Purpose
To characterize the noise and spatial resolution properties of a commercially available deep learning‐based computed tomography (CT) reconstruction algorithm.
Methods
Two phantom experiments ...were performed. The first used a multisized image quality phantom (Mercury v3.0, Duke University) imaged at five radiation dose levels (CTDIvol: 0.9, 1.2, 3.6, 7.0, and 22.3 mGy) with a fixed tube current technique on a commercial CT scanner (GE Revolution CT). Images were reconstructed with conventional (FBP), iterative (GE ASiR‐V), and deep learning‐based (GE True Fidelity) reconstruction algorithms. Noise power spectrum (NPS), high‐contrast (air–polyethylene interface), and intermediate‐contrast (water–polyethylene interface) task transfer functions (TTF) were measured for each dose level and phantom size and summarized in terms of average noise frequency (fav) and frequency at which the TTF was reduced to 50% (f50%), respectively. The second experiment used a custom phantom with low‐contrast rods and lung texture sections for the assessment of low‐contrast TTF and noise spatial distribution. The phantom was imaged at five dose levels (CTDIvol: 1.0, 2.1, 3.0, 6.0, and 10.0 mGy) with 20 repeated scans at each dose, and images reconstructed with the same reconstruction algorithms. The local noise stationarity was assessed by generating spatial noise maps from the ensemble of repeated images and computing a noise inhomogeneity index, η, following AAPM TG233 methods. All measurements were compared among the algorithms.
Results
Compared to FBP, noise magnitude was reduced on average (± one standard deviation) by 74 ± 6% and 68 ± 4% for ASiR‐V (at “100%” setting) and True Fidelity (at “High” setting), respectively. The noise texture from ASiR‐V had substantially lower noise frequency content with 55 ± 4% lower NPS fav compared to FBP while True Fidelity had only marginally different noise frequency content with 9 ± 5% lower NPS fav compared to FBP. Both ASiR‐V and True Fidelity demonstrated locally nonstationary noise in a lung texture background at all radiation dose levels, with higher noise near high‐contrast edges of vessels and lower noise in uniform regions. At the 1.0 mGy dose level η values were 314% and 271% higher in ASiR‐V and True Fidelity compared to FBP, respectively. High‐contrast spatial resolution was similar between all algorithms for all dose levels and phantom sizes (<3% difference in TTF f50%). Compared to FBP, low‐contrast spatial resolution was lower for ASiR‐V and True Fidelity with a reduction of TTF f50% of up to 42% and 36%, respectively.
Conclusions
The deep learning‐based CT reconstruction demonstrated a strong noise magnitude reduction compared to FBP while maintaining similar noise texture and high‐contrast spatial resolution. However, the algorithm resulted in images with a locally nonstationary noise in lung textured backgrounds and had somewhat degraded low‐contrast spatial resolution similar to what has been observed in currently available iterative reconstruction techniques.
Purpose To determine the effect of radiation dose and iterative reconstruction (IR) on noise, contrast, resolution, and observer-based detectability of subtle hypoattenuating liver lesions and to ...estimate the dose reduction potential of the IR algorithm in question. Materials and Methods This prospective, single-center, HIPAA-compliant study was approved by the institutional review board. A dual-source computed tomography (CT) system was used to reconstruct CT projection data from 21 patients into six radiation dose levels (12.5%, 25%, 37.5%, 50%, 75%, and 100%) on the basis of two CT acquisitions. A series of virtual liver lesions (five per patient, 105 total, lesion-to-liver prereconstruction contrast of -15 HU, 12-mm diameter) were inserted into the raw CT projection data and images were reconstructed with filtered back projection (FBP) (B31f kernel) and sinogram-affirmed IR (SAFIRE) (I31f-5 kernel). Image noise (pixel standard deviation), lesion contrast (after reconstruction), lesion boundary sharpness (average normalized gradient at lesion boundary), and contrast-to-noise ratio (CNR) were compared. Next, a two-alternative forced choice perception experiment was performed (16 readers six radiologists, 10 medical physicists). A linear mixed-effects statistical model was used to compare detection accuracy between FBP and SAFIRE and to estimate the radiation dose reduction potential of SAFIRE. Results Compared with FBP, SAFIRE reduced noise by a mean of 53% ± 5, lesion contrast by 12% ± 4, and lesion sharpness by 13% ± 10 but increased CNR by 89% ± 19. Detection accuracy was 2% higher on average with SAFIRE than with FBP (P = .03), which translated into an estimated radiation dose reduction potential (±95% confidence interval) of 16% ± 13. Conclusion SAFIRE increases detectability at a given radiation dose (approximately 2% increase in detection accuracy) and allows for imaging at reduced radiation dose (16% ± 13), while maintaining low-contrast detectability of subtle hypoattenuating focal liver lesions. This estimated dose reduction is somewhat smaller than that suggested by past studies.
