China
Africa
Explore chapters and articles related to this topic
Special Considerations in Pediatric Nuclear Medicine
Published in Michael Ljungberg, Handbook of Nuclear Medicine and Molecular Imaging for Physicists, 2022
Sofie Lindskov Hansen, Søren Holm, Liselotte Højgaard, Lise Borgwardt
Most commercial scanners allow the user to choose between filtered back projection (FBP) and iterative reconstruction methods. Furthermore, if the iterative reconstruction method is chosen, the “strength” can be set by the user. Iterative reconstruction methods have the potential to reduce the dose for a given image quality or to reduce the image noise for a given level of radiation exposure [30]. Although iterative reconstruction techniques increase the image quality when measured quantitatively, the images may appear smoother or glossier, which is disturbing to radiologists familiar with the texture known from FBP-reconstructed images. Consequently, vendors and scientists have embarked upon new ways of generating something that looks like a high-dose FBP image from low-dose images, and one promising path is to use deep-learning algorithms to up-convert low-dose images to high dose quality without changing the noise-texture of the images. One such commercially available deep-learning approach was recently presented. It is called True Fidelity and is developed by GE Healthcare. Their Deep Learning Image Reconstruction (DLIR) features a deep neural network that has been trained on high-quality FBP data sets in order to learn how to differentiate noise from signals, and thus to suppress noise without impacting anatomical and pathological structures [31].
Imaging modalities and challenges
Published in Rolf Behling, Modern Diagnostic X-Ray Sources, 2021
Though computationally expensive and challenging clinical workflow, current technology using additional dedicated processing hardware allows for handling multiple backprojection, forward-projection, and correction steps in iterative optimization processes. The physics of the CT system is modeled. Forward projections of the object function are simulated, based on starting values, e.g., from FBP. Discrepancies between simulated and measured data are iteratively minimized in correction steps, while non-plausible data are filtered out and data inaccuracies considered. Knowledge about the Poisson statistics of quantum noise is used to weight projection data according to their signal strength. Depending on the algorithm and clinical task, the number of iterations may reach several dozens until satisfying a stop criterion. Different vendor-specific concepts have been devised, e.g., model-based iterative reconstruction or maximum-likelihood iterative reconstruction (see, e.g., Ziegler et al., 2007 or Van Eyndhoven & Sijbers, 2018). Iterative reconstruction enables image quality improvement, dose reduction, and artifact suppression (see Noël et al., 2014; Sauter et al., 2016). The images generally display a more artificial look than those most radiologists are used to.
Introduction
Published in A Stewart Whitley, Jan Dodgeon, Angela Meadows, Jane Cullingworth, Ken Holmes, Marcus Jackson, Graham Hoadley, Randeep Kumar Kulshrestha, Clark’s Procedures in Diagnostic Imaging: A System-Based Approach, 2020
A Stewart Whitley, Jan Dodgeon, Angela Meadows, Jane Cullingworth, Ken Holmes, Marcus Jackson, Graham Hoadley, Randeep Kumar Kulshrestha
The CT scan is used for PET attenuation correction and the PET images are typically reconstructed using iterative reconstruction techniques. The minimum datasets required for the image interpreter are: CT scout, low-dose CT, PET AC and NAC. A 3D MIP of the PET also provides additional data for the interpreter as an overview. Additional fused PET–CT datasets with colour maps can be produced of transaxial, coronal and sagittal planes. These are useful for multidisciplinary team discussions, presenting the data in a format for immediate image interpretation. Figure 1.59a presents a sample of the minimum data set in one screen shot.
Sparse-View Cone-Beam CT Reconstruction by Bar-by-Bar Neural FDK Algorithm
Published in Nondestructive Testing and Evaluation, 2023
Siqi Wang, Tatsuya Yatagawa, Yutaka Ohtake, Hiromasa Suzuki
On the other hand, iterative reconstruction refers to a class of methods that update the volumetric image until it agrees with input projections. The iterative reconstruction typically solves a least-squares problem to minimise the difference between the volumetric output image and the input projections. Moreover, there are several variants such as simultaneous algebraic reconstruction techniques (SART) [12–14], minimisation-based iterative reconstruction [15,16] and model-based iterative reconstruction [17,18], depending on how the minimisation problem is formulated. Although these methods have been proven effective for reducing sparse-view artefacts in CBCT, their main drawback is the long processing time and limits their practical applications [19]. Moreover, they sometimes lead to oversmoothed reconstruction, which prevents the detailed examination of target objects [20].
Long axial field-of-view PET/CT devices: are we ready for the technological revolution?
Published in Expert Review of Medical Devices, 2022
Luca Filippi, Antonia Dimitrakopoulou-Strauss, Laura Evangelista, Orazio Schillaci
In last decades hybrid imaging, combining molecular and anatomical data in a unique, synergistic approach, has thoroughly changed the face of medical diagnostics [1,2]. In particular, positron emission computed tomography (PET/CT) has established itself as an essential tool in many oncological and non-oncological scenarios [3], providing the opportunity of investigating in vivo physio-pathological processes at a cellular and molecular level [4,5]. Notably, in recent years some technological improvements have been introduced in PET imaging, such as novel iterative reconstruction algorithms, or time-of-flight (TOF) PET/CT scanners operating in fully-3D mode [6]. Most importantly, the silicon photomultiplier (SiPM)-based detectors have been implemented instead of the ‘old-fashioned’ photomultiplier tubes (PMTs) [7,8], giving rise to the so-called digital PET/CT (dPET/CT). With respect to the PMT-equipped PET/CT, namely analogue PET/CT (aPET/CT), dPET/CT is characterized by higher sensitivity, spatial and temporal resolution, with a significantly greater detection rate of pathological lesions, also employing fast protocols [9–15].