Few-view CT reconstruction via a novel non-local means algorithm

PurposeNon-local means (NLM) based reconstruction method is a promising algorithm for few-view computed tomography (CT) reconstruction, but often suffers from over-smoothed image edges. To address this problem, an adaptive NLM reconstruction method based on rotational invariance (ART-RIANLM) is proposed.MethodsThe method consists of four steps: 1) Initializing parameters; 2) ART reconstruction using raw data; 3) Positivity constraint of the reconstructed image; 4) Image updating by RIANLM filtering. In RIANLM, two kinds of rotational invariance measures which are average gradient (AG) and region homogeneity (RH) are proposed to calculate the distance between two patches and a novel NLM filter is developed to avoid over-smoothed image. Moreover, the parameter h in RIANLM which controls the decay of the weights is adaptive to avoid over-smoothness, while it is constant in NLM during the whole reconstruction process. The proposed method is validated on two digital phantoms and real projection data.ResultsIn our experiments, the searching neighborhood size is set as 15 × 15 and the similarity window is set as 3 × 3. For the simulated case of Shepp-Logan phantom, ART-RIANLM produces higher SNR (36.23 dB > 24.00 dB) and lower MAE (0.0006 < 0.0024) reconstructed images than ART-NLM. The visual inspection demonstrated that the proposed method could suppress artifacts or noises more effectively and recover image edges better. The result of real data case is also consistent with the simulation result.ConclusionsA RIANLM based reconstruction method for few-view CT is presented. Compared to the traditional ART-NLM method, SNR and MAE from ART-RIANLM increases 51% and decreases 75%, respectively.     

Evaluation of a commercial Model Based Iterative reconstruction algorithm in computed tomography

IntroductionIterative reconstruction algorithms have been introduced in clinical practice to obtain dose reduction without compromising the diagnostic performance.PurposeTo investigate the commercial Model Based IMR algorithm by means of patient dose and image quality, with standard Fourier and alternative metrics.Materials and methodsA Catphan phantom, a commercial density phantom and a cylindrical water filled phantom were scanned both varying CTDI_vol and reconstruction thickness. Images were then reconstructed with Filtered Back Projection and both statistical (iDose) and Model Based (IMR) Iterative reconstruction algorithms.Spatial resolution was evaluated with Modulation Transfer Function and Target Transfer Function. Noise reduction was investigated with Standard Deviation. Furthermore, its behaviour was analysed with 3D and 2D Noise Power Spectrum. Blur and Low Contrast Detectability were investigated.Patient dose indexes were collected and analysed.ResultsAll results, related to image quality, have been compared to FBP standard reconstructions.Model Based IMR significantly improves Modulation Transfer Function with an increase between 12% and 64%. Target Transfer Function curves confirm this trend for high density objects, while Blur presents a sharpness reduction for low density details.Model Based IMR underlines a noise reduction between 44% and 66% and a variation in noise power spectrum behaviour. Low Contrast Detectability curves underline an averaged improvement of 35–45%; these results are compatible with an achievable reduction of 50% of CTDI_vol.A dose reduction between 25% and 35% is confirmed by median values of CTDI_vol.ConclusionIMR produces an improvement in image quality and dose reduction.     

Internal algorithm variability and among-algorithm discordance in statistical haplotype reconstruction

The potential effectiveness of statistical haplotype inference makes it an area of active exploration over the last decade. There are several complications of statistical inference, including: the same algorithm can produce different solutions for the same data set, which reflects the internal algorithm variability; different algorithms can give different solutions for the same data set, reflecting the discordance among algorithms; and the algorithms per se are unable to evaluate the reliability of the solutions even if they are unique, this being a general limitation of all inference methods. With the aim of increasing the confidence of statistical inference results, consensus strategy appears to be an effective means to deal with these problems. Several authors have explored this with different emphases. Here we discuss two recent studies examining the internal algorithm variability and among-algorithm discordance, respectively, and evaluate the different outcomes of these analyses, in light of Orzack (2009) comment. Until other, better methods are developed, a combination of these two approaches should provide a practical way to increase the confidence of statistical haplotyping results.     

