Segmentation of carotid arterial walls using neural networks.

BACKGROUND: Automated, accurate, objective, and quantitative medical image segmentation has remained a challenging goal in computer science since its inception. This study applies the technique of convolutional neural networks (CNNs) to the task of segmenting carotid arteries to aid in the assessment of pathology. AIM: To investigate CNN's utility as an ancillary tool for researchers who require accurate segmentation of carotid vessels. METHODS: An expert reader delineated vessel wall boundaries on 4422 axial T2-weighted magnetic resonance images of bilateral carotid arteries from 189 subjects with clinically evident atherosclerotic disease. A portion of this dataset was used to train two CNNs (one to segment the vessel lumen and the other to segment the vessel wall) with the remaining portion used to test the algorithm's efficacy by comparing CNN segmented images with those of an expert reader. RESULTS: Overall quantitative assessment between automated and manual segmentations was determined by computing the DICE coefficient for each pair of segmented images in the test dataset for each CNN applied. The average DICE coefficient for the test dataset (CNN segmentations compared to expert's segmentations) was 0.96 for the lumen and 0.87 for the vessel wall. Pearson correlation values and the intra-class correlation coefficient (ICC) were computed for the lumen (Pearson = 0.98, ICC = 0.98) and vessel wall (Pearson = 0.88, ICC = 0.86) segmentations. Bland-Altman plots of area measurements for the CNN and expert readers indicate good agreement with a mean bias of 1%-8%. CONCLUSION: Although the technique produces reasonable results that are on par with expert human assessments, our application requires human supervision and monitoring to ensure consistent results. We intend to deploy this algorithm as part of a software platform to lessen researchers' workload to more quickly obtain reliable results.

Predictors of change in carotid atherosclerotic plaque inflammation and burden as measured by 18-FDG-PET and MRI, respectively, in the dal-PLAQUE study.

Baseline predictors of response to treatment of patients with coronary heart disease (CHD) with respect to vascular inflammation and atherosclerotic plaque burden are poorly understood. From post hoc analysis of the dal-PLAQUE study (NCT00655473), 18F-fluorodeoxyglucose-positron emission tomography (18-FDG-PET) imaging and carotid black blood magnetic resonance imaging (MRI) were used to track changes in these vascular parameters. Baseline demographics, imaging, and biomarkers were collected/measured in 130 patients with CHD or CHD risk-equivalents, and imaging follow-up at 6 months (PET) and 24 months (MRI) was performed. Using stepwise linear regression, predictors of change in carotid plaque inflammation by PET [target-to-background ratio (TBR), n = 92] and plaque burden by MRI [wall area (WA) and total vessel area (TVA), n = 89] were determined. Variables with p < 0.05 in multivariable models were considered independently significant. Interleukin-6, systolic blood pressure and standard deviation of wall thickness (WT) at baseline were independently positively associated with 18-FDG uptake (mean of maximum [MeanMax] TBR change over 6 months). Mean of mean TBR, phospholipase A2, apolipoprotein A-I, and high-sensitivity C-reactive protein at baseline were independently negatively associated with MeanMax TBR change over 6 months. Mean WT and plasminogen activator inhibitor-1 (PAI-1) activity at baseline, and age, were independently associated with change in WA over 24 months. For TVA changes; mean WA and PAI-1 activity at baseline, age, and female gender were independent predictors. These findings may help determine patients most suitable for clinical trials employing plaque inflammation or burden changes as endpoints.

Iron oxide core oil-in-water emulsions as a multifunctional nanoparticle platform for tumor targeting and imaging.

Nanoemulsions are increasingly investigated for the delivery of hydrophobic drugs to improve their bioavailability or make their administration possible. In the current study, oil-in-water emulsions with three different mean diameters (30, 60, and 95 nm) were developed as a new multimodality nanoparticle platform for tumor targeting and imaging. To that aim, hydrophobically coated iron oxide particles were included in the soybean oil core of the nanoemulsions to enable their detection with magnetic resonance imaging (MRI), while the conjugation of a near infrared fluorophore allowed optical imaging. The accumulation of this novel nanocomposite in subcutaneous human tumors in nude mice was demonstrated with MRI and fluorescence imaging in vivo, and with Perl's staining of histological tumor sections ex vivo.

Macrophage-specific lipid-based nanoparticles improve cardiac magnetic resonance detection and characterization of human atherosclerosis.

