Ryan Brunsing, MD

Abstract

To develop and evaluate the performance of a fully-automated convolutional neural network (CNN)-based algorithm to evaluate hepatobiliary phase (HBP) adequacy of gadoxetate disodium (EOB)-enhanced MRI. Secondarily, we explored the potential of the proposed CNN algorithm to reduce examination length by applying it to EOB-MRI examinations.We retrospectively identified EOB-enhanced MRI-HBP series from examinations performed 2011-2018 (internal and external datasets). Our algorithm, comprising a liver segmentation and classification CNN, produces an adequacy score. Two abdominal radiologists independently classified series as adequate or suboptimal. The consensus determination of HBP adequacy was used as ground truth for CNN model training and validation. Reader agreement was evaluated with Cohen's kappa. Performance of the algorithm was assessed by receiver operating characteristics (ROC) analysis and computation of the area under the ROC curve (AUC). Potential examination duration reduction was evaluated descriptively.1408 HBP series from 484 patients were included. Reader kappa agreement was 0.67 (internal dataset) and 0.80 (external dataset). AUCs were 0.97 (0.96-0.99) for internal and 0.95 (0.92-96) for external and were not significantly different from each other (p = 0.24). 48 % (50/105) examinations could have been shorter by applying the algorithm.A proposed CNN-based algorithm achieves higher than 95 % AUC for classifying HBP images as adequate versus suboptimal. The application of this algorithm could potentially shorten examination time and aid radiologists in recognizing technically suboptimal images, avoiding diagnostic pitfalls.

View details for DOI 10.1016/j.ejrad.2020.108837

View details for PubMedID 31958630

Abstract

Current self-calibration and reconstruction methods for wave-encoded single-shot fast spin echo imaging (SSFSE) requires long computational time, especially when high accuracy is needed.To develop and investigate the clinical feasibility of data-driven self-calibration and reconstruction of wave-encoded SSFSE imaging for computation time reduction and quality improvement.Prospective controlled clinical trial.With Institutional Review Board approval, the proposed method was assessed on 29 consecutive adult patients (18 males, 11 females, range, 24-77 years).A wave-encoded variable-density SSFSE sequence was developed for clinical 3.0T abdominal scans to enable 3.5 acceleration with full-Fourier acquisitions. Data-driven calibration of wave-encoding point-spread function (PSF) was developed using a trained deep neural network. Data-driven reconstruction was developed with another set of neural networks based on the calibrated wave-encoding PSF. Training of the calibration and reconstruction networks was performed on 15,783 2D wave-encoded SSFSE abdominal images.Image quality of the proposed data-driven approach was compared independently and blindly with a conventional approach using iterative self-calibration and reconstruction with parallel imaging and compressed sensing by three radiologists on a scale from -2 to 2 for noise, contrast, sharpness, artifacts, and confidence. Computation time of these two approaches was also compared.Wilcoxon signed-rank tests were used to compare image quality and two-tailed t-tests were used to compare computation time with P values of under 0.05 considered statistically significant.An average 2.1-fold speedup in computation was achieved using the proposed method. The proposed data-driven self-calibration and reconstruction approach significantly reduced the perceived noise level (mean scores 0.82, P<0.0001).The proposed data-driven calibration and reconstruction achieved twice faster computation with reduced perceived noise, providing a fast and robust self-calibration and reconstruction for clinical abdominal SSFSE imaging.1 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2019.

