Hairong Zheng’s research while affiliated with University of Chinese Academy of Sciences and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (789)


Wave‐CAIPI Multiparameter MR Imaging in Neurology
  • Article

January 2025

·

14 Reads

NMR in Biomedicine

·

Yongquan Ye

·

Yifan Guo

·

[...]

·

Haifeng Wang

In clinical practice, particularly in neurology assessments, imaging multiparametric MR images with a single‐sequence scan is often limited by either insufficient imaging contrast or the constraints of accelerated imaging techniques. A novel single scan 3D imaging method, incorporating Wave‐CAIPI and MULTIPLEX technologies and named WAMP, has been developed for rapid and comprehensive parametric imaging in clinical diagnostic applications. Featuring a hybrid design that includes wave encoding, the CAIPIRINHA sampling pattern, dual time of repetition (TR), dual flip angle (FA), multiecho, and optional flow modulation, the WAMP method captures information on RF B1t fields, proton density (PD), T1, susceptibility, and blood flow. This method facilitates the synthesis of multiple qualitative contrast‐weighted images and relaxometric parametric maps. A single WAMP scan generates multiple contrast‐weighted images and relaxometric parametric maps, including PD‐weighted (PDW), T1‐weighted (T1W), T2*‐weighted (T2W), adjusted T1‐weighted (aT1W), susceptibility‐weighted imaging (SWI), B1t map, T1 map, T2/R2* map, PD map, and quantitative susceptibility mapping (QSM). Both phantom and in vivo experiments have demonstrated that the proposed method can achieve high image quality and quantification accuracy even at high acceleration factors of 4 and 9. The experiments have confirmed that the rapid single scan method can be effectively applied in clinical neurology, serving as a valuable diagnostic tool for conditions such as pediatric tuberous sclerosis complex (TSC)‐related epilepsy, adult Parkinson's disease, and suspected stroke patient. The WAMP method holds substantial potential for advancing multiparametric MR imaging in clinical neurology, promising significant improvements in both diagnostic speed and accuracy.


A bibliometric analysis of research publications on NIR-II imaging agents and ICG and its derivatives as NIR-II fluorophores. (A) Annual publications and citations of articles reporting NIR-II imaging agents from 2000 to 2024. (B) Annual number of highly cited papers and their percentage among annual publications on NIR-II imaging agents. (C) Total number of highly cited papers and corresponding citations from the most prolific authors studying NIR-II imaging. (D) Annual publications and citations of articles reporting ICG and its derivatives for NIR-II imaging from 2010 to 2024.
Schematic illustration of ICG-based probes for NIR-II fluorescence imaging and image-guided therapy in preclinical and clinical settings.
(A) Chemical structures of currently FDA-approved fluorescent molecular imaging probes. (B) Timeline of ICG for different applications after approval by the FDA.
Molecular engineering of ICG to fabricate its similarities.
ICG for NIR-II fluorescence imaging. (A) Full emission spectrum of ICG and the NIR-II photograph of the ICG solution. The InGaAs detector can recover the true emission tail of ICG. (B) Fluorescence signal of ICG in the NIR-II region. (C) NIR-I (up) and NIR-II (down) brain vascular fluorescence imaging using ICG and different long-pass filters. Reprinted from [30] with permission from the National Academy of Sciences. (D) Scheme of the hypotension process and (E) dynamic bioimaging of carotid artery beyond 1,500-nm window after Isoket administration. Reprinted from [87] with permission from the American Chemical Society.

+5

Recent Advances in Indocyanine Green-Based Probes for Second Near-Infrared Fluorescence Imaging and Therapy
  • Literature Review
  • Full-text available

January 2025

·

25 Reads

Fluorescence imaging, a highly sensitive molecular imaging modality, is being increasingly integrated into clinical practice. Imaging within the second near-infrared biological window (NIR-II; 1,000 to 1,700 nm), also referred to as shortwave infrared, has received substantial attention because of its markedly reduced autofluorescence, deeper tissue penetration, and enhanced spatiotemporal resolution as compared to traditional near-infrared (NIR) imaging. Indocyanine green (ICG), a US Food and Drug Administration-approved NIR fluorophore, has long been used in clinical applications, including blood vessel angiography, vascular perfusion monitoring, and tumor detection. Recent advancements in NIR-II imaging technology have revitalized interest in ICG, revealing its extended tail fluorescence beyond 1,000 nm and reaffirming its potential as a clinically translatable NIR-II fluorophore for in vivo imaging and theranostic applications for diagnosing various diseases. This review emphasizes the notable advances in the use of ICG and its derivatives for NIR-II imaging and image-guided therapy from both fundamental and clinical perspectives. We also provide a concise conclusion and discuss the challenges and future opportunities with NIR-II imaging using clinically approved fluorophores.

