Hassan Mostafavi

Quantifying the predictability of diaphragm motion during respiration with a noninvasive external marker

The aim of this work was to quantify the ability to predict intrafraction diaphragm motion from an external respiration signal during a course of radiotherapy. The data obtained included diaphragm motion traces from 63 fluoroscopic lung procedures for 5 patients, acquired simultaneously with respiratory motion signals (an infrared camera-based system was used to track abdominal wall motion). During these sessions, the patients were asked to breathe either (i) without instruction, (ii) with audio prompting, or (iii) using visual feedback. A statistical general linear model was formulated to describe the relationship between the respiration signal and diaphragm motion over all sessions and for all breathing training types. The model parameters derived from the first session for each patient were then used to predict the diaphragm motion for subsequent sessions based on the respiration signal. Quantification of the difference between the predicted and actual motion during each session determined our ability to predict diaphragm motion during a course of radiotherapy. This measure of diaphragm motion was also used to estimate clinical target volume (CTV) to planning target volume (PTV) margins for conventional, gated, and proposed four-dimensional (4D) radiotherapy. Results from statistical analysis indicated a strong linear relationship between the respiration signal and diaphragm motion (p<0.001) over all sessions, irrespective of session number (p=0.98) and breathing training type (p=0.19). Using model parameters obtained from the first session, diaphragm motion was predicted in subsequent sessions to within 0.1 cm (1 sigma) for gated and 4D radiotherapy. Assuming a 0.4 cm setup error, superior-inferior CTV-PTV margins of 1.1 cm for conventional radiotherapy could be reduced to 0.8 cm for gated and 4D radiotherapy. The diaphragm motion is strongly correlated with the respiration signal obtained from the abdominal wall. This correlation can be used to predict diaphragm motion, based on the respiration signal, to within 0.1 cm (1 sigma) over a course of radiotherapy.

Quantitation of respiratory motion during 4D-PET/CT acquisition

We report on the variability of the respiratory motion during 4D-PET/CT acquisition. The respiratory motion for five lung cancer patients was monitored by tracking external markers placed on the abdomen. CT data were acquired over an entire respiratory cycle at each couch position. The x-ray tube status was recorded by the tracking system, for retrospective sorting of the CT data as a function of respiration phase. Each respiratory cycle was sampled in ten equal bins. 4D-PET data were acquired in gated mode, where each breathing cycle was divided into ten 500 ms bins. For both CT and PET acquisition, patients received audio prompting to regularize breathing. The 4D-CT and 4D-PET data were then correlated according to their respiratory phases. The respiratory periods, and average amplitude within each phase bin, acquired in both modality sessions were then analyzed. The average respiratory motion period during 4D-CT was within 18% from that in the 4D-PET sessions. This would reflect up to 1.8% fluctuation in the duration of each 4D-CT bin. This small uncertainty enabled good correlation between CT and PET data, on a phase-to-phase basis. Comparison of the average-amplitude within the respiration trace, between 4D-CT and 4D- PET, on a bin-by-bin basis show a maximum deviation of approximately 15%. This study has proved the feasibility of performing 4D-PET/CT acquisition. Respiratory motion was in most cases consistent between PET and CT sessions, thereby improving both the attenuation correction of PET images, and co-registration of PET and CT images. On the other hand, in two patients, there was an increased partial irregularity in their breathing motion, which would prevent accurately correlating the corresponding PET and CT images.

Comparison of respiratory surrogates for gated lung radiotherapy without internal fiducials

An investigation was carried out to compare the ability of two respiratory surrogates to mimic actual lung tumor motion during audio coaching. The investigation employed video clips acquired after patients had had fiducial markers implanted in lung tumors to be used for image-guided stereoscopic radiotherapy. The positions of the markers in the clips were measured within the video frames and used as the standard for tumor volume motion. An external marker was tracked optically during the fluoroscopic acquisitions. An image correlation technique was developed to compute a gating signal from the fluoroscopic images. The correlation gating trace was similar to the optical gating trace in the phase regions of the respiratory cycle used for gating. A cross correlation analysis and comparison of the external optical marker gating with internal fluoroscopic gating was performed. The fluoroscopic image correlation surrogate was found to be superior to the external optical surrogate in the AP-views in four out of six cases. In one of the remaining two cases, the two surrogates performed comparably, while in the last case, the external fiducial trace performed best. It was concluded that fluoroscopic gating based on correlation of native image features in the fluoroscopic images will be adequate for respiratory gating.

