Advanced searches left 3/3

CT - Astrophysics Data System

Summarized by Plex Scholar
Last Updated: 23 July 2022

* If you want to update the article please login/register

Liver Segmentation using Turbolift Learning for CT and Cone-beam C-arm Perfusion Imaging

The liver should be properly segmented from CT scans in order to apply TST using prior knowledge obtained from CT perfusion results. Turbolift learning, which offers a customized version of the multi-scale Attention UNet on various liver segmentation tasks serially, following the order of the trainings, CT, CBCT, and CBCT TST, makes the previous trainings appear as pre-training stages for the upcoming ones, addressing the issue of a lack of datasets for training. Turbolift not only enhances the overall performance of the model but also makes it more robust against artefacts originating from embolisation materials and truncation artefacts, according to Experiments, which proved that Turbolift does not only improves the model's stability but also makes it tough against artefacts originating from the embolization materials and truncation artefacts. This paper shows the possibility of segmenting the liver from CT, CBCT, and CBCT TST, learning from the existing limited training results, which may be used in the future for the generation and evaluation of liver disease maps.

Source link: https://ui.adsabs.harvard.edu/abs/2022arXiv220710167H/abstract


Material decomposition from photon-counting CT using a convolutional neural network and energy-integrating CT training labels

With PCD filtered back reconstruction reconstruction, PCD iterative reconstruction, and PCD decomposition as the input, we use a 3D U-net architecture and compare networks. Regardless of the input, we discovered that our Iter2Decomp solution does the best, but matrix inversion decomposition outperforms matrix inversion decomposition. Iter2Decomp makes a 27. 5 percent lower root mean squared error in the iodine map and 59. 87% lower in the photoelectric effect map, relative to PCD matrix inversion decomposition. When using Iter2Decomp, one limitation is some blurring caused by our DL strategy, with a decrease from 1. 98 line pairs/mm at 50% modulation transfer function with PCD matrix decomposition to 1. 75 line pairs/mm at 50% MTF. Overall, this study shows that our DL strategy with high-dose multi-EID derived decomposition labels is effective at creating more accurate material maps from PCD results.

Source link: https://ui.adsabs.harvard.edu/abs/2022PMB....67o5003N/abstract


Cumulative radiation exposure, effective and organ dose estimation from multiple head CT scans in stroke patients

This report sought to determine cumulative radiation exposure, safe, and organ dose from the recurrent computed tomography head scan during the stroke. A retrospective review of the picture archiving and messaging system for all the patients who underwent at least three head scans when admitted to the hospital due to stroke. respectively, the mean CTDIvol and DLP values per scan were 21 0. 8 and 429 u00b1 85, respectively. 6. 4, 8. 5, and 10. 7 mSv were respectively calculated as effective doses for patients with three, four, and five scans. The highest organ doses have been found in the brain and the lowest at the breast are consistent. Referring physician education and promoting request justification and dose optimization for patients exposed to frequent radiation exams is key to reducing patient exposure. In addition, the cumulative effective dose should be included in referrer, radiologists, and radiographers' preparations.

Source link: https://ui.adsabs.harvard.edu/abs/2022RaPC..19910306A/abstract


An extended primal-dual algorithm framework for nonconvex problems: application to image reconstruction in spectral CT

We propose an extended primal-dual algorithm framework for solving a variety of nonconvex and presumably nonsmooth optimization problems in spectral computed tomography image reconstruction using the convexity of each component of the forward operator. In addition, when the suggested plans are applied to a particular problem in spectral CT image reconstruction, namely, total variation regularized nonlinear least-squares problem without nonnegative constraint, we can also show the specific convergence for these schemes by using some unique features.

