Gary Virshup

Comprehensive analysis of proton range uncertainties related to patient stopping-power-ratio estimation using the stoichiometric calibration

The purpose of this study was to analyze factors affecting proton stopping-power-ratio (SPR) estimations and range uncertainties in proton therapy planning using the standard stoichiometric calibration. The SPR uncertainties were grouped into five categories according to their origins and then estimated based on previously published reports or measurements. For the first time, the impact of tissue composition variations on SPR estimation was assessed and the uncertainty estimates of each category were determined for low-density (lung), soft, and high-density (bone) tissues. A composite, 95th percentile water-equivalent-thickness uncertainty was calculated from multiple beam directions in 15 patients with various types of cancer undergoing proton therapy. The SPR uncertainties (1σ) were quite different (ranging from 1.6% to 5.0%) in different tissue groups, although the final combined uncertainty (95th percentile) for different treatment sites was fairly consistent at 3.0-3.4%, primarily because soft tissue is the dominant tissue type in the human body. The dominant contributing factor for uncertainties in soft tissues was the degeneracy of Hounsfield numbers in the presence of tissue composition variations. To reduce the overall uncertainties in SPR estimation, the use of dual-energy computed tomography is suggested. The values recommended in this study based on typical treatment sites and a small group of patients roughly agree with the commonly referenced value (3.5%) used for margin design. By using tissue-specific range uncertainties, one could estimate the beam-specific range margin by accounting for different types and amounts of tissues along a beam, which may allow for customization of range uncertainty for each beam direction.

Multiple-gain-ranging readout method to extend the dynamic range of amorphous silicon flat-panel imagers

The dynamic range of many flat panel imaging systems are fundamentally limited by the dynamic range of the charge amplifier and readout signal processing. We developed two new flat panel readout methods that achieve extended dynamic range by changing the read out charge amplifier feedback capacitance dynamically and on a real-time basis. In one method, the feedback capacitor is selected automatically by a level sensing circuit, pixel-by-pixel, based on its exposure level. Alternatively, capacitor selection is driven externally, such that each pixel is read out two (or more) times, each time with increased feedback capacitance. Both methods allow the acquisition of X-ray image data with a dynamic range approaching the fundamental limits of flat panel pixels. Data with an equivalent bit depth of better than 16 bits are made available for further image processing. Successful implementation of these methods requires careful matching of selectable capacitor values and switching thresholds, with the imager noise and sensitivity characteristics, to insure X-ray quantum limited operation over the whole extended dynamic range. Successful implementation also depends on the use of new calibration methods and image reconstruction algorithms, to insure artifact free rebuilding of linear image data by the downstream image processing systems. The multiple gain ranging flat panel readout method extends the utility of flat panel imagers and paves the way to new flat panel applications, such as cone beam CT. We believe that this method will provide a valuable extension to the clinical application of flat panel imagers.

Efficient scatter correction using asymmetric kernels

X-ray cone-beam (CB) projection data often contain high amounts of scattered radiation, which must be properly modeled in order to produce accurate computed tomography (CT) reconstructions. A well known correction technique is the scatter kernel superposition (SKS) method that involves deconvolving projection data with kernels derived from pencil beam-generated scatter point-spread functions. The method has the advantages of being practical and computationally efficient but can suffer from inaccuracies. We show that the accuracy of the SKS algorithm can be significantly improved by replacing the symmetric kernels that traditionally have been used with nonstationary asymmetric kernels. We also show these kernels can be well approximated by combinations of stationary kernels thus allowing for efficient implementation of convolution via FFT. To test the new algorithm, Monte Carlo simulations and phantom experiments were performed using a table-top system with geometry and components matching those of the Varian On-Board Imager (OBI). The results show that asymmetric kernels produced substantially improved scatter estimates. For large objects with scatter-to-primary ratios up to 2.0, scatter profiles were estimated to within 10% of measured values. With all corrections applied, including beam hardening and lag, the resulting accuracies of the CBCT reconstructions were within ±25 Hounsfield Units (±2.5%).

