Search

OBJECTIVE. Previous advances over filtered back projection (FBP) have incorporated model-based iterative reconstruction. The purpose of this study was to characterize the latest advance in image reconstruction, that is, deep learning. The focus was on applying characterization results of a deep learning approach to decisions about clinical CT protocols. MATERIALS AND METHODS. A proprietary deep learning image reconstruction (DLIR) method was characterized against an existing advanced adaptive statistical iterative reconstruction method (ASIR-V) and FBP from the same vendor. The metrics used were contrast-to-noise ratio, spatial resolution as a function of contrast level, noise texture (i.e., noise power spectra [NPS]), noise scaling as a function of slice thickness, and CT number consistency. The American College of Radiology accreditation phantom and a uniform water phantom were used at a range of doses and slice thicknesses for both axial and helical acquisition modes. RESULTS. ASIR-V and DLIR were associated with improved contrast-to-noise ratio over FBP for all doses and slice thicknesses. No dose or contrast dependencies of spatial resolution were observed for ASIR-V or DLIR. NPS results showed DLIR maintained an FBP-like noise texture whereas ASIR-V shifted the NPS to lower frequencies. Noise changed with dose and slice thickness in the same manner for ASIR-V and FBP. DLIR slice thickness noise scaling differed from FBP, exhibiting less noise penalty with decreasing slice thickness. No clinically significant changes were observed in CT numbers for any measurement condition. CONCLUSION. In a phantom model, DLIR does not suffer from the concerns over reduction in spatial resolution and introduction of poor noise texture associated with previous methods.

6.1 System Overview Before giving a detailed analysis and description of major components in a CT scanner, this chapter will present a system overview to explain how the different components work together to produce CT images. Figure 6.1 presents a generic block diagram of a CT system. The actual system architecture for different commercial scanners may deviate from this diagram, but the general functionalities of all CT scanners are more or less the same. For a typical CT operation, an operator positions a patient on the CT table and prescribes a scanogram or “scout view.” The purpose of this scan is to determine the patient's anatomical landmarks and the exact location and range of CT scans. In this scan mode, both the x-ray tube and the detector remain stationary while the patient table travels at a constant speed. The scan is similar to a conventional x ray taken either at an A-P position (with the tube located in the 6 or 12 o'clock position) or a lateral position (with the tube located in the 3 or 9 o'clock position). Once such a scan is initiated, an operational control computer instructs the gantry to rotate to the desired orientation as prescribed by the operator. The computer then sends instructions to the patient table, the x-ray generation system, the x-ray detection system, and the image generation system to perform a scan. The table subsequently reaches the starting scan location and maintains a constant speed during the entire scanning process. The high-voltage generator quickly reaches the desired voltage and keeps both the voltage and the current to the x-ray tube at the prescribed level during the scan. The x-ray tube produces x-ray flux, and the x-ray photons are detected by an x-ray detector to produce electrical signals. At the same time, the data acquisition system samples the detector outputs at a uniform sampling rate and converts analog signals to digital signals. The sampled data are then sent to the image generation system for processing. Typically, the system contains high-speed computers and digital signal processing (DSP) chips. The acquired data are preprocessed and enhanced before being sent to the display device for operator viewing and to the data storage device for archiving.