Evaluation of respiration‐correlated digital tomosynthesis in lunga)

Digital tomosynthesis (DTS) with a linear accelerator-mounted imaging system provides a means of reconstructing tomographic images from radiographic projections over a limited gantry arc, thus requiring only a few seconds to acquire. Its application in the thorax, however, often results in blurred images from respiration-induced motion. This work evaluates the feasibility of respiration-correlated (RC) DTS for soft-tissue visualization and patient positioning. Image data acquired with a gantry-mounted kilovoltage imaging system while recording respiration were retrospectively analyzed from patients receiving radiotherapy for non-small-cell lung carcinoma. Projection images spanning an approximately 30 degrees gantry arc were sorted into four respiration phase bins prior to DTS reconstruction, which uses a backprojection, followed by a procedure to suppress structures above and below the reconstruction plane of interest. The DTS images were reconstructed in planes at different depths through the patient and normal to a user-selected angle close to the center of the arc. The localization accuracy of RC-DTS was assessed via a comparison with CBCT. Evaluation of RC-DTS in eight tumors shows visible reduction in image blur caused by the respiratory motion. It also allows the visualization of tumor motion extent. The best image quality is achieved at the end-exhalation phase of the respiratory motion. Comparison of RC-DTS with respiration-correlated cone-beam CT in determining tumor position, motion extent and displacement between treatment sessions shows agreement in most cases within 2-3 mm, comparable in magnitude to the intraobserver repeatability of the measurement. These results suggest the methods applicability for soft-tissue image guidance in lung, but must be confirmed with further studies in larger numbers of patients.

Respiratory gating of sliding window IMRT

Breathing motion may limit sites of applicability for intensity modulated radiation therapy (IMRT). Respiratory gating reduces motion by activating the beam during a chosen phase of respiration. We investigated the effect of gating on the dose distribution of sliding window dynamic multileaf collimator (DMLC) fields. A Varian 2100C beam was gated by the Varian real time position management system that tracks moving infrared reflective markers with a wall-mounted camera. A rotating eccentric wheel simulated marker breathing motion. Dose distributions from normal (ungated) and gated DMLC fields were compared using film dosimetry and also by calculating dose distributions from the leaf motion log files. Fields included prostate (15 MV) and H&N (6 MV) treatment fields and special test fields. Kodak V films under polystyrene phantom were exposed separately for gated and ungated operation. Additional ungated fields were filmed as controls. Log files for gated and ungated deliveries were copied to a program written in-house which uses the clinical treatment planning system to calculate the dose distribution delivered by the recorded leaf motion. The films were digitized, optical densities converted to dose, and matched pairs of films were registered. The largest differences (/spl ap/5% or /spl sim/1 mm) occurred in high dose gradient regions and were consistent with uncertainties in aligning pairs of films. Gated vs. ungated dose distributions calculated from the fog files agreed to better than 0.5%. These preliminary results show no deleterious effects of gating on DMLC dose distributions.

Subsecond and Submillimeter Resolution Positional Verification for Stereotactic Irradiation of Spinal Lesions

PURPOSE Spine stereotactic body radiation therapy (SBRT) requires highly accurate positioning. We report our experience with markerless template matching and triangulation of kilovoltage images routinely acquired during spine SBRT, to determine spine position. METHODS AND MATERIALS Kilovoltage images, continuously acquired at 7, 11 or 15 frames/s during volumetric modulated spine SBRT of 18 patients, consisting of 93 fluoroscopy datasets (1 dataset/arc), were analyzed off-line. Four patients were immobilized in a head/neck mask, 14 had no immobilization. Two-dimensional (2D) templates were created for each gantry angle from planning computed tomography data and registered to prefiltered kilovoltage images to determine 2D shifts between actual and planned spine position. Registrations were considered valid if the normalized cross correlation score was ≥0.15. Multiple registrations were triangulated to determine 3D position. For each spine position dataset, average positional offset and standard deviation were calculated. To verify the accuracy and precision of the technique, mean positional offset and standard deviation for twenty stationary phantom datasets with different baseline shifts were measured. RESULTS For the phantom, average standard deviations were 0.18 mm for left-right (LR), 0.17 mm for superior-inferior (SI), and 0.23 mm for the anterior-posterior (AP) direction. Maximum difference in average detected and applied shift was 0.09 mm. For the 93 clinical datasets, the percentage of valid matched frames was, on average, 90.7% (range: 49.9-96.1%) per dataset. Average standard deviations for all datasets were 0.28, 0.19, and 0.28 mm for LR, SI, and AP, respectively. Spine position offsets were, on average, -0.05 (range: -1.58 to 2.18), -0.04 (range: -3.56 to 0.82), and -0.03 mm (range: -1.16 to 1.51), respectively. Average positional deviation was <1 mm in all directions in 92% of the arcs. CONCLUSIONS Template matching and triangulation using kilovoltage images acquired during irradiation allows spine position detection with submillimeter accuracy at subsecond intervals. Although the majority of patients were not immobilized, most vertebrae were stable at the sub-mm level during spine SBRT delivery.

