M4D2  Non-emission Imaging Methods

Thursday, Nov. 5  16:30-18:00  Pacific Salon 1&2

Session Chair:  Paul Vaska, Stony Brook University, ; Xiaochuan Pan, The University of Chicago, United States

Show/Hide All Abstracts

(16:30) M4D2-1, Imaging Performance of a Spectral Photon-Counting CT Prototype with Small Field of View

I. Blevis1, Y. Younes1, A. Livne1, M. Rokni1, R. Levinson1, Y. Berman1, E. Roessl2, M. Bartels2, B. Brendel2, H. Daerr2, A. Thran2, C. Herrmann3, R. Steadman3, K.-J. Engels3, R. Proksa2, A. Altman1, O. Zarchin1

1Global Research and Advanced Development, Philips Healthcare CT, Haifa, Israel
2Philips Research Laboratories, Hamburg, Germany
3BG Imaging Systems, Philips Healthcare, Cleveland, OH, U.S.A.
Objective: We report on the imaging performance measured with phantoms for a research prototype for energy-resolving, photon-counting computed tomography. Method: The prototype is based on an assembly of CZT direct conversion semi-conductor detectors and custom electronics adapted and assembled into a medical Philips iCT gantry. Each pixel of the detection system operates in single-photon-counting energy-discrimination mode sorting the counts into several energy-bins with separate counters. This facilitates HU-imaging, two-material discrimination, mono-chromatic imaging as well as K-edge imaging for a wide range of atomic numbers. The full detector size provides a reconstruction transaxial field-of-view of about 16 cm and a z-coverage of about 2.5 mm, measured in the scanner’s iso-center. Results: We present preliminary imaging results at tube currents up to 100 mA. The imaging performance will be described in terms of spatial resolution, HU-image quality, material discrimination capabilities and the linearity of the response to various dilutions of iodine and gadolinium-based contrast materials. Conclusions: The ability to perform conventional attenuation based imaging as well as energy based imaging including multiple materials, and materials with a K-edge signature has been realized in a CZT based Spectral Photon Counting CT.

(16:45) M4D2-2, Experimental Evaluation of a Combined X-Ray Fluorescence Emission Tomography and X-Ray Luminescent Computed Tomography System Towards Quantifying and Imaging Therapeutic Nanoparticles

J. George1, D. Strat1, B. Quigley2, P. La Rivičre2, L.-J. Meng2

1Nuclear, Plasma, and Radiological Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States
2Department of Radiology, University of Chicago, Chicago, IL, United States
In recent years, new X-ray-mediated imaging methods, such as x-ray fluorescence computed tomography (XFCT) and X-ray luminescence computed tomography (XLCT), have garnered much attention for their potential as diagnostic modalities and for monitoring therapeutic delivery dosage and effects . In both XLCT and XFCT, the irradiation and stimulation of imaging agents by externally applied incident X-rays produces detectable secondary photons. For XLCT, this secondary signal is visible light produced by upconversion of lanthanide-based nanophosphors. When incident energy is received by the nanophosphor, either from an x-ray or a secondary electron, an orbital electron is raised to an excited state, and the subsequent decay to ground state releases energy as photons. XFCT, in contrast, relies on ionization of inner-shell electrons, whose replacement results in emission of material specific characteristic X-rays. XFCT and XLCT are largely complementary. The high quantum yield of XL allows for the production of several visible photons from a single incident X-ray. This allows for improved sensitivity and resolution at shallow depths. By contrast, XFCT offers more tissue penetration and could be used to image deeper tissues . However, the quantum yield for fluorescence is significantly smaller as only a single fluorescent photon can be achieved per incident ionizing photon. With this in mind, this proposed work explores a dual modality system to allow both XFET and XLCT imaging of the same sample. With this setup, we will compare the sensitivity and image resolution as a function of depth in 3D-printed nanoparticle-filled phantoms. We will also study the possibility of using both X-ray fluorescence and X-ray luminescence signals from the same sample to create a 3-D image of the nanophosphor that emit both fluorescence X-rays (for XFCT) and visible photons (for XLCT). In addition, results obtained with polychromatic and monochromatic X-ray sources will be discussed.

(17:00) M4D2-3, Photon Counting Systems for Breast Imaging

W. C. Barber1,2, J. C. Wessel1,2, N. Malakhov2, G. Wawrzyniak2, N. E. Hartsough1, E. Naess-Ulseth2, J. S. Iwanczyk1

1DxRay Inc., Nortridge, CA, USA
2Interon AS, Asker, Norway
We have developed silicon (Si) strip photon counting detectors for modalities used in clinical breast imaging including mammography, tomosynthesis, computed tomography, and lump imaging in the operating room. Typically, x-ray integrating detectors based on scintillating cesium iodide CsI(Tl) or amorphous selenium (a-Se) are used in most commercial systems. Recently, mammography instrumentation has been introduced based on photon counting silicon Si strip detectors with two adjustable energy discriminators for dual energy data acquisition with two contiguous energy windows. The systems presented here use four discriminators for up to four contiguous energy windows or two energy windows with separated upper and lower bounds. The high resolution detectors with 100 micron pixels produce a maximum output count rate of 100 Mcps/mm2 which is linear up to 40 Mcps/mm2 and a FWHM energy resolution of 2 keV which is uniform from the noise floor of 4 keV to the upper bound on the dynamic range at 60 keV. The Si strip detectors are packaged into modules which are tiled in different geometries for the different modalities used in breast imaging. We describe the detector optimization and the development of application specific integrated readout electronics that provide the required spatial resolution, low noise, high count rate, and dynamic range for each modality. We present results from prototype mammography, computed tomography, and lump imaging systems as well as describe methods for the absolute quantification of iodine concentration.