9.1 Introduction Helical CT (also called spiral CT) was introduced commercially in the late 1980s and early 1990s. Helical CT has expanded the traditional CT capability by enabling the scan of an entire organ in a single breath-hold. It is safe to state that helical CT is one of the key steps that moved CT from a slice-oriented imaging modality to an organ-oriented modality. The difference in the naming convention between helical and spiral CT is due mainly to different CT manufacturers. For all practical and technical purposes, there is no difference between the two. To avoid confusion, we will use the term “helical” throughout this chapter. 9.1.1 Clinical needs All previous chapters have focused on a single scanning protocol: the step-and-shoot mode. This scanning protocol contains both a data acquisition period and non-data-acquisition period. During the data acquisition period, the patient remains stationary while the x-ray tube and detector rotates about the patient at a constant speed. Once a complete projection dataset is acquired for the slice, the non-data-acquisition period starts. The x-ray tube is turned off and the patient is indexed to the next scanning location. For typical CT scanners, the minimum non-data-acquisition period is on the order of seconds as a result of both mechanical and patient constraints. The mechanical constraint is due to the fact that a typical patient weighs over 45 kg, and the patient table requires a certain amount of time to move a large mass from one location to another. The cause of the patient constraint may not be as obvious. From the law of physics we know that to move a resting object over a short distance, first we must accelerate the object up to a certain speed and decelerate the object when it is near the target location. Since the distance between adjacent scanning locations is typically a few millimeters, the amount of acceleration and deceleration is fairly large. A human body is not rigid (the internal organs can move and deform), so the acceleration and deceleration will likely induce motion in the patient. As a result, a certain amount of time must elapse to minimize motion artifacts. In the late 1980s, the CT scan speed approached one second per revolution.

When x-ray radiation penetrates an object, part of its energy is transferred to the object and causes changes in the object’s material. During the energy transfer, an x-ray can indirectly produce ion pairs in the tissue. The ion pairs react with other chemical systems and cause radiation damage. Alternatively, the x-rays may strike and break molecular bonds, such as those in DNA, and cause direct damage.

Radiation dose reduction is critical to the success of x-ray computed tomography (CT). Many advanced reconstruction techniques have been developed over the years to combat noise resulting from the low-dose CT scans. These algorithms rely on accurate local estimation of the image noise to determine reconstruction parameters or to select inferencing models. Because of the difficulties in the noise estimation for heterogeneous objects, the performance of many algorithms is inconsistent and suboptimal. Here, we propose a novel approach to overcome such shortcoming.By injecting appropriate amount of noise in the CT raw data, a computer simulation approach is capable of accurately estimating the local statistics of the raw data and the local noise in the reconstructed images. This information is then used to guide the noise-reduction process during the reconstruction. As an initial implementation, a scaling map is generated based on the noise predicted from the simulation and the noise estimated from existing reconstruction algorithms. Images generated with existing algorithms are subsequently modified based on the scaling map. In this study, both iterative reconstruction (IR) and deep learning image reconstruction (DLIR) algorithms are evaluated.Phantom experiments were conducted to evaluate the performance of the simulation-based noise estimation in terms of the standard deviation and noise power spectrum. Quantitative results have demonstrated that the noise measured from the original image matches well with the noise estimated from the simulation. Clinical datasets were utilized to further confirm the accuracy of the proposed approach under more challenging conditions. To validate the performance of the proposed reconstruction approach, clinical scans were used. Performance comparison was carried out qualitatively and quantitatively. Two existing advanced reconstruction techniques, IR and DLIR, were evaluated against the proposed approach. Results have shown that the proposed approach outperforms existing IR and DLIR algorithms in terms of noise suppression and, equally importantly, noise uniformity across the entire imaging volume. Visual assessment of the images also reveals that the proposed approach does not endure noise texture issues faced by some of the existing reconstruction algorithms today.Phantom and clinical results have demonstrated superior performance of the proposed approach with regard to noise reduction as well as noise homogeneity. Visual inspection of the noise texture further confirms the clinical utility of the proposed approach. Future enhancements on the current implementation are explored regarding image quality and computational efficiency. Because of the limited scope of this paper, detailed investigation on these enhancement features will be covered in a separate report.