Download

Myocardial T1 mapping at 5T using multi-inversion recovery real-time spoiled GRE

January 2025

·

7 Reads

Purpose: To develop an accurate myocardial T1 mapping technique at 5T using Look-Locker-based multiple inversion-recovery with the real-time spoiled gradient echo (GRE) acquisition. Methods: The proposed T1 mapping technique (mIR-rt) samples the recovery of inverted magnetization using the real-time GRE and the images captured during diastole are selected for T1 fitting. Multiple-inversion recoveries are employed to increase the sample size for accurate fitting. Furthermore, the inversion pulse (IR) was tailored for cardiac imaging at 5T, optimized to maximize the inversion efficiency over specified ranges of B1 and off-resonance. The T1 mapping method was validated using Bloch simulation, phantom studies, and in 16 healthy volunteers at 5T. Results: The optimized IR pulse based on the tangent/hyperbolic tangent pulse was found to outperform the conventional hyperbolic secant IR pulse within a limited peak amplitude of 10.6 {\mu}T at the 5T scanner. This optimized IR pulse achieves an average inversion factor of 0.9014 within a B0 range of +/-250Hz and a B1 range of -50% to 20%. In both simulation and phantom studies, the T1 values measured by mIR-rt closely approximate the reference T1 values, with errors less than 3%, while the conventional MOLLI sequence underestimates T1 values. The myocardial T1 values at 5T are 1553 +/- 52 ms, 1531 +/- 53 ms, and 1526 +/- 60 ms (mean +/- standard deviation) at the apex, middle, and base, respectively. Conclusion: The proposed method is feasible for myocardial T1 mapping at 5T and provides better accuracy than the conventional MOLLI sequence. Keywords: Myocardial T1 mapping, 5T, Look-Locker


In situ structural-functional synchronous dissection of dynamic neuromuscular system via an integrated multimodal wearable patch

January 2025

·

48 Reads

Science Advances

Neuromuscular abnormality is the leading cause of disability in adults. Understanding the complex interplay between muscle structure and function is crucial for effective treatment and rehabilitation. However, the substantial deformation of muscles during movement (up to 40%) poses challenges for accurate assessment. To address this, we developed a wearable structural-functional sensing patch (WSFP) that enables synchronous analysis of muscle structure and function. The WSFP incorporates a soft, stretchable electrode array for high-performance electrophysiological monitoring with low contact impedance and high stability. Its innovative design absorbs skin deformation stress, ensuring stable adhesion of a flexible ultrasound transducer array, offering higher-fidelity imaging. With dynamic tissue imaging, it allows real-time visualization of muscle structure. The WSFP achieves superior accuracy in dynamic action recognition and disease assessment compared to single-modal methods, maintaining stable operation during motion for up to 72 hours. This study advances neuromuscular system analysis and improves diagnostic precision.


Dual-Uncertainty Guided Multimodal MRI-Based Visual Pathway Extraction

January 2025

·

6 Reads

IEEE transactions on bio-medical engineering

Objective: This study aims to accurately extract the visual pathway (VP) from multimodal MR images while minimizing reliance on extensive labeled data and enhancing extraction performance. Method: We propose a novel approach that incorporates a Modality-Relevant Feature Extraction Module (MRFEM) to effectively extract essential features from T1-weighted and fractional anisotropy (FA) images. Additionally, we implement a mean-teacher model integrated with dual uncertainty-aware ambiguity identification (DUAI) to enhance the reliability of the VP extraction process. Results: Experiments conducted on the Human Connectome Project (HCP) and Multi-Shell Diffusion MRI (MDM) datasets demonstrate that our method reduces annotation efforts by at least one-third compared to fully supervised techniques while achieving superior extraction performance over six state-of-the-art semi-supervised methods. Conclusion: The proposed label-efficient approach alleviates the burdens of manual annotation and enhances the accuracy of multimodal MRI-based VP extraction. Significance: This work contributes to the field of medical imaging by facilitating more efficient and accurate visual pathway extraction, thereby improving the analysis and understanding of complex brain structures with reduced reliance on expert annotation.