(17:15) M4D2-4, Methods for Tissue and Iodine Quantification in Spectral Mammography

Y. Pavia1,2, A. Brambilla1, V. Rebuffel1, J. M. Letang2, N. Freud2, L. Verger1

1CEA LETI, MINATEC Campus, F-38054 Grenoble, France
2Universite de Lyon, CREATIS ; CNRS UMR5220 ; Inserm U1044 ; INSA-Lyon ;, Universite Claude Bernard Lyon 1, Centre Leon Berard, France
Breast density can be estimated from tissue decomposition using material-base methods. Emerging detectors for medical applications, like energy sensitive photon counting detectors are able to estimate tissue thicknesses with a single X-ray exposure. Nevertheless, these detectors must be associated with efficient methods. This study compares two main approaches: one based on polynomials (determined from a calibration-base), and another one based on Poisson-likelihood (comparison with the calibration-base). The proposed work aimed to benchmark several polynomials of second and third orders, with and without full terms, against the Poisson-likelihood method. A realistic simulation tool, including the energetic detector response matrix (DRM) and Poisson noise has especially been developed for this purpose, based on a scanning-slit system to reduce scattering effects in mammographic conditions (60 µm pixel pitch and low energy spectrum). Breast tissues were mimicked by PMMA and water (for adipose and glandular). The different polynomial methods and the Poisson-likelihood method were also assessed for a 3-material base decomposition, in the presence of iodine in the case of contrast enhanced k-edge imaging. Results show that third order polynomials allow good thickness estimations, even when iodine is involved. Moreover, the proposed third order polynomial slightly shifts the bias-noise compromise but can increase iodine quantification by 16%. The Poisson log-likelihood approach shows valuable results and can still be improved to optimize the compromise between noise and bias. Further, this method presents the advantage of being easily extended to more than three energy bins for spectroscopic detectors and could allow better quantification results.

(17:30) M4D2-5, Scatter and Attenuation Effects in X-ray Luminescence Optical Tomography

A. Martinez-Davalos1, S. Rosas-Gonzalez2, T. Bautista-Torres1, M. Rodriguez-Villafuerte1

1Instituto de Fisica, UNAM, D.F., Mexico
2Instituto Nacional de Neurologia y Neurocirugia, D.F., Mexico
X-ray luminescence optical tomography (XLOT) has been proposed to study problems related to deep-tissue small-animal imaging. In this technique luminescent nanoparticles emit optical photons when irradiated with a collimated x-ray beam. The use of this modality for small-animal imaging requires an accurate knowledge of the energy deposition map inside the subject for optimization of the optical imaging model used in the tomographic reconstruction. It is of particular interest to determine the contribution of scattered radiation to the luminescent signal, since this might limit the spatial and contrast resolution of the system. In this work we report the use of Monte Carlo simulation to study attenuation and scatter effects in mouse-size phantoms with embedded Gd2O2S inserts using a W target x-ray tube in the range of 30-90 kVp with an added 1.0 mm Al filtration. The results show that the scatter contribution is of the order of 25% of the total dose to the insert and that it scales linearly with kVp for a fixed concentration (1 mg/ml) of luminescent nanoparticles at a fixed air-kerma rate. The imaging performance of the system was evaluated by means of simulations of the NEMA NU4 image quality and micro-Derenzo phantoms. The results show that quantification of the luminescent particle concentration deteriorates with object size, up to 80% when going from 5 to 1 mm diameter objects at 1 mg/ml concentration. The optical spatial resolution for 1 mm step size and 10 degrees angular step is of the order of 1.25 mm if an iterative reconstruction algorithm is used to obtain the tomographic images.

(17:45) M4D2-6, Dual-Energy C-Arm CT in the Angiographic Suite

S. Datta1, J.-H. Choi2, C. Niebler3, A. Maier4, R. Fahrig2, K. Müller2

1Siemens Healthcare, Forchheim, Germany
2Dept. of Radiology, Stanford University, Stanford, CA, USA
3Dept. of Electrical Engineering, Technische Hochschule Nürnberg, Nürnberg, Germany
4Dept. of Computer Science, Friedrich-Alexander-University, Erlangen-Nürnberg, Germany
PURPOSE: Dual-energy CT techniques have shown tremendous clinical value thanks to their ability to distinguish materials based on atomic number. C-arm CT is currently used to guide interventional procedures but there are no commercially available systems that employ dual-energy material decomposition. This paper explores the feasibility of implementing a fast kV-switching technique to perform dual-energy C-arm CT on a clinical angiography system. METHODS: A software prototype on a Siemens Artis zeego C-arm angiography system was used to prescribe tube output parameters (voltage, current, and pulse width) on a per-projection basis during rotational acquisitions. As an initial proof of concept, energies of 90 kV and 125 kV were selected for all experiments. The peak energy produced during each pulse of kV-constant scans was measured using a kV sensor and compared to that of kV-switching scans, in which frames alternated between 90 kV and 125 kV. Next, a phantom containing iodine and water was scanned using the same protocols. The contrast-to-noise ratio in these resulting reconstructions (each using projections from a single nominal energy) measured the contrast penalty resulting from applying the kV-switching technique. RESULTS: During rapid kV-switching acquisitions, the energy produced by the tube at each pulse is up to 5% lower than the energy produced during kV-constant acquisitions. The instability of the produced kV, measured as the standard deviation of the kVp produced in each pulse, is up to 4 times higher for kV-switching acquisitions. However, these deficits did not consistently degrade the contrast resolution of 3D reconstructions. CONCLUSIONS: A clinical C-arm system is capable of performing dual-energy acquisitions using fast kV switching without significant degradation in contrast. Future work will better characterize the system behavior across the full range of energies and optimize the technique for clinically-relevant material decomposition.