What are the latest advances in image reconstruction algorithms for computed tomography?
Best insight from top research papers
Recent advances in image reconstruction algorithms for computed tomography (CT) include the development of deep learning-based methods derived from maximum a posteriori (MAP) estimation, which can produce decent sparse-view CT images . Another advancement is the use of backprojection-filtration (BPF)-based algorithms, such as S-BPF and D-BPF, for multiple source translation CT (mSTCT), which can achieve high-resolution reconstruction with fewer projections compared to virtual projection-based filtered backprojection (V-FBP) . Iterative reconstruction algorithms, both backward projection and both backward and forward projections, have also been developed to enable low-dose CT acquisitions with high image quality . Additionally, reconstruction methods based on artificial intelligence (AI) have emerged, aiming to construct high-quality images using low doses and fast reconstruction speed . These AI-based methods have the advantage of attaining multiple goals in one shot, but they also have limitations such as the requirement for large datasets, unstable performance, and weak generalizability .
Answers from top 5 papers
More filters
| Papers (5) | Insight |
|---|---|
01 Jan 2023 | The paper discusses the latest advances in image reconstruction algorithms for computed tomography, specifically focusing on the evolution from conventional reconstruction methods to artificial intelligence-based reconstruction methods. |
| The paper discusses the advances in image reconstruction algorithms for computed tomography, specifically focusing on iterative reconstruction techniques. | |
30 May 2023 | The latest advances in image reconstruction algorithms for computed tomography discussed in the paper are the virtual projection-based filtered backprojection (V-FBP) algorithm and the backprojection-filtration (BPF)-based algorithms, specifically the derivatives along source (S-BPF) and derivatives along detector (D-BPF) algorithms. |
Open access•Posted Content 23 Jun 2023 | The provided paper discusses a deep learning-based image reconstruction method for computed tomography (CT) that utilizes maximum a posteriori (MAP) estimation. It does not mention any other specific advances in image reconstruction algorithms for CT. |
1 Citations | The latest advances in image reconstruction algorithms for computed tomography discussed in the paper are the virtual projection-based filtered backprojection (V-FBP) algorithm and the backprojection-filtration (BPF)-based algorithms, specifically the S-BPF and D-BPF algorithms. |
Related Questions
How effective is machine learning in the area of tomographic reconstruction?4 answersMachine learning has shown effectiveness in the area of tomographic reconstruction. It has been used to enhance the efficiency of obtaining high-quality tomographic pictures in electrical impedance tomography (EIT). Machine learning techniques are utilized to translate voltage measurements into reconstruction pictures, and the selection of model hyperparameters plays a crucial role in the quality of the reconstruction. Various machine learning algorithms, such as k-nearest neighbors and deep learning, have been explored for tomographic reconstruction. These approaches have shown promising results in improving the accuracy and reliability of tomographic image reconstruction, even in challenging scenarios such as limited-angle and sparse-view data. Overall, machine learning has proven to be effective in enhancing tomographic reconstruction techniques and has the potential to further improve signal-to-noise ratio, spatial resolution, and computational efficiency in various applications.
What are DRR techniques for X-Ray rendering from CT Scan ?5 answersDeep learning techniques have been proposed to address the challenge of reconstructing a universal image from one filtered backprojection (FBP) image in CT scans. Wavelet-based compression algorithms such as Embedded Zero-Tree Wavelet (EZW), Set Partitioning in Hierarchical Trees (SPIHT), and Wavelet Difference Reduction (WDR) have been used to compress X-ray and CT images. Sparse-angle tomography is another approach for obtaining 3D reconstructions from limited data in CT scans. Techniques such as adaptive sampling, sparse casting, regula falsi method, Monte-Carlo ambient occlusion estimation, screen-space ambient occlusion, and depth of field have been used to improve the rendering quality of large volumetric data sets obtained from CT or MRI scans.
What are the most recent advances in image processing algorithms?5 answersRecent advances in image processing algorithms include the development of a library that allows students to learn practical methods in image processing and introduces them to image ideas. Generative adversarial networks (GANs) have also emerged as a powerful tool in image processing, helping to generate networks without losing properties of attributes such as face identity and orientation. Image processing techniques have been applied in the agricultural industry to enhance productivity, including plant detection, livestock detection, and recognition of vegetables and fruits. In the field of medical image processing, deep learning algorithms have shown improved performance in the classification of cancer cells, lesions, organ segmentation, and medical image enhancement. Pulse-coupled neural networks (PCNNs) have also been developed, aiming to imitate the information analysis process of the biological cortex, and have been widely used in image processing.
What are the recent advances in ultrasound imaging?5 answersRecent advances in ultrasound imaging include the use of artificial intelligence (AI) to assist in scanning, such as standard plane recognition and organ identification, extraction of clinical planes from 3D volumes, and guidance of acquisitions by humans or robots. Technological improvements have also led to the development of new tools and techniques in breast ultrasound, such as microvasculature imaging, elastography, contrast-enhanced imaging, and automated ultrasound. Ultrasound has been utilized to diagnose and treat superficial soft tissue lesions, providing non-invasive imaging for treatment planning, staging, and follow-up. Optoacoustic and ultrasound imaging have been combined to provide high-resolution images of deep-seated structures, with advances in image formation methods and the ability to map the speed of sound and acoustic attenuation in tissues. Tracking technologies have also been applied to ultrasound imaging, improving accuracy and intuitiveness in applications such as freehand 3D imaging, image fusion, and ultrasound-guided interventions.
What are the latest advances in machine learning?5 answersMachine learning has seen several recent advances. Optimal Transport has emerged as a probabilistic framework in machine learning, offering new solutions for generative modeling and transfer learning. Computational Optimal Transport has also developed, impacting machine learning practice. Another area of progress is the usage of machine learning for electrochemical sensors, particularly in the analysis of data generated by sensing and biosensing methods. Additionally, machine learning models have been widely used for predicting the toxicity of small molecules, aiding in drug discovery by filtering out molecules with a high probability of failing in clinical trials. These models have been applied to various toxic endpoints, such as acute oral toxicity, hepatotoxicity, and mutagenicity. Overall, these advancements highlight the growing importance and diverse applications of machine learning in different domains.
What are the latest advances in echocardiogram?5 answersRecent advances in echocardiography include the development of artificial intelligence (AI) models for accurate estimation of left ventricular ejection fraction (LVEF). AI can assist level 1 readers in interpreting systolic function on an echocardiogram, reducing interinstitutional variability. Automated segmentation and quantitative analysis of echocardiography images using AI techniques offer a solution to reduce the time crunch and reporting errors faced by echocardiographers. Echocardiography has become an essential imaging modality for structural heart disease interventions, providing guidance for pre-procedural selection, intra-procedural guidance, post-procedural evaluation, and long-term follow-up. The introduction of real-time 3D echocardiography imaging, enabled by advancements in computer and transducer technologies, has revolutionized cardiovascular ultrasound, allowing for real-time acquisition and presentation of cardiac structures from any view. These advancements have improved the accuracy, reliability, and efficiency of echocardiographic imaging, enhancing clinical decision-making and patient care.