Score-based Diffusion Models with Self-supervised Learning for Accelerated 3D Multi-contrast Cardiac MR Imaging

January 2025

·

8 Reads

IEEE Transactions on Medical Imaging

Long scan time significantly hinders the widespread applications of three-dimensional multi-contrast cardiac magnetic resonance (3D-MC-CMR) imaging. This study aims to accelerate 3D-MC-CMR acquisition by a novel method based on score-based diffusion models with self-supervised learning. Specifically, we first establish a mapping between the undersampled k-space measurements and the MR images, utilizing a self-supervised Bayesian reconstruction network. Secondly, we develop a joint score-based diffusion model on 3D-MC-CMR images to capture their inherent distribution. The 3D-MC-CMR images are finally reconstructed using the conditioned Langenvin Markov chain Monte Carlo sampling. This approach enables accurate reconstruction without fully sampled training data. Its performance was tested on the dataset acquired by a 3D joint myocardial T1 and T mapping sequence. The T1 and T maps were estimated via a dictionary matching method from the reconstructed images. Experimental results show that the proposed method outperforms traditional compressed sensing and existing self-supervised deep learning MRI reconstruction methods. It also achieves high quality T1 and T parametric maps close to the reference maps, even at a high acceleration rate of 14.


Reproducibility of automatic adipose tissue segmentation using proton density fat fraction images between 1.5 and 3.0 T magnetic resonance

January 2025

·

12 Reads

Quantitative Imaging in Medicine and Surgery

Background Deep learning (DL)-based adipose tissue segmentation methods have shown great performance and efficacy for adipose tissue distribution analysis using magnetic resonance (MR) images, an important indicator of metabolic health and disease. The aim of this study was to evaluate the reproducibility of whole-body adipose tissue distribution analysis using proton density fat fraction (PDFF) images at different MR strengths. Methods A total of 24 volunteers were imaged using both 1.5 and 3.0 T clinical MR imaging (MRI) scanners at two sites. Whole-body PDFF images were acquired covering from neck to knee, and grouped into three subparts: thorax, abdomen, and thigh. The PDFF images were then segmented automatically into subcutaneous adipose tissue (SAT) and internal adipose tissue (IAT) using a U-Net DL model. The volumes of whole body (WH), total adipose tissue (TAT), SAT, and IAT for total body and each subpart were measured, and the volume ratio of TAT/WH, SAT/WH, IAT/WH, SAT/TAT, and IAT/SAT were also calculated. Additionally, the reproducibility of PDFF values of SAT and IAT for total body and subparts were evaluated. Results The intraclass correlation coefficient (ICC) and Pearson correlation coefficient of these volumes and volume ratios in whole-body between the two scanners were very close to one. The paired t-test and Bland-Altman plots for all comparisons showed no significant differences (P>0.05) when comparing the results from the 1.5 T scanner minus those from the 3.0 T scanner. The mean bias for WH, TAT, SAT, and IAT was −6.89 cm³ (P=0.95), −67.21 cm³ (P=0.40), 19.31 cm³ (P=0.74), and −18.84 cm³ (P=0.69), respectively. Good reproducibility performances were also found in each subpart, except for the indices of IAT volume, TAT/WH ratio, and SAT/TAT ratio in the thorax due to different susceptibility effects across MR strengths. The results also demonstrated good reproducibility between PDFF values of the two scanners with the mean bias for WH, thorax, abdomen, and thigh being −0.19% (P=0.219), −0.30% (P=0.118), 0.086% (P=0.494), and 0.24% (P=0.186) for SAT, respectively, as well as 0.35% (P=0.136), 0.46% (P=0.150), 0.58% (P=0.255), and 0.40% (P=0.169) for IAT, respectively. Conclusions Good reproducibility of whole-body adipose tissue distribution analysis using the DL method between 1.5 and 3.0 T MR images was demonstrated, which may facilitate the whole-body adipose tissue distribution analysis using the quantitative MR-PDFF images.