Mercuric iodide and lead iodide x-ray detectors for radiographic and fluoroscopic medical imaging

Mercuric iodide (HgI2) and lead iodide (PbI2) have been under development for several years as direct converter layers in digital x-ray imaging. Previous reports have covered the basic electrical and physical characteristics of these and several other materials. We earlier reported on 5cm x 5cm and 10cm x 10cm size imagers, direct digital radiography X-ray detectors, based on photoconductive polycrystalline mercuric iodide deposited on a flat panel thin film transistor (TFT) array, as having great potential for use in medical imaging, NDT, and security applications. This paper, presents results and comparison of both lead iodide and mercuric iodide imagers scaled up to 20cm x 25cm sizes. Both the mercuric iodide and lead iodide direct conversion layers are vacuum deposited onto TFT array by Physical Vapor Deposition (PVD). This process has been successfully scaled up to 20cm x 25cm -- the size required in common medical imaging applications. A TFT array with a pixel pitch of 127 microns was used for this imager. In addition to increasing detector size, more sophisticated, non-TFT based small area detectors were developed in order to improve analysis methods of the mercuric and lead iodide photoconductors. These small area detectors were evaluated in radiographic mode, continuous fluoroscopic mode and pulsed fluoroscopic mode. Mercuric iodide coating thickness ranging between 140 microns and 300 microns and lead iodide coating thickness ranging between 100 microns and 180 microns were tested using beams with energies between 40 kVp and 100 kVp, utilizing exposure ranges typical for both fluoroscopic and radiographic imaging. Diagnostic quality radiographic and fluoroscopic images have been generated at up to 15 frames per second. Mercuric iodide image lag appears adequate for fluoroscopic imaging. The longer image lag characteristics of lead iodide make it only suitable for radiographic imaging. For both material the MTF is determined primarily by the aperture and pitch of the TFT array (Nyquist frequency of ~3.93 mm-1 (127 micron pixel pitch).

Does kV-MV dual-energy computed tomography have an advantage in determining proton stopping power ratios in patients?

Conventional kilovoltage (kV) x-ray-based dual-energy CT (DECT) imaging using two different x-ray energy spectra is sensitive to image noise and beam hardening effects. The purpose of this study was to evaluate the theoretical advantage of the DECT method for determining proton stopping power ratios (SPRs) using a combination of kV and megavoltage (MV) x-ray energies. We investigated three representative x-ray energy pairs: 100 and 140 kVp comprised the kV-kV pair, 100 kVp and 1 MV comprised the kV-MV pair, and two 1 MV x-ray beams-one with and one without external filtration-comprised the MV-MV pair. The SPRs of 34 human tissues were determined using the DECT method with these three x-ray energy pairs. Small perturbations were introduced into the CT numbers and x-ray spectra used for the DECT calculation to simulate the effects of random noise and beam hardening. An error propagation analysis was performed on the DECT calculation algorithm to investigate the propagation of CT number uncertainty to final SPR estimation and to suggest the best x-ray energy combination. We found that the DECT method using each of the three beam pairs achieved similar accuracy in determining the SPRs of human tissues in ideal conditions. However, when CT number uncertainties and artifacts such as imaging noise and beam hardening effects were considered, the kV-MV DECT improved the accuracy of SPR estimation substantially over the kV-kV or MV-MV DECT methods. Furthermore, our error propagation analysis showed that the combination of 100 kVp and 1 MV beams was close to the optimal selection when using the DECT method to determine SPRs. Overall, the kV-MV combination makes the DECT method more robust in resolving the effective atomic numbers for biological tissues than the traditional kV-kV DECT method.