SU‐FF‐I‐04: A Fast Variable‐Intensity Ring Suppression Algorithm

Purpose: Gain drifts and nonlinearities in amorphous silicon flat‐panel x‐ray detectors can produce ring artifacts in reconstructed cone‐beam computed tomography(CBCT)images. We have found that the magnitude of these artifacts can exceed 50 HU in clinical situations, and that the intensity of a given ring may not be uniform throughout an image. In some cases (e.g. half‐fan pelvic scans), discrete arcs may be produced. The goal of this study was to develop a post‐processing algorithm to efficiently suppress such variable‐intensity rings in axial slices. Method and Materials: Our approach builds upon the work of Sijbers and Postnov who showed that constant‐intensity rings can be estimated via radial median filtering of the input image after its transformation to polar coordinates. To characterize variable‐intensity rings and arcs, we developed a 2‐D estimation technique that uses a combination of row‐based (radial) and column‐based (angular) filters operating in the polar domain. The 2‐D estimates were transformed back to Cartesian space for subtraction from the original image. The new algorithm was implemented in C++ and tested on clinical and phantom CBCTimages acquired using a Varian 4030CB detector.Results: Correction times (3.2GHz Intel Pentium4 processor), including coordinate transformations, averaged 55 msec/slice for 512×512 matrix sizes. Rings and arcs were reduced in intensity by more than an order of magnitude to levels well below the background noise intensity. By subtracting ring estimates in Cartesian space, the polar matrix size could be reduced without sacrificing spatial resolution in the final image. This permitted for a 4× reduction in execution time compared to the original Sijbers‐Postnov approach where subtraction occurs in polar space. Conclusion: The Sijbers‐Postnov algorithm ring suppression algorithm was modified to provide improved image quality and fast execution times suitable for clinical implementation. Conflict of Interest: Funding provided by Varian Medical Systems.

Automatic tracking of arbitrarily shaped implanted markers in kilovoltage projection images: A feasibility study

PURPOSE Certain types of commonly used fiducial markers take on irregular shapes upon implantation in soft tissue. This poses a challenge for methods that assume a predefined shape of markers when automatically tracking such markers in kilovoltage (kV) radiographs. The authors have developed a method of automatically tracking regularly and irregularly shaped markers using kV projection images and assessed its potential for detecting intrafractional target motion during rotational treatment. METHODS Template-based matching used a normalized cross-correlation with simplex minimization. Templates were created from computed tomography (CT) images for phantom studies and from end-expiration breath-hold planning CT for patient studies. The kV images were processed using a Sobel filter to enhance marker visibility. To correct for changes in intermarker relative positions between simulation and treatment that can introduce errors in automatic matching, marker offsets in three dimensions were manually determined from an approximately orthogonal pair of kV images. Two studies in anthropomorphic phantom were carried out, one using a gold cylindrical marker representing regular shape, another using a Visicoil marker representing irregular shape. Automatic matching of templates to cone beam CT (CBCT) projection images was performed to known marker positions in phantom. In patient data, automatic matching was compared to manual matching as an approximate ground truth. Positional discrepancy between automatic and manual matching of less than 2 mm was assumed as the criterion for successful tracking. Tracking success rates were examined in kV projection images from 22 CBCT scans of four pancreas, six gastroesophageal junction, and one lung cancer patients. Each patient had at least one irregularly shaped radiopaque marker implanted in or near the tumor. In addition, automatic tracking was tested in intrafraction kV images of three lung cancer patients with irregularly shaped markers during 11 volumetric modulated arc treatments. Purpose-built software developed at our institution was used to create marker templates and track the markers embedded in kV images. RESULTS Phantom studies showed mean ± standard deviation measurement uncertainty of automatic registration to be 0.14 ± 0.07 mm and 0.17 ± 0.08 mm for Visicoil and gold cylindrical markers, respectively. The mean success rate of automatic tracking with CBCT projections (11 frames per second, fps) of pancreas, gastroesophageal junction, and lung cancer patients was 100%, 99.1% (range 98%-100%), and 100%, respectively. With intrafraction images (approx. 0.2 fps) of lung cancer patients, the success rate was 98.2% (range 97%-100%), and 94.3% (range 93%-97%) using templates from 1.25 mm and 2.5 mm slice spacing CT scans, respectively. Correction of intermarker relative position was found to improve the success rate in two out of eight patients analyzed. CONCLUSIONS The proposed method can track arbitrary marker shapes in kV images using templates generated from a breath-hold CT acquired at simulation. The studies indicate its feasibility for tracking tumor motion during rotational treatment. Investigation of the causes of misregistration suggests that its rate of incidence can be reduced with higher frequency of image acquisition, templates made from smaller CT slice spacing, and correction of changes in intermarker relative positions when they occur.