Prompt-Agent-Driven Integration of Foundation Model Priors for Low-Count PET Reconstruction

January 2025

·

17 Reads

IEEE Transactions on Medical Imaging

Low-count Positron Emission Tomography reconstruction is critical for maintaining high imaging quality while minimizing tracer doses and radiation exposure. Although integrating structural information from CT and MR data has been shown to enhance PET reconstruction, this typically requires simultaneous PET and CT/MRI scans, complicating workflows and increasing radiation exposure. Recent advancements in foundation models offer a promising alternative to in-person CT/MRI imaging, potentially overcoming these limitations. However, the use of foundation models’ segmentation masks as semantic guides has been observed to introduce erroneous structures in low-count PET reconstructions. To address this challenge, this work introduces an innovative prompting agent-based framework that dynamically interacts with the foundation model to retrieve and refine priors, minimizing undue influence on the reconstruction process. Specifically, a box agent is designed for single-instance local area information retrieval, while a point agent is introduced to progressively prompt broader semantic structures globally, utilizing history point prompts. Additionally, an MDP paradigm has been developed to address the challenges of utilizing historical point prompts while maintaining the independence required by MDPs. Evaluated on both simulated and real datasets, the proposed method demonstrates superior qualitative and quantitative performance compared to state-of-the-art methods, even those leveraging in-person CT/MRI priors.