Deep silicon-based photon-counting CT (Si-PCCT) is an emerging detector technology that provides improved spatial resolution by virtue of its reduced pixel sizes. This article reports the outcomes of the first simulation study evaluating the impact of this advantage over energy-integrating CT (ECT) for estimation of morphological radiomics features in lung lesions.A dynamic nutrient-access-based stochastic model was utilized to generate three distinct morphologies for lung lesions. The lesions were inserted into the lung parenchyma of an anthropomorphic phantom (XCAT - 50Compared to ECT, the mean estimation error was lower for Si-PCCT (independent features: 35.9% vs. 54.0%, all features: 54.5% vs. 68.1%) with statistically significant reductions in errors for 8/14 features. For both systems, the estimation accuracy was minimally affected by dose and distance from the isocenter while reconstruction kernel and pixel size were observed to have a relatively stronger effect.For all lesions and imaging conditions considered, Si-PCCT exhibited improved estimation accuracy for morphological radiomics features over a conventional ECT system, demonstrating the potential of this technology for improved quantitative imaging.

Purpose: We provide a review of the key computed tomography (CT) technologies developed since the late 1980s and offer an overview of one of the future technologies under development. The focus of this review is mainly on the hardware and system development. The topics on the historical event linked to the early days of CT development and other innovations that contributed to the CT development, such as advanced image reconstruction techniques, are covered by companion papers in this special issue. Approach: The review is divided into five major sections, each linked to a key innovation in CT: helical spiral data acquisition, multi-slice CT, wide-cone CT, dual-source CT, and spectral CT. Given the limited scope of this review, only one of the future technologies, photon-counting CT, is discussed in detail. Whenever possible, both theory of operation and clinical examples are provided. Results: Theoretical analyses, phantom results, and clinical examples clearly demonstrate the efficacy and clinical relevancy of five historical technology developments and one future technology in CT. These technologies have improved and will continue to improve CT performance in terms of isotropic volume coverage, improved temporal resolution, and material differentiation and characterization capabilities. Conclusions: Over the past 30 years, technological developments of CT have contributed to the success of CT in many clinical applications such as trauma, oncology, cardiac imaging, and stroke. Advanced clinical applications have and will continue to demand more advanced technology development.

To avoid severe limited-view artifacts in reconstructed CT images, current multi-row detector CT (MDCT) scanners with a single x-ray source-detector assembly need to limit table translation speeds such that the pitch ${p}$ (viz., normalized table translation distance per gantry rotation) is lower than 1.5. When ${p}>{1.5}$ , it remains an open question whether one can reconstruct clinically useful helical CT images without severe artifacts. In this work, we show that a synergistic use of advanced techniques in conventional helical filtered backprojection, compressed sensing, and more recent deep learning methods can be properly integrated to enable accurate reconstruction up to ${p}={4}$ without significant artifacts for single source MDCT scans.

Low Performing Pixel (LPP)/bad pixel in CT detectors cause ring and streaks artifacts, structured non-uniformities and deterioration of the image quality. These artifacts make the image unusable for diagnostic purposes. A missing/defective detector pixel translates to a channel missing across all views in sinogram domain and its effect gets spill over entire image in reconstruction domain as artifacts. Most of the existing ring and streak removal algorithms perform correction only in the reconstructed image domain. In this work, we propose a supervised deep learning algorithm that operates in sinogram domain to remove distortions cause by the LPP. This method leverages CT scan geometry, including conjugate ray information to learn the interpolation in sinogram domain. While the experiments are designed to cover the entire detector space, we emphasize on LPPs near detector iso-center as these have most adverse impact on image quality specially if the LPPs fall on the high frequency region (bone-tissue interface). We demonstrated efficacy of the proposed method using data acquired on GE RevACT multi-slice CT system with flat-panel detector. Experimental results on head scans show significant reduction in ring artifacts regardless of LPP location in the detector geometry. We have simulated isolated LPPs accounting for 5% and 10% of total channels. Detailed statistical analysis illustrates approximately 5dB improvement in SNR in both sinogram and reconstruction domain as compared to classical bicubic and Lagrange interpolation methods. Also, with reduction in ring and streak artifacts, the perceptual image quality is improved across all the test images.