“Select and retrieve via direct upsampling” network (SARDU-Net): a data-driven, model-free, deep learning approach for quantitative MRI protocol design

2020 ◽  
Author(s):  
Francesco Grussu ◽  
Stefano B. Blumberg ◽  
Marco Battiston ◽  
Lebina S. Kakkar ◽  
Hongxiang Lin ◽  
...  
Keyword(s):  
Goodness Of Fit ◽  
Diffusion Tensor ◽  
Protocol Design ◽  
Quantitative Mri ◽  
Data Driven ◽  
Acquisition Time ◽  
Model Free ◽  
Spherical Mean ◽  
Mri Protocol

AbstractPurposeWe introduce “Select and retrieve via direct upsampling” network (SARDU-Net), a data-driven framework for model-free quantitative MRI (qMRI) protocol design, and demonstrate it on in vivo brain and prostate diffusion-relaxation imaging (DRI).MethodsSARDU-Net selects subsets of informative measurements within lengthy pilot scans, without the requirement to identify tissue parameters for which to optimise for. The algorithm consists of a selector, identifying measurement subsets, and a predictor, estimating fully-sampled signals from the subsets. We implement both using deep neural networks, which are trained jointly end-to-end. We demonstrate the algorithm on brain (32 diffusion-/T1-weightings) and prostate (16 diffusion-/T2-weightings) DRI scans acquired on 3 healthy volunteers on two separate 3T Philips systems each. We used SARDU-Net to identify sub-protocols of fixed size, assessing the reproducibility of the procedure and testing sub-protocols for their potential to inform multi-contrast analyses via T1-weighted spherical mean diffusion tensor (T1-SMDT, brain) and hybrid multi-dimensional MRI (HM-MRI, prostate) modelling.ResultsIn both brain and prostate, SARDU-Net identifies sub-protocols that maximise information content in a reproducible manner across training instantiations. The sub-protocols enable multi-contrast modelling for which they were not optimised explicitly, providing robust T1-SMDT and HM-MRI maps and goodness-of-fit in the top 5% against extensive sub-protocol comparisons.ConclusionsSARDU-Net gives new opportunities to identify economical but informative qMRI protocols from a subset of the pilot scans that can be used for acquisition-time-sensitive applications. The simple architecture makes the algorithm easy to train when exhaustive searches are intractable, and applicable to a variety of anatomical contexts.

scholarly journals Feasibility of Data-Driven, Model-Free Quantitative MRI Protocol Design: Application to Brain and Prostate Diffusion-Relaxation Imaging

2021 ◽  
Vol 9 ◽  
Author(s):  
Francesco Grussu ◽  
Stefano B. Blumberg ◽  
Marco Battiston ◽  
Lebina S. Kakkar ◽  
Hongxiang Lin ◽  
...  
Keyword(s):  
Diffusion Tensor ◽  
Protocol Design ◽  
Quantitative Mri ◽  
Data Driven ◽  
Acquisition Time ◽  
Diffusion Relaxation ◽  
Model Free ◽  
Spherical Mean ◽  
Mri Protocol

Purpose: We investigate the feasibility of data-driven, model-free quantitative MRI (qMRI) protocol design on in vivo brain and prostate diffusion-relaxation imaging (DRI).Methods: We select subsets of measurements within lengthy pilot scans, without identifying tissue parameters for which to optimise for. We use the “select and retrieve via direct upsampling” (SARDU-Net) algorithm, made of a selector, identifying measurement subsets, and a predictor, estimating fully-sampled signals from the subsets. We implement both using artificial neural networks, which are trained jointly end-to-end. We deploy the algorithm on brain (32 diffusion-/T1-weightings) and prostate (16 diffusion-/T2-weightings) DRI scans acquired on three healthy volunteers on two separate 3T Philips systems each. We used SARDU-Net to identify sub-protocols of fixed size, assessing reproducibility and testing sub-protocols for their potential to inform multi-contrast analyses via the T1-weighted spherical mean diffusion tensor (T1-SMDT, brain) and hybrid multi-dimensional MRI (HM-MRI, prostate) models, for which sub-protocol selection was not optimised explicitly.Results: In both brain and prostate, SARDU-Net identifies sub-protocols that maximise information content in a reproducible manner across training instantiations using a small number of pilot scans. The sub-protocols support T1-SMDT and HM-MRI multi-contrast modelling for which they were not optimised explicitly, providing signal quality-of-fit in the top 5% against extensive sub-protocol comparisons.Conclusions: Identifying economical but informative qMRI protocols from subsets of rich pilot scans is feasible and potentially useful in acquisition-time-sensitive applications in which there is not a qMRI model of choice. SARDU-Net is demonstrated to be a robust algorithm for data-driven, model-free protocol design.