Measurement of muscle glycogen at 5T
a, b Spectral quantification and concentration calibration of glycoNOE from Z-spectra of glycogen at 26 nm in vitro (pH 7.3, 37 °C) both in neat solution and in agar to simulate tissue. The average and standard values of glycoNOE(26 nm) signals were obtained from two separate acquisitions (n = 2 acquisitions). c The relative glycoNOE signal (α\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{\alpha }}}$$\end{document}) as a function of particle size. The signals were normalized using glycoNOE at 26 nm. The curve was fitted using Eq. 2. d Extraction of in vivo glycoNOE signal from the red area under the curve of the difference signal between Z-spectrum (black circles) with Lorentzian water line fit (blue dashes). e The glycoNOE signal in vivo as a function of B1 (n = 16 subjects), showing a reduction above 0.2 µT due to background interference. f Outline of region of interest in human skeletal muscle used for the spectra. g, h The B0 and B1 maps. i Glycogen concentration map over a slice through the calf, based on glycoNOE after B0 and B1 correction and calibrated using (b). Slice thickness = 5 mm, B1 = 0.2 μT, tsat = 3 s, TR = 4 s. Data in (b) and (e) are presented as means ±  SD. Source data are provided as a Source Data file.
In vivo mapping of glycogen levels in human skeletal muscle before and after 31-min lower intensity plantar flexion exercise
Whole muscle glycoNOE signals were fitted from the Z-spectra for a representative subject (#1, male) at a rest, b 13 min post exercise, and c 6 h post exercise. d Muscle assignment from T2w MRI and glycogen concentration maps before and after exercise obtained as a function of time. e Glycogen levels as a function of time in muscle groups labeled on the T2W image. f Map of amount of glycogen depletion after exercise: Δ[Gly(0)]depex=[Gly]base−[Gly0]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta [{{\rm{Gly}}}(0)]}_{{{\rm{dep}}}}^{{{\rm{ex}}}}={[{{\rm{Gly}}}]}^{{{\rm{base}}}}-[{{\rm{Gly}}}\left(0\right)]$$\end{document}. g Map of amount of glycogen repletion over the first 3 h: Δ[Gly(3hr)]rep=[Gly(3hr)]−[Gly(0)]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {[{{\rm{Gly}}}(3{{\rm{hr}}})]}_{{{\rm{rep}}}}=[{{\rm{Gly}}}(3{{\rm{hr}}})]-[{{\rm{Gly}}}(0)]$$\end{document}. h Map of glycogen initial repletion rate Rrepinitial\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{{\rm{rep}}}}^{{{\rm{initial}}}}$$\end{document} of the exponential recovery. i, j, k Group average (n = 14 subjects per muscle group) of glycogen levels as a function of time for LG, MG, and soleus. Group analysis (n = 14 subjects per muscle group) for different muscle groups of glycogen l depletion amount Δ[Gly(0)]depex\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta [{{\rm{Gly}}}(0)]}_{{{\rm{dep}}}}^{{{\rm{ex}}}}$$\end{document} and m linear repletion rate Rreplinear\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{{\rm{rep}}}}^{{{\rm{linear}}}}$$\end{document}. Data are presented as mean ±  SD. Statistical significance was determined using two-tailed unpaired Wilcoxon rank-sum test. p < 0.05 were considered statistically significant. n Correlation and linear regression between glycogen depletion and glycogen repletion over 3 h post exercise (n = 14 subjects). The data points were indicated using the same color coding as in (m) for different muscle groups. The gray shadows in (i–k) indicate SD, and in (n) indicate 95% confidence interval (CI). Slice thickness = 5 mm, B1 = 0.2 μT, tsat = 3 s, TR = 4 s. LG lateral gastrocnemius, MG medial gastrocnemius. Source data are provided as a Source Data file.
In vivo mapping of glycogen levels in human skeletal muscle before and after 31-min higher intensity plantar flexion exercise
Muscle a T2w MRI and maps of b glycogen at t = 0, 3 h post exercise, c glycogen depletion, and d repletion over first 3 h for a representative subject (#15). e–g Glycogen levels as a function of time in muscle groups labeled on the T2W image. Group average (n = 14 subjects per muscle group) of glycogen levels as a function of time for h LG, i MG, and j soleus. Group analysis (n = 14 subjects per muscle group) for different muscle groups of glycogen k depletion Δ[Gly(0)]depex\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta [{{\rm{Gly}}}(0)]}_{{{\rm{dep}}}}^{{{\rm{ex}}}}$$\end{document} and l linear repletion rate Rreplinear\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{{\rm{rep}}}}^{{{\rm{linear}}}}$$\end{document}. Data are presented as means ± SD. Statistical significance was determined using two-tailed unpaired Wilcoxon rank-sum test. p < 0.05 were considered statistically significant. m Correlation and linear regression between glycogen depletion and glycogen repletion over first 3 h post exercise (n = 14 subjects). The data points were indicated using the same color coding as in (l) for different muscle groups. The gray shadows in (h–j) indicate SD, and in (m) indicate 95% CI. Slice thickness = 20 mm, B1 = 0.2 μT, tsat = 1775 ms, TR = 2 s. Source data are provided as a Source Data file.
Regional glycogen depletion and repletion patterns in muscle after higher intensity exercise
a T2W image in the axial view showing regions with the three types of glycogen repletion. The estimated initial rate (Rrep1initial\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{{\rm{rep}}}1}^{{{\rm{initial}}}}$$\end{document} and Rrep2initial\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{{\rm{rep}}}2}^{{{\rm{initial}}}}$$\end{document}) maps in b phase I and c phase II, respectively, for the multiphase recoveries (types b and c) in subject #15. d T2W images with overlay showing different types of repletion along the length of MG and soleus for a representative subject (#23). The measured glycogen levels as a function of time in muscle regions with type a (e), type b (f), and type c (g) kinetics, as indicated in (a). h–j Group average (n = 14, 7, 9, and 12 subjects for type a slow, type a rapid, type b, and type c, respectively) of glycogen levels as a function of time in regions with type a, b, and c kinetics. k Group analysis (n = 14, 7, 9, and 12 subjects for type a slow, type a rapid, type b, and type c, respectively) of glycogen depletion amount Δ[Gly(0)]depex\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta [{{\rm{Gly}}}(0)]}_{{{\rm{dep}}}}^{{{\rm{ex}}}}$$\end{document} for the three types of kinetics. Data are presented as mean ± SD. Statistical significance was determined using two-tailed unpaired Wilcoxon rank-sum test. p < 0.05 were considered statistically significant. Correlation and linear regression between glycogen depletion and initial repletion rates of phase I (l) and glycogen repletion over 3 h post exercise (m) (n = 14 subjects). The data points were indicated using the same color coding as in (a) for the respective types of kinetics. The gray shadows indicate SD in (h–j) and 95% CI in (l, m). Source data are provided as a Source Data file.
In vivo imaging of glycogen in human muscle