RSNA, 2017 Online supplemental material is available for this article.
The purpose of this study was to develop a CT simulation platform that is: 1) compatible with voxel-based computational phantoms; 2) capable of modeling the geometry and physics of commercial CT ...scanners; and 3) computationally efficient. Such a simulation platform is designed to enable the virtual evaluation and optimization of CT protocols and parameters for achieving a targeted image quality while reducing radiation dose. Given a voxelized computational phantom and a parameter file describing the desired scanner and protocol, the developed platform DukeSim calculates projection images using a combination of ray-tracing and Monte Carlo techniques. DukeSim includes detailed models for the detector quantum efficiency, quantum and electronic noise, detector crosstalk, subsampling of the detector and focal spot areas, focal spot wobbling, and the bowtie filter. DukeSim was accelerated using GPU computing. The platform was validated using physical and computational versions of a phantom (Mercury phantom). Clinical and simulated CT scans of the phantom were acquired at multiple dose levels using a commercial CT scanner (Somatom Definition Flash; Siemens Healthcare). The real and simulated images were compared in terms of image contrast, noise magnitude, noise texture, and spatial resolution. The relative error between the clinical and simulated images was less than 1.4%, 0.5%, 2.6%, and 3%, for image contrast, noise magnitude, noise texture, and spatial resolution, respectively, demonstrating the high realism of DukeSim. The runtime, dependent on the imaging task and the hardware, was approximately 2-3 minutes per rotation in our study using a computer with 4 GPUs. DukeSim, when combined with realistic human phantoms, provides the necessary toolset with which to perform large-scale and realistic virtual clinical trials in a patient and scanner-specific manner.
Purpose:
To develop and validate an automated technique for evaluating the spatial resolution characteristics of clinical computed tomography (CT) images.
Methods:
Twenty one chest and abdominopelvic ...clinical CT datasets were examined in this study. An algorithm was developed to extract a CT resolution index (RI) analogous to the modulation transfer function from clinical CT images by measuring the edge-spread function (ESF) across the patient’s skin. A polygon mesh of the air-skin boundary was created. The faces of the mesh were then used to measure the ESF across the air-skin interface. The ESF was differentiated to obtain the line-spread function (LSF), and the LSF was Fourier transformed to obtain the RI. The algorithm’s ability to detect the radial dependence of the RI was investigated. RIs measured with the proposed method were compared with a conventional phantom-based method across two reconstruction algorithms (FBP and iterative) using the spatial frequency at 50% RI, f
50, as the metric for comparison. Three reconstruction kernels were investigated for each reconstruction algorithm. Finally, an observer study was conducted to determine if observers could visually perceive the differences in the measured blurriness of images reconstructed with a given reconstruction method.
Results:
RI measurements performed with the proposed technique exhibited the expected dependencies on the image reconstruction. The measured f
50 values increased with harder kernels for both FBP and iterative reconstruction. Furthermore, the proposed algorithm was able to detect the radial dependence of the RI. Patient-specific measurements of the RI were comparable to the phantom-based technique, but the patient data exhibited a large spread in the measured f
50, indicating that some datasets were blurrier than others even when the projection data were reconstructed with the same reconstruction algorithm and kernel. Results from the observer study substantiated this finding.
Conclusions:
Clinically informed, patient-specific spatial resolution can be measured from clinical datasets. The method is sufficiently sensitive to reflect changes in spatial resolution due to different reconstruction parameters. The method can be applied to automatically assess the spatial resolution of patient images and quantify dependencies that may not be captured in phantom data.