The effect of microvascular obstruction on the myocardial microstructure: a diffusion tensor imaging study

2021 ◽  
Vol 22 (Supplement_1) ◽  
Author(s):  
A Das ◽  
K Kelly ◽  
M Aldred ◽  
I Teh ◽  
CK Stoeck ◽  
...  
Keyword(s):  
Microvascular Obstruction ◽  
Diffusion Tensor ◽  
Spin Echo ◽  
Mean Diffusivity ◽  
Imaging Study ◽  
Free Breathing ◽  
Helix Angle ◽  
Acquisition Time ◽  
Heart Research

Abstract Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): Heart Research UK Background Diffusion tensor cardiac magnetic resonance (DT-CMR) imaging allows for characterising myocardial microstructure in-vivo using mean diffusivity (MD), fractional anisotropy (FA), secondary eigenvector angle (E2A) and helix angle (HA) maps. Following myocardial infarction (MI), alterations in MD, FA and HA proportions have previously been reported. E2A depicts the contractile state of myocardial sheetlets, however the behaviour of E2A in infarct segments, and all DTI markers in areas of microvascular obstruction (MVO) is also not fully understood.  Purpose We performed spin echo DTI in patients following ST-elevation MI (STEMI) in order to investigate acute changes in DTI parameters in remote and infarct segments both with and without MVO. Method Twenty STEMI patients (16 men, 4 women, mean age 59) had acute (5 ± 2d) 3T CMR scans. CMR protocol included: second order motion compensated (M012) free-breathing spin echo DTI (3 slices, 18 diffusion directions at b-values 100s/mm2[3], 200s/mm2[3] and 500s/mm2[12], reconstructed resolution was 1.66x1.66x8mm); cine and late gadolinium enhancement (LGE) imaging. Average MD, FA, E2A HA parameters were calculated on a  16 AHA segmental level. HA maps were described by dividing values into left-handed HA (LHM, -90° < HA < -30°), circumferential HA (CM, -30° < HA < 30°), and right-handed HA (RHM, 30° < HA < 90°) and reported as relative proportions. Segments were defined as infarct (positive for LGE) and remote (opposite to the infarct).  Results DTI acquisition was successful in all patients (acquisition time 13 ± 5mins). Ten patients had evidence of MVO on LGE images. MD was significantly higher in infarct regions in comparison to remote; MVO-ve infarct segments had significantly higher MD than MVO + ve infarct segments (MD remote= 1.46 ± 0.12x10-3mm2/s, MD MVO + ve = 1.59 ± 0.12x10-3mm2/s, MD MVO-ve  = 1.75 ± 0.12x10-3mm2/s, ANOVA p < 0.01). FA was reduced in infarct segments in comparison to remote; MVO-ve infarct segments had significantly lower FA than MVO + ve infarct segments (FAremote= 0.37 ± 0.02, FA MVO + ve = 0.31 ± 0.02 x 10-3mm2/s, MD MVO-ve =0.25 ± 0.02, ANOVA p < 0.01). E2A values were significantly lower in infarct segments compared to remote; MVO + ve infarct segments had significantly lower values than MVO-ve. (E2A remote= 57.4 ± 5.2°, E2A MVO-ve = 46.8 ± 2.5°, E2A MVO + ve = 36.8 ± 3.1°, ANOVA p < 0.001). RHM% (corresponding to subendocardium) was significantly lower in infarct segments compared to remote; MVO + ve infarct segments had significantly lower RHM% than MVO-ve. (RHM remote= 37 ± 3%, RHM RHM MVO-ve= 28 ± 7%, MVO + ve= 8 ± 5%, ANOVA p < 0.001). Conclusion The presence of MVO results in a decrease in MD and increase in FA in comparison to surrounding infarct segments. However, the reduction in E2A and right-handed myocytes on HA in infarct segments is further exacerbated by the presence of MVO. Further study is required to investigate the underlying mechanisms for such alterations in signal intensity. Abstract Figure. A case of transmural septal MI with MVO