December 2024

·

89 Reads

Probing regional glycogen metabolism in humans non-invasively has been challenging due to a lack of sensitive approaches. Here we studied human muscle glycogen dynamics post-exercise with a spatial resolution of millimeters and temporal resolution of minutes, using relayed nuclear Overhauser effect (glycoNOE) MRI. Data at 5T showed a homogeneous distribution of glycogen in resting muscle, with an average concentration of 99 ± 13 mM. After plantar flexion exercise following fasting with recovery under fasting conditions, the calf muscle showed spatially heterogeneous glycogen depletion and repletion kinetics that correlated with the severity of this depletion. Three types of regional glycogen kinetics were observed: (i) single exponential repletion (type a); (ii) biphasic recovery of rapid repletion followed by additional depletion (type b); (iii) biphasic recovery where continued depletion is followed by an exponential recovery (type c). The study of the complex patterns of glycogen kinetics suggests that glycogen breakdown may be quantitatively important during the initial recovery.


On the origin of MTF reduction in grating‐based x‐ray differential phase contrast CT imaging

December 2024

·

16 Reads

Background The complementary absorption contrast CT (ACT) and differential phase contrast CT (DPCT) can be generated simultaneously from an x‐ray computed tomography (CT) imaging system incorporated with grating interferometer. However, it has been reported that ACT images exhibit better spatial resolution than DPCT images. By far, the primary cause of such discrepancy remains unclear. Purpose The purpose of this study is to investigate the underlying cause of the resolution discrepancy between ACT and DPCT in a grating interferometer CT imaging system. Methods In this study, theoretical derivations were performed with a ππ\pi‐phase Talbot–Lau grating interferometer system to model the signal formation mechanism of absorption imaging and phase imaging, respectively. In addition, physical, and numerical experiments were conducted to verify the theoretical findings and assess the resolution discrepancy between ACT and DPCT under various conditions. Herein, the ACT and DPCT images were reconstructed from the filtered‐back‐projection algorithm using a standard Ramp filter and a standard Hilbert filter, respectively. Results Experiments demonstrated that the spatial resolution of ACT and DPCT images are primarily impacted by the beam diffraction induced signal splitting. In particular, lower modulation transfer function (MTF) was observed for DPCT than ACT due to the opposite‐superposition of phase signals. In addition, factors such as focal spot size, beam spectra, object composition, sample size, and detector pixel size were found to have minor impacts on the MTFs of both ACT and DPCT. Conclusions In conclusion, this study reveals that the opposite‐superposition of split phase signals causes the spatial resolution reduction in DPCT imaging.


Citations (27)


... While achieving good performance, further improvements might be achieved by incorporating these constraints, along with kspace consistency, during the TSMI reconstruction process itself (see e.g. [22,23,54] for non-DDPM based MRF models that account for these constraints, also [55,56] for MRI reconstructions DDPM models that account for k-space consistency). ...

Reference:

Denoising Diffusion Probabilistic Models for Magnetic Resonance Fingerprinting
SPIRiT-Diffusion: Self-Consistency Driven Diffusion Model for Accelerated MRI
  • Citing Article
  • October 2024

IEEE Transactions on Medical Imaging

... While it does not yet achieve state-of-the-art performance, it serves as a valuable benchmark and provides open-source code, facilitating further research [27]. Furthermore, the tri-orientated Mamba (ToM) module [28] is proposed for high-dimensional medical image segmentation tasks, and Swin-UMamba [29] leverages pre-trained models. These innovations not only improve the accuracy of medical image segmentation but also introduce novel solutions to the field of medical image analysis. ...

Swin-UMamba: Mamba-Based UNet with ImageNet-Based Pretraining

... In the mouse and monkey models of epilepsy, ultrasound stimulation of epileptogenic regions located in the hippocampus or frontal lobe has been able to reduce seizure severity Zou et al., 2020;Zou et al., 2021). Even extending to the peripheral nervous system, ultrasound modulation of the vagus nerve has exerted considerable anti-epileptic effects (Zou et al., 2024). In human motor cortex (M1), similar to findings in isolated brain slices, neuronal excitation/inhibition ratios can also be non-invasively modulated by ultrasound (Zhang et al., 2023). ...