Pin-photodiode array for the measurement of fan-beam energy and air kerma distributions of X-ray CT scanners

PurposePatient dose estimation in X-ray computed tomography (CT) is generally performed by Monte Carlo simulation of photon interactions within anthropomorphic or cylindrical phantoms. An accurate Monte Carlo simulation requires an understanding of the effects of the bow-tie filter equipped in a CT scanner, i.e. the change of X-ray energy and air kerma along the fan-beam arc of the CT scanner. To measure the effective energy and air kerma distributions, we devised a pin-photodiode array utilizing eight channels of X-ray sensors arranged at regular intervals along the fan-beam arc of the CT scanner.MethodsEach X-ray sensor consisted of two plate type of pin silicon photodiodes in tandem – front and rear photodiodes – and of a lead collimator, which only allowed X-rays to impinge vertically to the silicon surface of the photodiodes. The effective energy of the X-rays was calculated from the ratio of the output voltages of the photodiodes and the dose was calculated from the output voltage of the front photodiode using the energy and dose calibration curves respectively.ResultsThe pin-photodiode array allowed the calculation of X-ray effective energies and relative doses, at eight points simultaneously along the fan-beam arc of a CT scanner during a single rotation of the scanner.ConclusionsThe fan-beam energy and air kerma distributions of CT scanners can be effectively measured using this pin-photodiode array.     

Measurement-based model of a wide-bore CT scanner for Monte Carlo dosimetric calculations with GMCTdospp software

The aim of this work was to create a model of a wide-bore Siemens Somatom Sensation Open CT scanner for use with GMCTdospp, which is an EGSnrc-based software tool dedicated for Monte Carlo calculations of dose in CT examinations.The method was based on matching spectrum and filtration to half value layer and dose profile, and thus was similar to the method of Turner et al. (Med. Phys. 36, pp. 2154–2164). Input data on unfiltered beam spectra were taken from two sources: the TASMIP model and IPEM Report 78. Two sources of HVL data were also used, namely measurements and documentation. Dose profile along the fan-beam was measured with Gafchromic RTQA-1010 (QA+) film. Two-component model of filtration was assumed: bow-tie filter made of aluminum with 0.5 mm thickness on central axis, and flat filter made of one of four materials: aluminum, graphite, lead, or titanium.Good agreement between calculations and measurements was obtained for models based on the measured values of HVL. Doses calculated with GMCTdospp differed from the doses measured with pencil ion chamber placed in PMMA phantom by less than 5%, and root mean square difference for four tube potentials and three positions in the phantom did not exceed 2.5%. The differences for models based on HVL values from documentation exceeded 10%. Models based on TASMIP spectra and IPEM78 spectra performed equally well.     

Image quality for radiotherapy CT simulators with different scanner bore size

PurposeWe compare image quality parameters derived from phantom images taken on three commercially available radiotherapy CT simulators. To make an unbiased evaluation, we assured images were obtained with the same surface dose measured using XR-QA2 model GafChromic™ film placed at the imaging phantom surface for all three CT-simulators.MethodsRadiotherapy CT simulators GE LS 16, Philips Brilliance Big Bore, and Toshiba Aquilion LB were compared in terms of spatial resolution, low contrast detectability, image uniformity, and contrast to noise ratio using CATPHAN-504 phantom, scanned with Head and Pelvis protocols. Dose was measured at phantom surface, with CT scans repeated until doses on all scanners were within 2%.ResultsIn terms of spatial resolution, the GE simulator appears slightly better, while Philips CT images are superior in terms of SNR for both scanning protocols. The CNR results show that Philips CT images appear to be better, except for high Z material, while Toshiba appears to fit in between the two simulators.ConclusionsWhile the image quality parameters for three RT CT simulators show comparable results, the scanner bore size is of vital importance in various radiotherapy applications. Since the image quality is a function of a large number of confounding parameters, any loss in image quality due to scanner bore size could be compensated by the appropriate choice of scanning parameters, including the exposure and by balancing between the additional imaging dose to the patient and high image quality required in highly conformal RT techniques.     