scholarly journals Extended perfusion protocol for MS lesion quantification

Open Medicine ◽  
2020 ◽  
Vol 15 (1) ◽  
pp. 520-530
Author(s):  
Eleftherios Kontopodis ◽  
Kostas Marias ◽  
Georgios C. Manikis ◽  
Katerina Nikiforaki ◽  
Maria Venianaki ◽  
...  
Keyword(s):  
Statistical Analysis ◽  
Magnetic Resonance ◽  
Goodness Of Fit ◽  
Accurate Determination ◽  
Acquisition Time ◽  
Minimal Extension ◽  
Dce Mri ◽  
Contrast Enhanced ◽  
Demyelinating Lesions ◽  
Mri Protocol

AbstractThis study aims to examine a time-extended dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) protocol and report a comparative study with three different pharmacokinetic (PK) models, for accurate determination of subtle blood–brain barrier (BBB) disruption in patients with multiple sclerosis (MS). This time-extended DCE-MRI perfusion protocol, called Snaps, was applied on 24 active demyelinating lesions of 12 MS patients. Statistical analysis was performed for both protocols through three different PK models. The Snaps protocol achieved triple the window time of perfusion observation by extending the magnetic resonance acquisition time by less than 2 min on average for all patients. In addition, the statistical analysis in terms of adj-R2 goodness of fit demonstrated that the Snaps protocol outperformed the conventional DCE-MRI protocol by detecting 49% more pixels on average. The exclusive pixels identified from the Snaps protocol lie in the low ktrans range, potentially reflecting areas with subtle BBB disruption. Finally, the extended Tofts model was found to have the highest fitting accuracy for both analyzed protocols. The previously proposed time-extended DCE protocol, called Snaps, provides additional temporal perfusion information at the expense of a minimal extension of the conventional DCE acquisition time.

Diffusion Tensor Imaging and 3D Tractography of the Cervical Spinal Cord Using the ECG-Gated Line-Scan Technique

2007 ◽  
Vol 20 (5) ◽  
pp. 574-579 ◽  
Author(s):  
M. Hori ◽  
K. Ishigame ◽  
S. Aoki ◽  
H. Kumagai ◽  
T. Araki
Keyword(s):  
Spinal Cord ◽  
Clinical Symptoms ◽  
Diffusion Tensor ◽  
Cervical Spinal Cord ◽  
Three Dimensional ◽  
Acquisition Time ◽  
Clinical Use ◽  
Ecg Gating ◽  
Line Scan

Diffusion tensor (DT) magnetic resonance (MR) imaging in addition to conventional MR images provide valuable information on the brain. This study compared line scan DT imaging with and without the ECG-gating technique to estimate clinical usefulness of the line scan diffusion tensor image (LSDTI) with ECG-gating in evaluating spinal cord diseases in vivo. First, five healthy volunteers participated in the comparison study. LSDWI was performed in three to five sagittal sections with a pulsed-field-gradient diffusion preparation pulse employing two different b-values (0 and 700 s/mm2) along six directions. Apparent diffusion coefficient (ADC) maps and fractional anisotropy (FA) were calculated and three-dimensional tract reconstruction and color schemes of the spinal cord were obtained. Image quality and the acquisition time of each LSDTI were compared. Second, LSDTI with ECG-gating was performed in eighteen patients with cervical spinal cord disorders and evaluated by two neuroradiologists. Images with the ECG-gated technique were all superior to those without ECG—gating. Mean extended time for LSDTI with ECG-gating was approximately two minutes. In clinical use, the ADC and FA of spinal cord in patients with cervical spondylotic myelopathy statically changed. Moreover, demonstration of fibers was correlated with clinical symptoms. ECG-gating technique is preferable to LSDTI. The ADC and FA measurements and 3D fiber tracking of LSDTI with ECG-gating are promising methods to estimate cervical spinal cord pathology in clinical use.