Improved properties of PbI2 x-ray imagers with tighter process control and using positive bias voltage

Vapor deposited lead iodide films show a wide range of physical attributes dependant upon fabrication conditions. High density is most readily achieved with films less than 100 μm. Thicker films, with lessening density, often show lower response (gain) as charge collection becomes less efficient. Lack of consistency in density throughout a deposition invariably leads to non-uniform electronic properties, which is challenging to both model and predict. To overcome this, tighter control of deposition parameters is required during the slow growth process (<10 μm/hour). Lead iodide films are characterized in forms of planar devices deposited onto conductive glass and active pixel arrays deposited onto a-Si TFT arrays1. Electronic properties (e.g. leakage current, gain) show little variation that can be traced to substrate choice. Films generally provide less than 100 pA/mm2 leakage current as they show saturation in gain (at approximate fields of 1 V/μm). We recently modified our readout electronics to accept positive bias. Using positive bias on the top electrode provides better charge collection for the lower mobility electrons and (despite process variability) better quality films can provide sensitivities greater than 6 μC/R*cm2, with only partial x-ray absorption, and show less than 20 pA/mm2 dark current.

Investigation into the optimal linear time-invariant lag correction for radar artifact removal

PURPOSE Detector lag, or residual signal, in amorphous silicon (a-Si) flat-panel (FP) detectors can cause significant shading artifacts in cone-beam computed tomography (CBCT) reconstructions. To date, most correction models have assumed a linear, time-invariant (LTI) model and lag is corrected by deconvolution with an impulse response function (IRF). However, there are many ways to determine the IRF. The purpose of this work is to better understand detector lag in the Varian 4030CB FP and to identify the IRF measurement technique that best removes the CBCT shading artifact. METHODS We investigated the linearity of lag in a Varian 4030CB a-Si FP operating in dynamic gain mode at 15 frames per second by examining the rising step-response function (RSRF) followed by the falling step-response function (FSRF) at ten incident exposures (0.5%-84% of a-Si FP saturation exposure). We implemented a multiexponential (N = 4) LTI model for lag correction and investigated the effects of various techniques for determining the IRF such as RSRF versus FSRF, exposure intensity, length of exposure, and spatial position. The resulting IRFs were applied to (1) the step-response projection data and (2) CBCT acquisitions of a large pelvic phantom and acrylic head phantom. For projection data, 1st and 50th frame lags were measured pre- and postcorrection. For the CBCT reconstructions, four pairs of ROIs were defined and the maximum and mean errors within each pair were calculated for the different exposures and step-response edge techniques. RESULTS A nonlinearity greater than 50% was observed in the FSRF data. A model calibrated with RSRF data resulted in overcorrection of FSRF data. Conversely, models calibrated with FSRF data applied to RSRF data resulted in undercorrection of the RSRF. Similar effects were seen when LTI models were applied to data collected at different incident exposures. Some spatial variation in lag was observed in the step-response data. For CBCT reconstructions, an average error range of 3-21 HU was observed when using IRFs from different techniques. For our phantoms and FP, the lowest average error occurred for the FSRF-based techniques at exposures of 1.6 or 3.4% a-Si FP saturation, depending on the phantom used. CONCLUSIONS The choice of step-response edge (RSRF versus FSRF) and exposure intensity for IRF calibration could leave large residual lag in the step-response data. For the CBCT reconstructions, IRFs derived from FSRF data at low exposure intensities (1.6 and 3.4%) best removed the CBCT shading artifact. Which IRF to use for lag correction could be selected based on the object size.

Improving accuracy of electron density measurement in the presence of metallic implants using orthovoltage computed tomography

The goal of this study was to evaluate the improvement in electron density measurement and metal artifact reduction using orthovoltage computed tomography (OVCT) imaging compared with conventional kilovoltage CT (KVCT). For this study, a bench-top system was constructed with adjustable x-ray tube voltage up to 320 kVp. A commercial tissue-characterization phantom loaded with inserts of various human tissue substitutes was imaged using 125 kVp (KVCT) and 320 kVp (OVCT) x rays. Stoichiometric calibration was performed for both KVCT and OVCT imaging using the Schneider method. The metal inserts-titanium rods and aluminum rods-were used to study the impact of metal artifacts on the electron-density measurements both inside and outside the metal inserts. It was found that the relationships between Hounsfield units and relative electron densities (to water) were more predictable for OVCT than KVCT. Unlike KVCT, the stoichiometric calibration for OVCT was insensitive to the use of tissue substitutes for direct electron density calibration. OVCT was found to significantly reduce metal streak artifacts. Errors in electron-density measurements within uniform tissue substitutes were reduced from 42% (maximum) and 18% (root-mean-square) in KVCT to 12% and 2% in OVCT, respectively. Improvements were also observed inside the metal implants. For the detectors optimized for KVCT, the imaging dose is almost doubled for OVCT for the image quality comparable to KVCT. OVCT may be a good option for high-precision radiotherapy treatment planning, especially for patients with metal implants and especially for charged particle therapy, such as proton therapy.