OBJECTIVES: We sought to determine whether gadolinium (Gd)-containing lipid-based nanoparticles (NPs) targeting the macrophage scavenger receptor-B (CD36) improve cardiac magnetic resonance (CMR) detection and characterization of human atherosclerosis. BACKGROUND: Gd-containing lipid-based NPs targeting macrophages have improved MR detection of murine atherosclerosis. METHODS: Gadolinium-containing untargeted NPs, anti-CD36 NPs, and nonspecific Fc-NPs were created. Macrophages were incubated with fluorescent targeted and nontargeted NPs to determine uptake via confocal microscopy and inductively coupled plasma mass spectroscopy (ICP-MS) quantified Gd uptake. Human aortic specimens were harvested at autopsy. With a 1.5-T scanner, T1, T2, and PDW 3-dimensional scans were performed along with post-contrast scans after 24 h incubation. The T1 and cluster analyses were performed and compared with immunohistopathology. RESULTS: The NPs had a mean diameter of 125 nm and 14,900 Gd-ions, and relaxivity was 37 mmol/l(-1)s(-1) at 1.5-T and 37 degrees C. Confocal microscopy and ICP-MS demonstrated significant in vitro macrophage uptake of targeted NPs, whereas non-targeted NPs had minimal uptake. On T1 imaging, targeted NPs increased contrast-to-noise ratio (CNR) by 52.5%, which was significantly greater than Fc-NPs (CNR increased 17.2%) and nontargeted NPs (CNR increased 18.7%) (p = 0.001). Confocal fluorescent microscopy showed that NPs target resident macrophages, whereas the untargeted NPs and Fc-NPs are found diffusely throughout the plaque. Targeted NPs had a greater signal intensity increase in the fibrous cap compared with non-targeted NPs. CONCLUSIONS: Macrophage-specific (CD36) NPs bind human macrophages and improve CMR detection and characterization of human aortic atherosclerosis. Thus, macrophage-specific NPs could help identify high-risk human plaque before the development of an atherothrombotic event.

Cross-sectional, prospective study of MRI reproducibility in the assessment of plaque burden of the carotid arteries and aorta.

BACKGROUND: The reliability of imaging techniques to assess early atherosclerosis remains unclear. We did a cross-sectional, prospective study to test reproducibility of MRI when imaging arteries, to assess risk of cardiovascular disease and correlations with age and sex. METHODS: Between January 2003 and December 2006 we performed black-blood MRI of both common carotid arteries and the thoracic descending aorta in patients with cardiovascular risk factors who were referred from clinics in New York, NY, USA. Mean wall area, wall thickness, lumen area, total vessel area, and ratio of the mean wall area to the mean total vessel area (WA/TVA) were manually measured. Reproducibility within and between readers was tested on subsets of images from randomly chosen patients. RESULTS: MRI was performed on 300 patients. Intrareader reproducibility, assessed in images from 20 patients, was high for all parameters (intraclass correlation coefficients >0.8), except WA/TVA ratio in the descending aorta. The inter-reader reproducibility, assessed in images from 187 patients, was acceptable (intraclass correlation coefficients >0.7) for the mean wall, lumen, and total vessel areas. Values for all MRI parameters in all vessels increased with increasing age for both sexes (all P <0.0005) but were always significantly higher in men than in women, except for aortic mean wall thickness and WA/TVA ratio in the carotid arteries. Mean wall area values correlated well between the carotid arteries and aorta, reflecting the systemic nature of atherosclerosis. CONCLUSIONS: Our findings support MRI as a reproducible measurement of plaque burden and demonstrate the systemic distribution of atherosclerosis.

"Stroop concordant" coloring of letters for remediation of dyslexia.

In the modern society of the printed word dyslexia can be distressing and disabling. Commonly dyslexia manifests as a difficulty reading often caused by confusion or reversal of certain letters such as 'b' and 'd', and 'p' and 'q' and 'g'. Here we suggest that one method of remediation or amelioration is to print letters in color such as a 'b' in blue print, an 'r' in red, 'g' in green and 'p' in pink. Such coloring of letters takes advantage of the well-known association of vision of the color with the color's name--and therefore enunciation of the color's first letter.

Automated classification of atherosclerotic plaque from magnetic resonance images using predictive models.