scholarly journals Microstructural characteristics of chronic infarct segments assessed using diffusion tensor imaging

2021 ◽  
Vol 22 (Supplement_2) ◽  
Author(s):  
A Das ◽  
C Kelly ◽  
I Teh ◽  
C Stoeck ◽  
S Kozerke ◽  
...  
Keyword(s):  
Diffusion Tensor Imaging ◽  
Diffusion Tensor ◽  
Spin Echo ◽  
Mean Diffusivity ◽  
Free Breathing ◽  
Helix Angle ◽  
Acquisition Time ◽  
Transmural Infarct ◽  
Subendocardial Mi

Abstract Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): British Heart Foundation Background The microstructural changes following myocardial infarction (MI) can be characterised in-vivo with cardiac diffusion tensor imaging (cDTI) imaging, using mean diffusivity (MD), fractional anisotropy (FA), secondary eigenvector angle (E2A) and helix angle (HA) maps. In this study, we use cDTI to explore the microstructural differences between subendocardial and transmural chronic infarct segments. Method Twenty STEMI patients (15 men, 5 women, mean age 59) underwent 3T CMR scan at 3 months following presentation (mean interval 107 ± 18 days). Scan protocol included: second order motion compensated (M012) free-breathing spin echo DTI (3 slices, 18 diffusion directions at b-values 100s/mm2[3], 200s/mm2[3] and 500s/mm2[12], acquired resolution was 2.20x2.27x8mm3; cine gradient echo and LGE imaging. Average MD, FA, E2A and HA parameters were calculated on a 16-AHA-segmental level. HA maps were described by dividing values into left-handed HA (LHM, -90° < HA < -30°), circumferential HA (CM, -30° < HA < 30°), and right-handed HA (RHM, 30° < HA < 90°) and reported as relative proportions. Infarct segments were identified using LGE; patients were categorised according to the maximal transmurality of their infarct segments, into subendocardial (<50% LGE) or transmural (>50% LGE) MI. Results DTI acquisition was successful in all patients (acquisition time 13 ± 5mins). Ten patients had transmural MI. The results are shown in table 1. Transmurally infarcted segments had significantly lower FA (FA subendocardial MI = 0.27 ± 0.04, FA transmural MI = 0.23 ± 0.02, p < 0.01), lower E2A (E2A subendocardial MI = 47 ± 7°, E2A transmural MI = 38 ± 6°, p < 0.01) and lower proportions of right-handed cardiomyocytes (RHM subendocardial MI = 21 ± 5%, RHM transmural MI = 14 ± 5%, p < 0.01) than subendocardial infarct segments.  Conclusion Compared to subendocardial MI segments, the diffusion of water molecules is more isotropic in transmurally infarcted myocardium as evidenced by lower FA values, signifying increased structural disarray. The significantly lower E2A values suggest that laminar sheetlets of transmural infarct segments remain fixed at shallower angles during systole and are unable to reach their usual contractile configuration. The lower proportions of RHM on HA maps highlight the significantly greater loss of subendocardial cardiomyocytes in transmural infarct segments. Further studies are required to assess if these segmental changes can be predictive of long-term LV remodelling.