Noninvasive closed-loop acoustic brain-computer interface for seizure control

Theranostics

... By incorporating granular implementation details, providing quantitative results, and utilizing visual aids, this section effectively showcases the practical application and effectiveness of our proposed methodologies. These enhancements not only validate the research but also provide valuable resources and insights for domain experts seeking to adopt and extend privacy-preserving data mining techniques in their respective fields (Huang et al., 2024). ...

Enhancing representation in radiography-reports foundation model: a granular alignment algorithm using masked contrastive learning

... Unsupervised Local Contrastive Learning Correlating a dense visual representation with finegrained semantic meaning is not only helpful for image understanding but vital to tasks like semantic segmentation. Recent work address this problem in the challenging unsupervised scenario [19,51,59,52,31,57,40,30]. Some methods rely on a pre-trained object detector or segmentation model to extract the region of interest [57]. ...

Mlip: Medical Language-Image Pre-Training With Masked Local Representation Learning
  • Citing Conference Paper
  • May 2024

... One promising solution is to develop medical foundation models that can handle multiple clinical applications simultaneously and leverage pre-trained models to reduce the dependency on large annotated datasets [5][6][7][8][9][10][11] . These models can be trained on diverse and representative image-based datasets using self-supervised methods that do not require annotations, allowing them to learn robust and transferable feature representations that can be used across various tasks and domains 12,13 . By incorporating simple task-based heads with the well-learned feature representations from the foundation model, these methods can achieve good performance in specific tasks without the need for extensive manual annotations 14 . ...

Enhancing Representation in Medical Vision-Language Foundation Models Via Multi-Scale Information Extraction Techniques
  • Citing Conference Paper
  • May 2024

... As the medical field is a significant branch for VLM-based applications [22]- [24], numerous methods have recently emerged to effectively learn visual and textual information (e.g. radiology reports) using VLMs to enhance the performance of medical imaging analysis [25]- [28], medical report generation [29]- [31], and medical visual question answering (Med-VQA) [32], [33]. Nevertheless, the dominant settings were single-turn [34], [35] or multi-turn stitched by independent single-turn pairs of visual questions and textual answers [36], [37]. ...

MAKEN: Improving Medical Report Generation with Adapter Tuning and Knowledge Enhancement in Vision-Language Foundation Models
  • Citing Conference Paper
  • May 2024

... However, word-level alignment is constrained by contextual variability, hindering accurate capture of pathological descriptions. To overcome this limitation, Yang et al. [7] proposed a framework combining global and local contrastive learning, it leverages global features while precisely aligning sentencelevel features with local image regions. ...

Multimodal Self-Supervised Learning for Lesion Localization
  • Citing Conference Paper
  • May 2024

... This phenomenon could be attributed to several underlying mechanisms. First, the progressive hypertrophic and fibrotic remodeling of the left ventricle associated with chronic hypertension may directly compromise both longitudinal and circumferential myocardial contractility, resulting in a concurrent decline in both strain parameters (22)(23)(24). Second, prolonged or poorly controlled hypertension may overwhelm the heart's compensatory mechanisms, particularly in patients with advanced or long-standing disease, where the expected compensatory Frontiers in Cardiovascular Medicine enhancement of GCS fails to occur (25)(26)(27). ...

Machine Learning in Hypertrophic Cardiomyopathy
  • Citing Article
  • July 2024

JACC Cardiovascular Imaging

... Previous studies have demonstrated the utility of multimodality data for DL-based LC PET image denoising (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). Xu et al. (26) incorporated multi-contrast information from simultaneous magnetic resonance imaging (MRI) for ultra-LC PET denoising, reducing the dose level to 1/200 FC. ...

Accurate Whole-Brain Image Enhancement for Low-Dose Integrated PET/MR Imaging Through Spatial Brain Transformation
  • Citing Article
  • May 2024

IEEE Journal of Biomedical and Health Informatics