Search

Total variation minimization has been extensively researched for image denoising and sparse view reconstruction. These methods show superior denoising performance for simple images with little texture, but result in texture information loss when applied to more complex images. It could thus be beneficial to use other regularizers within medical imaging. We propose a general regularization method, based on a split-Bregman approach. We show results for this framework combined with a total variation denoising operator, in comparison to ASD-POCS. We show that sparse-view reconstruction and noise regularization is possible. This general method will allow us to investigate other regularizers in the context of regularized CT reconstruction, and decrease the acquisition times in µCT.

We studied the resolution limit that can be obtained for a whole body PET scanner. The results were obtained using a Monte Carlo based simulation program. The influence of two parameters was investigated: the crystal pixel size and the number of layers used for Depth-Of-Interaction (DOI) correction.

Objectives: PET Contrast Recovery depends on lesion’s background. Identical lesions in different backgrounds can recover differently due to different convergence and spill in/out effects. In this work we investigate the effects of Time-of-flight PET for identical lesions in different backgrounds in terms of both CRC and threshold based volume definition. Methods: A phantom with lesions and organ inserts simulating an obese patient at high count rate was acquired on a TOF-PET scanner. Identical lesions were placed both in hot and cold background. We performed listmode TOF and non-TOF reconstruction for varying scan times (30 and 60 s with 10 realizations each). We compared the performances of the two methods in measuring lesions volume by applying a 42% threshold method. CRC was calculated for all lesions in PET and TOF for a range of clinically used iterations. Results: TOF CRC converges faster and to higher values for all cold and hot lesions except one (23 mm), which converges 10 % higher for non-TOF. For this lesion the difference between TOF and non-TOF CRCs is less than 5 % at 3 iterations, clinically the most used on site. Identical lesions in different backgrounds converge to different values, the difference being 20% smaller for TOF. Volume thresholding for lesions in a cold background gives up to a 30% better volume estimation when using TOF. Conclusions: TOF PET shows better characteristics for quantization tasks in different backgrounds and volume definition for cold background lesions. Count and contrast recovery is less depending on surrounding activity than with regular PET.

xxx

Although there is increasing interest in the automatic processing of tomographic radionuclide ventriculography (TRV) studies, validation is mainly limited to a comparison of TRV results with data from planar radionuclide ventriculography (PRV) or gated perfusion. single photon emission computed tomography (SPECT). The aim of this study was to use a dynamic physical cardiac phantom to validate the ejection fraction (EF) and volumes from PRV and TRV studies. A new dynamic left ventricular phantom was constructed and used to obtain 21 acquisitions in the planar and tomographic mode. The directly measured volumes and EFs of the phantom during the acquisitions were considered as the gold standard for comparison with TRV and PRV. EFs were calculated from PRV by background-corrected end-diastolic and end-systolic frames. Volumes and EFs were calculated from TRV by region growing with different lower thresholds to search for the optimal threshold. EF from PRV correlated significantly with the real EF (r = 0.94, P = 0.00). The optimal threshold value for volume calculation from TRV in 336 cases was 50% (r = 0.98, P = 0.00) yielding the best slope after linear regression. When considering these calculated end-diastolic and end-systolic volumes, EF correlated well (r=0.99, P=0.00) with the real EF, and this correlation was significantly (P = 0.04) higher than that of the EF from PRV. Our experiments prove that EF measured by TRV yields more accurate results compared with PRV in dynamic cardiac phantom studies. (

Positron emission tomography (PET) is currently performed using either a dedicated PET scanner or scintillation gamma camera equipped with electronic circuitry for coincidence detection of 511 keV annihilation quanta (gamma camera PET system). Although the resolution limits of these two instruments are comparable, the sensitivity and count rate performance of the gamma camera PET system are several times lower than that of the PET scanner. Most gamma camera PET systems are manufactured as dual-detector systems capable of performing dual-head coincidence imaging. One possible step towards the improvement of the sensitivity of the gamma camera PET system is to add another detector head. This work investigates the characteristics of one such triple-head gamma camera PET system capable of performing triple-head coincidence imaging. The following performance characteristics of the system were assessed: spatial resolution, sensitivity, count rate performance. The spatial resolution, expressed as the full width at half-maximum (FWHM), at 1 cm radius is 5.9 mm ; at 10 cm radius, the transverse radial resolution is 5.3 mm, whilst the transverse tangential and axial resolutions are 8.9 mm and 13.3 mm, respectively. The sensitivity for a standard cylindrical phantom is 255 counts×s71×MBq71, using a 30% width photopeak energy window. An increase of 35% in the PET sensitivity is achievable by opening an additional 30% width energy window in the Compton region. The count rate in coincidence mode, at the upper limit of the systems optimal performance, is 45 kc×s71 (kc=kilocounts) using the photopeak energy window only, and increases to 60 kc×s71 using the photopeak + Compton windows. Sensitivity results are compared with published data for a similar dualhead detector system.

Positron emission tomography (PET) is currently performed using either dedicated PET scanner or gamma camera equipped with electronic circuitry for coincidence detection of 511 keV annihilation quanta (gamma camera PET system). While resolution limits of these two instruments are comparable, sensitivity and count rate performance of the gamma camera PET system is several times lower than that of the PET scanner. Most gamma camera PET systems are manufactured as dual-detector systems capable of performing dual-head coincidence imaging. This work investigates performance characteristics of triple-head gamma camera PET system and compares them with 2-head system. Following performance characteristics of the system were assessed: spatial resolution, sensitivity, count rate performance. Spatial resolution, expressed as full-width-at-half-maximum (FWHM), at 1 cm radius is 5.9 mm ; at 10 cm radius transverse radial resolution is 5.3 mm while transverse tangential and axial resolution are 8.9 mm and 13.3 mm respectively. Sensitivity for standard cylindrical phantom is 255 c×s-1×MBq-1, cşcounts, using 30%-width photopeak energy window. Increase of 35 % in the PET sensitivity is achievable by opening an additional 30%-width energy window in the Compton region. Count rate in coincidence mode, at the upper limit of system&#8217 ; s optimal performance, is 45 kc×s-1 using only photopeak energy window, and increases to 60 kc×s-1 when using photopeak + Compton windows. Resolution when imaging cold phantom objects (rods and spheres) is 10 mm. In conclusion, our triple-head gamma camera PET system has system sensitivity two times higher in comparison with a similar dual-head system. Spatial resolution is in the same range as in the most dual-head systems.