In vivo diffusion tensor imaging of the neonatal rat brain development

2013 ◽  
Vol 44 (S 01) ◽  
Author(s):  
M Breu ◽  
D Reisinger ◽  
D Wu ◽  
Y Zhang ◽  
A Fatemi ◽  
...  
Keyword(s):  
Diffusion Tensor Imaging ◽  
Brain Development ◽  
Rat Brain ◽  
Diffusion Tensor ◽  
Neonatal Rat ◽  
Neonatal Rat Brain

scholarly journals Optimization of spectral-domain optical coherence tomography with a supercontinuum source for in vivo motion detection of low reflective outer hair cells in guinea pig cochleae

2021 ◽  
Author(s):  
Fumiaki Nin ◽  
Samuel Choi ◽  
Takeru Ota ◽  
Zhang Qi ◽  
Hiroshi Hibino
Keyword(s):  
Optical Coherence Tomography ◽  
High Resolution ◽  
Guinea Pig ◽  
Hair Cells ◽  
Sensory Epithelium ◽  
Outer Hair Cells ◽  
Acquisition Time ◽  
Total Acquisition Time ◽  
Sd Oct

AbstractSound evokes sub-nanoscale vibration within the sensory epithelium. The epithelium contains not only immotile cells but also contractile outer hair cells (OHCs) that actively shrink and elongate synchronously with the sound. However, the in vivo motion of OHCs has remained undetermined. The aim of this work is to perform high-resolution and -accuracy vibrometry in live guinea pigs with an SC-introduced spectral-domain optical coherence tomography system (SD-OCT). In this study, to reveal the effective contribution of SC source in the recording of the low reflective materials with the short total acquisition time, we compare the performances of the SC-introduced SD-OCT (SCSD-OCT) to that of the conventional SD-OCT. As inanimate comparison objects, we record a mirror, a piezo actuator, and glass windows. For the measurements in biological materials, we use in/ex vivo guinea pig cochleae. Our study achieved the optimization of a SD-OCT system for high-resolution in vivo vibrometry in the cochlear sensory epithelium, termed the organ of Corti, in mammalian cochlea. By introducing a supercontinuum (SC) light source and reducing the total acquisition time, we improve the axial resolution and overcome the difficulty in recording the low reflective material in the presence of biological noise. The high power of the SC source enables the system to achieve a spatial resolution of 1.72 ± 0.00 μm on a mirror and reducing the total acquisition time contributes to the high spatial accuracy of sub-nanoscale vibrometry. Our findings reveal the vibrations at the apical/basal region of OHCs and the extracellular matrix, basilar membrane.

scholarly journals Meyer’s Loop Anatomy Demonstrated Using Diffusion Tensor MR Imaging and Fiber Tractography at 3T

2014 ◽  
Vol 60 (5) ◽  
pp. 215-222 ◽  
Author(s):  
Cristina Goga ◽  
Zeynep Firat ◽  
Klara Brinzaniuc ◽  
Is Florian
Keyword(s):  
Diffusion Tensor Imaging ◽  
Diffusion Tensor ◽  
Region Of Interest ◽  
Fiber Tractography ◽  
Brain Images ◽  
3 Dimensional ◽  
Software Release ◽  
Magnetic Resonance Imaging Mri ◽  
Meyer’S Loop

Abstract Objective: The ultimate anatomy of the Meyer’s loop continues to elude us. Diffusion tensor imaging (DTI) and diffusion tensor tractography (DTT) may be able to demonstrate, in vivo, the anatomy of the complex network of white matter fibers surrounding the Meyer’s loop and the optic radiations. This study aims at exploring the anatomy of the Meyer’s loop by using DTI and fiber tractography. Methods: Ten healthy subjects underwent magnetic resonance imaging (MRI) with DTI at 3 T. Using a region-of-interest (ROI) based diffusion tensor imaging and fiber tracking software (Release 2.6, Achieva, Philips), sequential ROI were placed to reconstruct visual fibers and neighboring projection fibers involved in the formation of Meyer’s loop. The 3-dimensional (3D) reconstructed fibers were visualized by superimposition on 3-planar MRI brain images to enhance their precise anatomical localization and relationship with other anatomical structures. Results: Several projection fiber including the optic radiation, occipitopontine/parietopontine fibers and posterior thalamic peduncle participated in the formation of Meyer’s loop. Two patterns of angulation of the Meyer’s loop were found. Conclusions: DTI with DTT provides a complimentary, in vivo, method to study the details of the anatomy of the Meyer’s loop.

Sign in / Sign up

Export Citation Format

Share Document