The purpose of this study was to assess image noise, spatial resolution, lesion detectability, and the dose reduction potential of a proprietary third-generation adaptive statistical iterative ...reconstruction (ASIR-V) technique.
A phantom representing five different body sizes (12-37 cm) and a contrast-detail phantom containing lesions of five low-contrast levels (5-20 HU) and three sizes (2-6 mm) were deployed. Both phantoms were scanned on a 256-MDCT scanner at six different radiation doses (1.25-10 mGy). Images were reconstructed with filtered back projection (FBP), ASIR-V with 50% blending with FBP (ASIR-V 50%), and ASIR-V without blending (ASIR-V 100%). In the first phantom, noise properties were assessed by noise power spectrum analysis. Spatial resolution properties were measured by use of task transfer functions for objects of different contrasts. Noise magnitude, noise texture, and resolution were compared between the three groups. In the second phantom, low-contrast detectability was assessed by nine human readers independently for each condition. The dose reduction potential of ASIR-V was estimated on the basis of a generalized linear statistical regression model.
On average, image noise was reduced 37.3% with ASIR-V 50% and 71.5% with ASIR-V 100% compared with FBP. ASIR-V shifted the noise power spectrum toward lower frequencies compared with FBP. The spatial resolution of ASIR-V was equivalent or slightly superior to that of FBP, except for the low-contrast object, which had lower resolution. Lesion detection significantly increased with both ASIR-V levels (p = 0.001), with an estimated radiation dose reduction potential of 15% ± 5% (SD) for ASIR-V 50% and 31% ± 9% for ASIR-V 100%.
ASIR-V reduced image noise and improved lesion detection compared with FBP and had potential for radiation dose reduction while preserving low-contrast detectability.
Purpose:
To investigate a measurement method for evaluating the resolution properties of CT imaging systems across reconstruction algorithms, dose, and contrast.
Methods:
An algorithm was developed ...to extract the task-based modulation transfer function (MTF) from disk images generated from the rod inserts in the ACR phantom (model 464 Gammex, WI). These inserts are conventionally employed for HU accuracy assessment. The edge of the disk objects was analyzed to determine the edge-spread function, which was differentiated to yield the line-spread function and Fourier-transformed to generate the object-specific MTF for task-based assessment, denoted MTFTask. The proposed MTF measurement method was validated against the conventional wire technique and further applied to measure the MTF of CT images reconstructed with an adaptive statistical iterative algorithm (ASIR) and a model-based iterative (MBIR) algorithm. Results were further compared to the standard filtered back projection (FBP) algorithm. Measurements were performed and compared across different doses and contrast levels to ascertain the MTFTask dependencies on those factors.
Results:
For the FBP reconstructed images, the MTFTask measured with the inserts were the same as the MTF measured from the wire-based method. For the ASIR and MBIR data, the MTFTask using the high contrast insert was similar to the wire-based MTF and equal or superior to that of FBP. However, results for the MTFTask measured using the low-contrast inserts, the MTFTask for ASIR and MBIR data was lower than for the FBP, which was constant throughout all measurements. Similarly, as a function of mA, the MTFTask for ASIR and MBIR varied as a function of noise–-with MTFTask being proportional to mA. Overall greater variability of MTFTask across dose and contrast was observed for MBIR than for ASIR.
Conclusions:
This approach provides a method for assessing the task-based MTF of a CT system using conventional and iterative reconstructions. Results demonstrated that the object-specific MTF can vary as a function of dose and contrast. The analysis highlighted the paradigm shift for iterative reconstructions when compared to FBP, where iterative reconstructions generally offer superior noise performance but with varying resolution as a function of dose and contrast. The MTFTask generated by this method is expected to provide a more comprehensive assessment of image resolution across different reconstruction algorithms and imaging tasks.
Purpose:
To use novel voxel-based 3D printed textured phantoms in order to compare low-contrast detectability between two reconstruction algorithms, FBP (filtered-backprojection) and SAFIRE (sinogram ...affirmed iterative reconstruction) and determine what impact background texture (i.e., anatomical noise) has on estimating the dose reduction potential of SAFIRE.