Detection and localization of radiotherapy targets by template matching

Radio opaque fiducials are implanted in tumors for the purpose of tracking the target motion using X-ray projections during radiation therapy dose delivery. In this paper we describe and evaluate a novel method based on template matching for detection and localization of arbitrary shaped fiducials. Segmentation methods are not adequate for these fiducials because their appearance in online X-ray projections can vary greatly as a function of imaging angle. The algorithm is based on using the planning CT image to generate templates that correspond to the imaging angles of the online images. We demonstrate successful tracking of complex shape fiducials in clinical images of lung and abdomen. We also validate the algorithm by comparing the results with a segmentation approach for one case in which the fiducials could be tracked by both methods. We also show how by adaptive thresholding of the match scores, we can control the false detection rate.

Digital tomosynthesis (DTS) for verification of target position in early stage lung cancer patients.

PURPOSE The ability to verify intrafraction tumor position is clinically useful for hypofractionated treatments. Short arc kV digital tomosynthesis (DTS) could facilitate more frequent target verification. The authors used DTS combined with triangulation to determine the mean temporal position of small-volume lung tumor targets treated with stereotactic radiotherapy. DTS registration results were benchmarked against online clinical localization using registration between free-breathing cone-beam computed tomography (CBCT) and the average intensity projection (AvIP) of the planning 4DCT. METHODS In this retrospective study, 76 sets of kV-projection images from online CBCT scans of 13 patients were used to generate DTS image slices (CB-DTS) with nonclinical research software (DTS Toolkit, Varian Medical Systems). Three-dimensional tumor motion was 1.3-4 mm in six patients and 6.1-25.4 mm in seven patients on 4DCT (significant difference in the mean of the groups, P < 0.01). The 4DCT AvIP was used to digitally reconstruct the Reference-DTS. DTS registration and DTS registration combined with triangulation were investigated. Progressive shortening of total DTS arc lengths from 95° to 35° around 0° gantry position was evaluated for different scenarios: DTS registration using the entire arc; DTS registration plus triangulation using two nonoverlapping arcs; and for 55° and 45° total gantry rotation, DTS registration plus triangulation using two overlapping arcs. Finally, DTS registration plus triangulation performed at eight gantry angles, each separated by 45° was evaluated using full fan kV projection data for one patient with an immobile tumor and five patients with mobile tumors. RESULTS For DTS registration alone, shortening arc length did not influence accuracy in X- and Y-directions, but in Z-direction, mean deviations from online CBCT localization systematically increased for shorter arc length (P < 0.05). For example, using a 95° arc mean DTS-CBCT difference was 0.8 mm (1 SD = 0.6 mm) and for a 35° arc the mean was 2.4 mm (1 SD = 1.7 mm). DTS plus triangulation using nonoverlapping-arcs increased accuracy in Z-direction for tested arc lengths ≤55° (P < 0.01). Overlapping arcs increased accuracy in Y-direction for tumors with motion >4 mm (P < 0.02) but increased Z-direction accuracy was only observed with 55° total gantry rotation. The 95th percentile deviations with this overlapping technique in X-, Y-, and Z-directions were 1.3, 2.0, and 2.5 mm, respectively. For the five patients with mobile tumors where DTS + triangulation was performed with 45° intervals, the pooled deviation from online CBCT correction showed, for X-, Y-, and Z-directions, mean of 1.1 mm, standard deviations (SD) of 0.9, 1.0, and 0.9 mm, respectively. The mean + 2 SD was <3 mm for each direction. CONCLUSIONS Short-arc DTS verification of time averaged lung tumor position is feasible using free-breathing kV projection data and the AvIP of the 4DCT as a reference. Observed differences between DTS and online CBCT registration with AvIP were ≤3 mm (mean + 2 SD), however, the increased temporal resolution of DTS + triangulation also identified short period deviations from the average target position on the CBCT. Short-arc DTS appears promising for intrafraction tumor position monitoring during stereotactic lung radiotherapy delivered with a rotational technique.