Image quality assessment of an iterative reconstruction algorithm applied to abdominal CT imaging

PurposeTo compare the noise and accuracy on images of the whole porcine liver acquired with iterative reconstruction (IMR, Philips Healthcare, Cleveland, OH, USA) and filtered back projection (FBP) methods.Materials and methodsWe used non-enhanced porcine liver to simulate the human liver and acquired it 100 times to obtain the average FBP value as the ground-truth. The mean and the standard deviation (“inter-scan SD”) of the pixel values on the 100 image acquisitions were calculated for FBP and for three levels of IMR (L1, L2, and L3). We also calculated the noise power spectrum (NPS) and the normalized NPS for the 100 image acquisitions.ResultsThe spatial SD for the porcine liver parenchyma on these slices was 9.92, 4.37, 3.63, and 2.30 Hounsfield units with FBP, IMR-L1, IMR-L2, and IMR-L3, respectively. The detectability of small faint features was better on single IMR than single FBP images. The inter-scan SD value for IMR-L3 images was 53% larger at the liver edges than at the liver parenchyma; it was only 10% larger on FBP images. Assessment of the normalized NPS showed that the noise on IMR images was comprised primarily of low-frequency components.ConclusionIMR images yield the same structure informations as FBP images and image accuracy is maintained. On level 3 IMR images the image noise is more strongly suppressed than on IMR images of the other levels and on FBP images.     

Interpreting the flock algorithm from a statistical perspective

We show that the algorithm in the program flock (Duchesne & Turgeon 2009) can be interpreted as an estimation procedure based on a model essentially identical to the structure (Pritchard et al. 2000) model with no admixture and without correlated allele frequency priors. Rather than using MCMC, the flock algorithm searches for the maximum a posteriori estimate of this structure model via a simulated annealing algorithm with a rapid cooling schedule (namely, the exponent on the objective function →∞). We demonstrate the similarities between the two programs in a two‐step approach. First, to enable rapid batch processing of many simulated data sets, we modified the source code of structure to use the flock algorithm, producing the program flockture . With simulated data, we confirmed that results obtained with flock and flockture are very similar (though flockture is some 200 times faster). Second, we simulated multiple large data sets under varying levels of population differentiation for both microsatellite and SNP genotypes. We analysed them with flockture and structure and assessed each program on its ability to cluster individuals to their correct subpopulation. We show that flockture yields results similar to structure albeit with greater variability from run to run. flockture did perform better than structure when genotypes were composed of SNPs and differentiation was moderate ( $F_{\mathrm{ST}}=$ 0.022–0.032). When differentiation was low, structure outperformed flockture for both marker types. On large data sets like those we simulated, it appears that flock 's reliance on inference rules regarding its ‘plateau record’ is not helpful. Interpreting flock 's algorithm as a special case of the model in structure should aid in understanding the program's output and behaviour.     

An algorithm for abdominal wall reconstruction

Acquired abdominal wall defects result from trauma, previous surgery, infection, and tumor resection. The correction of complex defects is a challenge to both plastic and reconstructive and general surgeons. The anatomy of the abdominal wall, as well as considerations in patient assessment and surgical planning, are discussed. A simple classification of abdominal wall defects based on size, depth, and location is provided. Publications regarding the various abdominal reconstruction techniques are reviewed and summarized to familiarize the reader with the treatment options for each particular defect. Finally, an algorithm is presented to guide the surgeon in selecting the optimal reconstructive technique.     

Conditional generative adversarial networks to generate pseudo low monoenergetic CT image from a single-tube voltage CT scanner

PurposeTo generate pseudo low monoenergetic CT images of the abdomen from 120-kVp CT images with cGAN.Materials and MethodsWe retrospectively included 48 patients who underwent contrast-enhanced abdominal CT using dual-energy CT. We reconstructed paired data sets of 120 kVp CT images and virtual low monoenergetic (55-keV) CT images. cGAN was prepared to generate pseudo 55-keV CT images from 120-kVp CT images. The pseudo 55 keV CT images in epoch 10, 50, 100, and 500 were compared to the 55 keV images generated using peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM).ResultsThe PSNRs were 28.0, 28.5, 28.6, and 28.8 at epochs 10, 50, 100, and 500, respectively. The SSIM was approximately constant from epochs 50 to 500.ConclusionPseudo low monoenergetic abdominal CT images were generated from 120-kVp CT images using cGAN, and the images had good quality similar to that of monochromatic images obtained with DECT software.     