Methods:
Liver volumes were segmented from 23 abdominal CT cases. The volumes were characterized in terms of texture features from gray-level co-occurrence and run-length matrices. Using a 3D clustered lumpy background (CLB) model, a fitting technique based on a genetic optimization algorithm was used to find CLB textures that were reflective of the liver textures, accounting for CT system factors of spatial blurring and noise. With the modeled background texture as a guide, four cylindrical phantoms (Textures A-C and uniform, 165 mm in diameter, and 30 mm height) were designed, each containing 20 low-contrast spherical signals (6 mm diameter at nominal contrast levels of ∼3.2, 5.2, 7.2, 10, and 14 HU with four repeats per signal). The phantoms were voxelized and input into a commercial multimaterial 3D printer (Object Connex 350), with custom software for voxel-based printing (using principles of digital dithering). Images of the textured phantoms and a corresponding uniform phantom were acquired at six radiation dose levels (SOMATOM Flash, Siemens Healthcare) and observer model detection performance (detectability index of a multislice channelized Hotelling observer) was estimated for each condition (5 contrasts × 6 doses × 2 reconstructions × 4 backgrounds = 240 total conditions). A multivariate generalized regression analysis was performed (linear terms, no interactions, random error term, log link function) to assess whether dose, reconstruction algorithm, signal contrast, and background type have statistically significant effects on detectability. Also, fitted curves of detectability (averaged across contrast levels) as a function of dose were constructed for each reconstruction algorithm and background texture. FBP and SAFIRE were compared for each background type to determine the improvement in detectability at a given dose, and the reduced dose at which SAFIRE had equivalent performance compared to FBP at 100% dose.
Results:
Detectability increased with increasing radiation dose (P = 2.7 × 10−59) and contrast level (P = 2.2 × 10−86) and was higher in the uniform phantom compared to the textured phantoms (P = 6.9 × 10−51). Overall, SAFIRE had higher d′ compared to FBP (P = 0.02). The estimated dose reduction potential of SAFIRE was found to be 8%, 10%, 27%, and 8% for Texture-A, Texture-B, Texture-C and uniform phantoms.
Conclusions:
In all background types, detectability was higher with SAFIRE compared to FBP. However, the relative improvement observed from SAFIRE was highly dependent on the complexity of the background texture. Iterative algorithms such as SAFIRE should be assessed in the most realistic context possible.
Clinical Imaging Physics Ehsan Samei, Douglas E. Pfeiffer / Ehsan Samei, Douglas E. Pfeiffer
2020, 20200101, 2020-04-23, 2020-04-20
eBook
Clinical Medical Imaging Physics: Current and Emerging Practice is the first text of its kind--a comprehensive reference work covering all imaging modalities in use in clinical medicine today. ...Destined to become a classic in the field, this book provides state-of-practice descriptions for each imaging modality, followed by special sections on new and emerging applications, technologies, and practices. Authored by luminaries in the field of medical physics, this resource is a sophisticated, one-volume handbook to a fast-advancing field that is becoming ever more central to contemporary clinical medicine. * Summarizes the current state of clinical medical imaging physics in one volume, with a focus on emerging technologies and applications * Provides comprehensive coverage of all key clinical imaging modalities, taking into account the new realities in healthcare practice * Features a strong focus on clinical application of principles and technology, now and in the future * Contains authoritative text compiled by world-renowned editors and contributors responsible for guiding the development of the field Practicing radiologists and medical physicists will appreciate Clinical Medical Imaging Physics as a peerless everyday reference work. Additionally, graduate students and residents in medical physics and radiology will find this book essential as they study for their board exams.
An updated extension of effective dose was recently introduced, namely relative effective dose (
), incorporating age and sex factors. In this study we extended
application to a population of about ...9000 patients who underwent multiple CT imaging exams, and we compared it with other commonly used radiation protection metrics in terms of their correlation with radiation risk. Using Monte Carlo methods,
, dose-length-product based effective dose (
), organ-dose based effective dose (
), and organ-dose based risk index (
) were calculated for each patient. Each metric's dependency to
was assessed in terms of its sensitivity and specificity.
showed the best sensitivity, specificity, and agreement with
(R
= 0.97); while
yielded the lowest specificity and, along with
, the lowest sensitivity. Compared to other metrics,
provided a closer representation of patient and group risk also incorporating age and sex factors within the established framework of effective dose.