Comparison of scatter correction methods for CBCT

In contrast to the narrow fan of clinical Computed Tomography (CT) scanners, Cone Beam scanners irradiate a much larger proportion of the object, which causes additional X-ray scattering. The most obvious scatter artefact is that the middle area of the object becomes darker than the outer area, as the density in the middle of the object is underestimated (cupping). Methods for estimating scatter were investigated that can be applied to each single projection without requiring a preliminary reconstruction. Scatter reduction by the Uniform Scatter Fraction method was implemented in the Varian CBCT software version 2.0. This scatter correction method is recommended for full fan scans using air norm. However, this method did not sufficiently correct artefacts in half fan scans and was not sufficiently robust if used in combination with a Single Norm. Therefore, a physical scatter model was developed that estimates scatter for each projection using the attenuation profile of the object. This model relied on laboratory experiments in which scatter kernels were measured for Plexiglas plates of varying thicknesses. Preliminary results suggest that this kernel model may solve the shortcomings of the Uniform Scatter Fraction model.

A nonlinear lag correction algorithm for a‐Si flat‐panel x‐ray detectors

PURPOSE Detector lag, or residual signal, in a-Si flat-panel (FP) detectors can cause significant shading artifacts in cone-beam computed tomography reconstructions. To date, most correction models have assumed a linear, time-invariant (LTI) model and correct lag by deconvolution with an impulse response function (IRF). However, the lag correction is sensitive to both the exposure intensity and the technique used for determining the IRF. Even when the LTI correction that produces the minimum error is found, residual artifact remains. A new non-LTI method was developed to take into account the IRF measurement technique and exposure dependencies. METHODS First, a multiexponential (N = 4) LTI model was implemented for lag correction. Next, a non-LTI lag correction, known as the nonlinear consistent stored charge (NLCSC) method, was developed based on the LTI multiexponential method. It differs from other nonlinear lag correction algorithms in that it maintains a consistent estimate of the amount of charge stored in the FP and it does not require intimate knowledge of the semiconductor parameters specific to the FP. For the NLCSC method, all coefficients of the IRF are functions of exposure intensity. Another nonlinear lag correction method that only used an intensity weighting of the IRF was also compared. The correction algorithms were applied to step-response projection data and CT acquisitions of a large pelvic phantom and an acrylic head phantom. The authors collected rising and falling edge step-response data on a Varian 4030CB a-Si FP detector operating in dynamic gain mode at 15 fps at nine incident exposures (2.0%-92% of the detector saturation exposure). For projection data, 1st and 50th frame lag were measured before and after correction. For the CT reconstructions, five pairs of ROIs were defined and the maximum and mean signal differences within a pair were calculated for the different exposures and step-response edge techniques. RESULTS The LTI corrections left residual 1st and 50th frame lag up to 1.4% and 0.48%, while the NLCSC lag correction reduced 1st and 50th frame residual lags to less than 0.29% and 0.0052%. For CT reconstructions, the NLCSC lag correction gave an average error of 11 HU for the pelvic phantom and 3 HU for the head phantom, compared to 14-19 HU and 2-11 HU for the LTI corrections and 15 HU and 9 HU for the intensity weighted non-LTI algorithm. The maximum ROI error was always smallest for the NLCSC correction. The NLCSC correction was also superior to the intensity weighting algorithm. CONCLUSIONS The NLCSC lag algorithm corrected for the exposure dependence of lag, provided superior image improvement for the pelvic phantom reconstruction, and gave similar results to the best case LTI results for the head phantom. The blurred ring artifact that is left over in the LTI corrections was better removed by the NLCSC correction in all cases.