The information contained within multicontrast magnetic resonance images (MRI) promises to improve tissue classification accuracy, once appropriately analyzed. Predictive models capture relationships empirically, from known outcomes thereby combining pattern classification with experience. In this study, we examine the applicability of predictive modeling for atherosclerotic plaque component classification of multicontrast ex vivo MR images using stained, histopathological sections as ground truth. Ten multicontrast images from seven human coronary artery specimens were obtained on a 9.4 T imaging system using multicontrast-weighted fast spin-echo (T1-, proton density-, and T2-weighted) imaging with 39-mum isotropic voxel size. Following initial data transformations, predictive modeling focused on automating the identification of specimen's plaque, lipid, and media. The outputs of these three models were used to calculate statistics such as total plaque burden and the ratio of hard plaque (fibrous tissue) to lipid. Both logistic regression and an artificial neural network model (Relevant Input Processor Network-RIPNet) were used for predictive modeling. When compared against segmentation resulting from cluster analysis, the RIPNet models performed between 25 and 30% better in absolute terms. This translates to a 50% higher true positive rate over given levels of false positives. This work indicates that it is feasible to build an automated system of plaque detection using MRI and data mining.

Carotid black blood MRI burden of atherosclerotic disease assessment correlates with ultrasound intima-media thickness.

The aim of this study was to correlate carotid black blood MRI based measurements with those obtained by ultrasound intima-media thickness (IMT). Seventeen patients with intermediate to high Framingham cardiovascular risk score underwent both carotid ultrasound and rapid extended coverage double inversion recovery black blood carotid MRI. Overall, there was good correlation between wall area, wall thickness, and plaque index measured by MRI and the IMT measurements obtained from the ultrasound images (max r2 = 0.72, p < 0.05). Patients with mean IMT > or = 1.2 mm had significantly higher values of wall area, plaque index and wall thickness compared to patients with mean IMT < 1.2 mm. Vessel wall measurements assessed by black-blood MRI may be potentially used clinically to evaluate plaque progression and regression.

Gradient echo acquisition for superparamagnetic particles with positive contrast (GRASP): sequence characterization in membrane and glass superparamagnetic iron oxide phantoms at 1.5T and 3T.

Iron oxides are used for cell trafficking and identification of macrophages in plaque using MRI. Due to the negative contrast, differentiation between signal loss caused by iron and native low signal in tissue may be problematic. It is, therefore, preferable to achieve positive contrast. The purpose of this study was to test the efficacy of a new MRI sequence GRASP (GRe Acquisition for Superparamagnetic Particles) to generate a positive signal in phantoms containing iron. Membrane phantoms were constructed containing Ferumoxide at 7 concentrations. Standard GRE sequences were modified with user controlled z-gradient rephasing (+/- 100%). CNR values were determined as a function of echo time (TE) and % rephasing at 1.5T and 3T. T(2)* values were determined using multiple double-echo GRE. The GRASP sequence generated positive signal enhancement in phantoms containing iron. For all rephasing values

Comparison of gated and non-gated fast multislice black-blood carotid imaging using rapid extended coverage and inflow/outflow saturation techniques.

PURPOSE: To comparatively analyze two fast in vivo multislice black-blood carotid artery vessel wall imaging techniques with and without cardiac gating. MATERIALS AND METHODS: Eight subjects with carotid artery atherosclerosis, and four healthy subjects were studied using two black-blood multislice techniques: rapid extended coverage double inversion recovery (REX-DIR), and inflow/outflow saturation band (IOSB) rapid acquisition with relaxation enhancement (RARE) multislice acquisitions. Quantitative, qualitative, and morphometric analyses were performed on images. RESULTS: Gating produced significantly lower values for the REX-DIR sequence with respect to signal intensity in muscle and the carotid artery wall, whereas it had no effect on flow suppression compared to non-gated images. For the IOSB sequences, gating had no significant effect on signal intensity of muscle and the carotid artery wall, but worsened flow suppression. REX-DIR and IOSB sequences were statistically different with respect to signal intensity of muscle (with REX-DIR sequences having lower values), while no statistical significance was observed for flow suppression and wall delineation. A morphologic analysis of the vessel wall and lumen comparing REX-DIR gated, IOSB gated, REX-DIR non-gated, and IOSB non-gated sequences revealed no significant differences between the acquisition techniques tested. CONCLUSION: Non-gated sequences may be used instead of gated sequences in atherosclerotic vessel wall imaging without compromising image quality. This may shorten examination time and improve patient comfort.

Quantification of human atherosclerotic plaques using spatially enhanced cluster analysis of multicontrast-weighted magnetic resonance images.

