration, could be always resolved for coplanar treatments. Overall rms of the rms uncertainty was 0.13 cm for all treatment fractions and 0.18 cm for the treatment fractions whose average breathing peak-trough ranges were more than 0.5 cm. The geometric uncertainty for 2D imaging varies depending on the tumor site, tumor motion range, time interval and beam angle as well as between patients, between fractions and within a fraction.

Quantitative Measurement of CyberKnife Robotic Arm Steering

Respiratory motion is a significant and challenging problem for radiation medicine. Without adequate compensation for respiratory motion, it is impossible to deliver highly conformal doses to tumors in the thorax and abdomen. The CyberKnife frameless stereotactic radiosurgery system with Synchrony provides respiratory motion adaptation by monitoring skin motion and dynamically steering the beam to follow the moving tumor. This study quantitatively evaluated this beam steering technology using optical tracking of both the linear accelerator and a ball-cube target. Respiratory motion of the target was simulated using a robotic motion platform and movement patterns recorded from previous CyberKnife patients. Our results show that Synchrony respiratory tracking can achieve sub-millimeter precision when following a moving object.

Cyberknife Image-Guided Delivery and Quality Assurance

The CyberKnife is a complex, emerging technology that is a significant departure from current stereotactic radiosurgery and external beam radiotherapy technologies. In its clinical application and quality assurance (QA) approach, the CyberKnife is currently situated somewhere in between stereotactic radiosurgery and radiotherapy. The clinical QA for this image-guided treatment delivery system typically follows the vendors guidance, mainly because of the current lack of vendor-independent QA recommendations. The problem has been exacerbated because very little published data are available for QA for the CyberKnife system, especially for QA of the interaction between individual system components. The tools and techniques for QA of the CyberKnife are under development and will continue to improve with longer clinical experience of the users. The technology itself continues to evolve, forcing continuous changes and adaptation of QA. To aid in the process of developing comprehensive guidance on CyberKnife QA, a database of errors based on users reporting incidents and corrective actions would be desirable. The goal of this work was to discuss the status of QA guidelines in the clinical implementation of the CyberKnife system. This investigation was done from the perspective of an active clinical and research site using the CyberKnife.

Respiratory motion tracking of skin and liver in swine for Cyberknife motion compensation

In this study, we collected respiratory motion data of external skin markers and internal liver fiducials from several swine. The POLARIS infrared tracking system was used for recording reflective markers placed on the swine’s abdomen. The AURORA electromagnetic tracking system was used for recording 2 tracked needles implanted into the liver. This data will be used to develop correlation models between external skin movement and internal organ movement, which is the first step towards the ability to compensate for respiratory movement of the lesion. We are also developing a motion simulator for validation of our model and dose verification of mobile lesions in the CYBERKNIFE Suite. We believe that this research could provide significant information towards the development of precise radiation treatment of mobile target volumes.

Skin respiratory motion tracking for stereotactic radiosurgery using the CyberKnife

Abstract The purpose of this study is to report and analyze patient skin motion data collected during CyberKnife stereotactic radiosurgery (SRS). The CyberKnife is a radiation treatment system that incorporates a robotic arm to precisely position a linear accelerator. While this capability is not currently commercially available, the CyberKnife could be programmed to move the radiation beam in real time to compensate for organ motion. This study serves as a first step towards our long-term goal of using skin motion to predict internal organ motion. This may lead to more precise radiation treatment delivery for mobile target volumes. To date, we have collected and analyzed skin motion data on four patients (two sacrum, one lung, one pancreas). The movement amplitudes of individual skin markers ranged from 3.1 to 14.8 mm with a median of 7.5 mm.

Tumor Motion Ranges Due to Respiration and Respiratory Motion Characteristics

Many soft-tissue tumors targeted with extracranial SRS move during respiration. New imaging technologies, motion compensation strategies, and treatment planning algorithms are being developed which enable tracking and treatment of moving tumors in real-time. For this chapter we reviewed the literature to determine known tumor motion amplitudes for lung, liver, and pancreas. Then we analyzed predicted tumor motion for 36 patients and 117 treatment fractions that were previously saved in CyberKnife® (Accuray Incorporated, Sunnyvale, CA) treatment logfiles. These represent 27 tumors in the lung (16 upper lung, 4 middle lung, 7 lower lung) and 9 pancreas patients. For each treatment, the location of the target at end inspiration and end expiration was determined in the patient coordinate system. The origin of the patient coordinate system is at the center of mass of the fiducials as marked on the simulation CT, +x is patient inferior, +y patient left, and +z anterior in a right-handed coordinate system. The mean and variance of respiratory cycle extrema positions were calculated using a program written in MatLab code. Observed motion ranges for all sites except pancreas agree very well with the literature. The largest motion amplitudes of up to 38.7 mm were observed in the lower lung. Twenty-five percent of tumors in the upper lung could have been treated without Synchrony® (Accuray Incorporated, Sunnyvale, CA) with a PTV margin of 2 mm, because the uncertainty is in the range of the technical tracking accuracy of Synchrony of 1.5 mm. Possible causes of large fluctuations around the mean motion could be fiducial tracking errors or irregular breathing. We concluded that a subset of all patients could have been treated using skeletal structure tracking, rather than implanted fiducials, and a PTV margin in the range of the stated tracking accuracy for Synchrony. Defining meaningful parameters to characterize the effects of free breathing is part of ongoing research, since published data from non-dynamic SBRT is limited to short fluoroscopic studies or Cine-CT. The results can be transferred to other treatment modalities to determine PTV margins in standard external beam treatments as well as defining the PTV in the third dimension for 2D motion compensation [1].

Percutaneous placement of fiducial markers for thoracic malignancies

Image-guided placement of fiducial markers is in some ways an extension of percutaneous procedures such as needle biopsy of lung pathology, which are native to most interventional radiology practices. To that extent, learning the procedure is not difficult for those who are familiar with the basic principles of image-guided lung nodule biopsy. However, there are significant modifications in the procedure that are necessary in order to ensure appropriate placement and distribution of the fiducial markers. Proper positioning of fiducial markers in specific geometric configurations is essential for accurate targeting of the nodule. This chapter focuses on the principles of CT-guided percutaneous placement of fiducial markers. For the most part, this procedure is performed on a consultative basis by interventional radiologists, physicians who specialize in minimally invasive image-guided therapy. Special considerations for patient selection, pre-procedural preparation, techniques, and post-procedural care are explained.

Respiratory motion compensation studies using a 3D robotic motion simulator and optical/electromagnetic tracking technologies

Respiratory motion can degrade the quality of nuclear medicine images, especially when attempting to identify small abnormalities, make quantitative estimates of activity concentration, or track the time-varying location of a tumor. Thus, we are developing methods for respiratory motion compensation and testing these methods using robotic devices. The testing device is a computer controlled 3-axis motion simulator that can hold activity-filled phantoms or spheres and move them along pre-programmed paths to simulate respiratory motion. The motion of the platform is programmed in advance and can also be monitored using an optical tracking system, thus providing a solid ground truth for the time-dependent activity concentration. Registration between the coordinate space of the PET or SPET scanner and the optical tracker coordinate system is based on a set of common points (mapped out using the motion simulator) that are visible to both systems. We have also used electromagnetic tracking and optical tracking to obtain realistic respiratory motion data from patients and animal models. These data can be transformed into motion simulator paths, thus providing us with breathing patterns that accurately reflect the nonstationary and variable nature of human respiration. The simulator thus provides a highly useful tool for repeatably testing different approaches to motion-compensated image reconstruction or gated acquisition schemes.