Source link: https://ui.adsabs.harvard.edu/abs/2022InvPr..38h5011G/abstract


Feasibility study of three-material decomposition in dual-energy cone-beam CT imaging with deep learning

In this work, a dedicated end-to-end deep convolution neural network, coded as Triple-CBCT, is shown to show the possibility of reconstructing three separate material distribution volumes from the dual-energy CBCT projection results. This Triple-CBCT network was developed by numerically synthesized dual-energy CBCT data, and it was validated on an in-house benchtop device with experimental dual-energy CBCT results of the Iodine-CaCl 2 solution and pig leg specimens. Both the sinogram and CT image domains can be used together to improve the decomposition quality of several products from the dual-energy projections, according to results.

Source link: https://ui.adsabs.harvard.edu/abs/2022PMB....67n5012Z/abstract


Crossbar-Net: A Novel Convolutional Neural Network for Kidney Tumor Segmentation in CT Images

Accurate segmentation of kidney tumor in CT images is a difficult and difficult task due to the continuous motion, similar appearance, and various shapes. Firstly, considering that the traditional learning-based segmentation techniques normally use either whole images or squared patches as the training samples, we experimentally sampled the orthogonal non-squared patches to completely cover the whole kidney tumors in either horizontal or vertical directions. These sampled crossbar patches can not only display detailed local information about kidney disease as the common patches, but also describe the global appearance from either horizontal or vertical direction using contextual data. Second, we developed a convolutional neural network with two sub-models in a cascaded manner in order to integrate the segmentation results from two directions. We'll try our method on a real CT kidney tumor registry database, which includes 3,500 images.

Source link: https://ui.adsabs.harvard.edu/abs/2019ITIP...28.4060Y/abstract


Real-Time Local Noise Filter in 3-D Visualization of CT Data

To enhance the final display's quality, remove noise in computer tomography data for real-time 3D visualization. However, the CT noise cannot be removed by straight averaging because the noise has a broadband spatial frequency that is overlapping with the latest signal frequencies. Our filter is compared to four other similar spatially similar filters in terms of entropy and processing time.

Source link: https://ui.adsabs.harvard.edu/abs/2019ITNS...66.1296T/abstract


Visual Attention Network for Low-Dose CT

Low dose CT data acquisition is largely responsible for noise and artifacts acquisition, and will have a huge effect on imaging results. To give a competitive advantage ahead of the noise distribution, our main aim is to bring visual attention to the learning process of GAN. Both the generator and discriminator networks are boosted with visual attention so they will not only pay particular attention to noisy areas and nearby buildings, but also investigate the local consistency of the recovered regions, which in doing so.

Source link: https://ui.adsabs.harvard.edu/abs/2019ISPL...26.1152D/abstract


Clinical CT densitometry for wooden cultural heritage analysis validated by FTIR and Raman spectroscopies

The identification and characterization of wood products are an ongoing challenge to archaeological artefacts of cultural heritage. The X-ray Computed Tomography (CT medical scanner) was used in this project for the first time in this series to distinguish three types of wood, which were traditionally used in the construction of heritage objects. In CT photos processed through the volumetric rendering process, wood interior features have been identified, without the need to destructively "take out" a piece of the sample.

Source link: https://ui.adsabs.harvard.edu/abs/2022RaPC..19910376L/abstract


In-Situ/Operando X-Ray CT Characterisation of Lithium-Ion Pouch Cells during Thermal Failure

High-speed imaging with in-situ/operando X-ray CT has been used extensively to investigate various lithium-ion battery safety characteristics and failure mechanisms [2][3], including thermal damage [4]. Several modifications to a cell structure leading to thermal runaway that can take minutes are often missed during lithium-ion battery failure, and as a result, many changes to a cell structure leading to thermal runaway can take minutes, and as a result, are often missed. Investigations are looking into the failure mechanisms inside a completely charged LiCoO 2 cathode and graphite anode pouch cell rated at 210 mAh.

Source link: https://ui.adsabs.harvard.edu/abs/2022ECSMA2022..349P/abstract

* Please keep in mind that all text is summarized by machine, we do not bear any responsibility, and you should always check original source before taking any actions

* Please keep in mind that all text is summarized by machine, we do not bear any responsibility, and you should always check original source before taking any actions