One of the current limitations of magnetic resonance imaging (MRI) is the lack of an objective method to classify plaque components. Here we present a cluster analysis technique that can objectively quantify and classify MR images of atherosclerotic plaques. We obtained three-dimensional (3D) images from 12 human coronary artery specimens on a 9.4T imaging system using multicontrast-weighted fast spin-echo (T1-, proton density-, and T2-weighted) imaging with an isotropic voxel size of 39 micro. Spatially enhanced cluster analysis (SECA) was performed on multicontrast MR images, and the resulting segmentation was evaluated against histological tracings. To visualize the overall structure of plaques, the MR images were rendered in 3D. The specimens exhibited lesions of American Heart Association (AHA) plaque classification types I-VI. Both MR images and histological sections were independently reviewed, categorized, and compared. Overall, the classification obtained from the cluster-analyzed MR and histopathology images showed very good agreement for all AHA types (92%, Cohen's kappa = 0.89, P < 0.0001). All plaque types were identified and quantified by SECA with a high degree of correlation between cluster-analyzed MR and manually traced histopathology data. MRI combined with SECA provides an objective method for atherosclerotic plaque component characterization and quantification.

Rapid extended coverage simultaneous multisection black-blood vessel wall MR imaging.

A two-dimensional rapid extended coverage (REX) rapid acquisition with relaxation enhancement (RARE) pulse sequence for simultaneous multisection double inversion-recovery (DIR) black-blood vessel wall magnetic resonance (MR) imaging was developed. Aortic vessel wall MR imaging was performed in five healthy subjects (mean age, 33 years +/- 4 [SD]) and five patients with atherosclerotic disease (mean age, 67 years +/- 11.7). Shortening of blood inversion time and imaging of multiple sections after single DIR block resulted in simultaneous acquisition of up to 20 aortic wall sections in less than 1 minute (spatial resolution, 0.97 x 0.97 x 3 mm(3)). Higher signal-to-noise ratios per unit time per section (16.0 +/- 2.45 vs 7.5 +/- 1.10, P <.05), no significant changes in contrast-to-noise ratios (15.0 +/- 5.3 vs 20.1 +/- 3.9, P >.05), and 17-fold improvement in acquisition time compared with those at conventional single-section DIR RARE imaging was achieved. Use of the REX method significantly shortened aortic imaging acquisition times without degrading image quality.

Parallel and nonparallel simultaneous multislice black-blood double inversion recovery techniques for vessel wall imaging.

PURPOSE: To reduce long examination times of black-blood vessel wall imaging by acquiring multiple slices simultaneously and by using parallel acquisition techniques. MATERIALS AND METHODS: DIR-rapid acquisition with relaxation enhancement (RARE) techniques imaging up to 10 simultaneous slices per acquisition with single and multiple 180 degrees -reinversion pulses were developed. A slab-selective reinversion multislice DIR-RARE sequence incorporating generalized autocalibrating partially parallel acquisitions (GRAPPA) imaging was implemented. Four-channel and eight-channel carotid coils were built to test these sequences. A total of 11 subjects were studied. Contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) efficiency factor (SEF, SNR/unit time/slice) were measured from aortic images of three healthy subjects to determine optimal MR parameters. The DIR-RARE-GRAPPA sequence was run on aortas and carotid arteries of the five remaining healthy subjects and three atherosclerotic patients with optimal parameters (acquisition times 12-21 seconds). RESULTS: SEFs of slab-selective protocols were significantly higher than those of slice-selective protocols, and SEFs of DIR-RARE-GRAPPA protocols were significantly higher than corresponding non-GRAPPA protocols (P < 0.05). CNR was not significantly different for all imaging protocols. The DIR-RARE-GRAPPA multislice sequence showed 8.35-fold time improvement vs. single-slice DIR-2RARE sequence. CONCLUSION: Future MRI atherosclerotic plaque studies can be performed in substantially shorter times using these methods.

Magnetic resonance microscopy quantifies the disease progression in Marfan syndrome mice.

PURPOSE: To use noninvasive magnetic resonance microscopy (MRM) to examine aneurysmal disease in the mouse model of Marfan syndrome (MFS). MATERIALS AND METHODS: A total of eight wild-type (WT) and MFS mice were imaged using MRM; four of them at three different time points over an 8-week period and the remaining animals were imaged at one time point. The maximal cross-sectional area of the aorta was measured by manual tracing and by automated means from combined cardiac and respiratory-gated bright-blood images. Relationships between aortic size and age and the differences between WT and mutant mice aortic size were established. RESULTS: Maximal cross-sectional aortic areas differed significantly (P < 0.05) between WT and mutant mice for all time points, with MFS mice having larger aortic size. There was a positive correlation between aortic size and age in MFS mice (r = 0.80) with a significant increase from the 14th to the 22nd week (P < 0.05). CONCLUSION: MRM detected the differences in aortic size between WT and mutant mice over time, demonstrating a potential for noninvasive technique for the assessment of potential therapies in MFS mice.