Experimental evaluation of depth-of-interaction correction in a small-animal positron emission tomography scanner

Human and small-animal positron emission tomography (PET) scanners with cylindrical geometry and conventional detectors exhibit a progressive reduction in radial spatial resolution with increasing radial distance from the geometric axis of the scanner. This "depth-of-interaction" (DOI) effect is sufficiently deleterious that many laboratories have devised novel schemes to reduce the magnitude of this effect and thereby yield PET images of greater quantitative accuracy. Here we examine experimentally the effects of a particular DOI correction method (dual-scintillator phoswich detectors with pulse shape discrimination) implemented in a small-animal PET scanner by comparing the same phantom and same mouse images with and without DOI correction. The results suggest that even this relatively coarse, two-level estimate of radial gamma ray interaction position significantly reduces the DOI parallax error. This study also confirms two less appreciated advantages of DOI correction: a reduction in radial distortion and radial source displacement as a source is moved toward the edge of the field of view and a resolution improvement detectable in the central field of view likely owing to improved spatial sampling.     

Validation and practical implementation of seated position radiotherapy in a commercial TPS for proton therapy

PurposeThis work aims to validate new 6D couch features and their implementation for seated radiotherapy in RayStation (RS) treatment planning system (TPS).Materials and methodsIn RS TPS, new 6D couch features are (i) chair support device, (ii) patient treatment option of “Sitting: face towards the front of the chair”, and (iii) patient support pitch and roll capabilities. The validation of pitch and roll was performed by comparing TPS generated DRRs with planar x-rays. Dosimetric tests through measurement by 2D ion chamber array were performed for beams created with varied scanning and treatment orientation and 6D couch rotations. For the implementation of 6D couch features for treatments in a seated position, the TPS and oncology information system (Mosaiq) settings are described for a commercial chair. An end-to-end test using an anthropomorphic phantom was performed to test the complete workflow from simulation to treatment delivery.ResultsThe 6D couch features were found to have a consistent implementation that met IEC 61712 standard. The DRRs were found to have an acceptable agreement with planar x-rays based on visual inspection. For dose map comparison between measured and calculated, the gamma index analysis for all the beams was >95% at a 3% dose-difference and 3 mm distance-to-agreement tolerances. For an end-to end-testing, the phantom was successfully set up at isocenter in the seated position and treatment was delivered.ConclusionsChair-based treatments in a seated position can be implemented in RayStation through the use of newly released 6D couch features.     

An improved algorithm for statistical alignment of sequences related by a star tree

The insertion-deletion model developed by Thorne, Kishino and Felsenstein (1991, J. Mol. Evol., 33, 114–124; the TKF91 model) provides a statistical framework of two sequences. The statistical alignment of a set of sequences related by a star tree is a generalization of this model. The known algorithm computes the probability of a set of such sequences in O(l ^2k ) time, where l is the geometric mean of the sequence lengths and k is the number of sequences. An improved algorithm is presented whose running time is only O(2^2k l ^k).     

Experimental tissue molding for soft-tissue reconstruction: a preliminary report

Molding of pig omentum to create external structures, such as the ear or penis, was performed using silicone rubber molds. The omentum was mobilized on its gastroepiploic pedicle and brought out through a fascial defect. The molds containing the omentum were placed beneath the skin and panniculus carnosus for 4 weeks. Split-thickness skin grafts were placed on both surfaces of the mold, and a series of synthetic materials was placed within the omentum to provide structural support. When an appropriate amount of tissue had been placed within the device, omentum took on the contour of the mold. Excessive pressure associated with swelling resulted in necrosis, while an inadequate amount of tissue gave poor contour. Omentum molded without any support material rapidly lost its contour once removed from the mold. Omentum molded in combination with a synthetic material maintained its shape up to 1 month after mold removal in a limited number of animals. This model was designed to create structures with the fine detail of a prosthesis using autologous tissue. This "prosthesis" made from living tissue could then be transferred secondarily to its recipient area by microvascular technique.     

A new statistical method for haplotype reconstruction from population data

Current routine genotyping methods typically do not provide haplotype information, which is essential for many analyses of fine-scale molecular-genetics data. Haplotypes can be obtained, at considerable cost, experimentally or (partially) through genotyping of additional family members. Alternatively, a statistical method can be used to infer phase and to reconstruct haplotypes. We present a new statistical method, applicable to genotype data at linked loci from a population sample, that improves substantially on current algorithms; often, error rates are reduced by > 50%, relative to its nearest competitor. Furthermore, our algorithm performs well in absolute terms, suggesting that reconstructing haplotypes experimentally or by genotyping additional family members may be an inefficient use of resources.