View details for DOI 10.1002/jmri.26871

View details for PubMedID 31322799

Abstract

Liver alignment between series/exams is challenged by dynamic morphology or variability in patient positioning or motion. Image registration can improve image interpretation and lesion co-localization. We assessed the performance of a convolutional neural network algorithm to register cross-sectional liver imaging series and compared its performance to manual image registration.Three hundred fourteen patients, including internal and external datasets, who underwent gadoxetate disodium-enhanced magnetic resonance imaging for clinical care from 2011 to 2018, were retrospectively selected. Automated registration was applied to all 2,663 within-patient series pairs derived from these datasets. Additionally, 100 within-patient series pairs from the internal dataset were independently manually registered by expert readers. Liver overlap, image correlation, and intra-observation distances for manual versus automated registrations were compared using paired t tests. Influence of patient demographics, imaging characteristics, and liver uptake function was evaluated using univariate and multivariate mixed models.Compared to the manual, automated registration produced significantly lower intra-observation distance (p < 0.001) and higher liver overlap and image correlation (p < 0.001). Intra-exam automated registration achieved 0.88 mean liver overlap and 0.44 mean image correlation for the internal dataset and 0.91 and 0.41, respectively, for the external dataset. For inter-exam registration, mean overlap was 0.81 and image correlation 0.41. Older age, female sex, greater inter-series time interval, differing uptake, and greater voxel size differences independently reduced automated registration performance (p 0.020).A fully automated algorithm accurately registered the liver within and between examinations, yielding better liver and focal observation co-localization compared to manual registration.

View details for DOI 10.1186/s41747-019-0120-7

View details for PubMedID 31655943

Abstract

To enable rapid imaging with a scan time-efficient 3D cones trajectory with a deep-learning off-resonance artifact correction technique.A residual convolutional neural network to correct off-resonance artifacts (Off-ResNet) was trained with a prospective study of pediatric MRA exams. Each exam acquired a short readout scan (1.18 ms 0.38) and a long readout scan (3.35 ms 0.74) at 3 T. Short readout scans, with longer scan times but negligible off-resonance blurring, were used as reference images and augmented with additional off-resonance for supervised training examples. Long readout scans, with greater off-resonance artifacts but shorter scan time, were corrected by autofocus and Off-ResNet and compared with short readout scans by normalized RMS error, structural similarity index, and peak SNR. Scans were also compared by scoring on 8 anatomical features by two radiologists, using analysis of variance with post hoc Tukey's test and two one-sided t-tests. Reader agreement was determined with intraclass correlation.The total scan time for long readout scans was on average 59.3% shorter than short readout scans. Images from Off-ResNet had superior normalized RMS error, structural similarity index, and peak SNR compared with uncorrected images across 1 kHz off-resonance (P < .01). The proposed method had superior normalized RMS error over -677 Hz to +1 kHz and superior structural similarity index and peak SNR over 1 kHz compared with autofocus (P < .01). Radiologic scoring demonstrated that long readout scans corrected with Off-ResNet were noninferior to short readout scans (P < .05).The proposed method can correct off-resonance artifacts from rapid long-readout 3D cones scans to a noninferior image quality compared with diagnostically standard short readout scans.

View details for DOI 10.1002/mrm.27825

View details for PubMedID 31115936

Abstract

Restriction spectrum imaging (RSI) is a novel diffusion-weighted MRI technique that uses the mathematically distinct behavior of water diffusion in separable microscopic tissue compartments to highlight key aspects of the tissue microarchitecture with high conspicuity. RSI can be acquired in less than 5min on modern scanners using a surface coil. Multiple field gradients and high b-values in combination with postprocessing techniques allow the simultaneous resolution of length-scale and geometric information, as well as compartmental and nuclear volume fraction filtering. RSI also uses a distortion correction technique and can thus be fused to high resolution T2-weighted images for detailed localization, which improves delineation of disease extension into critical anatomic structures. In this review, we discuss the acquisition, postprocessing, and interpretation of RSI for prostate MRI. We also summarize existing data demonstrating the applicability of RSI for prostate cancer detection, in vivo characterization, localization, and targeting.5 J. Magn. Reson. Imaging 2017;45:323-336.

View details for DOI 10.1002/jmri.25419

View details for Web of Science ID 000393654500001

View details for PubMedID 27527500

View details for PubMedCentralID PMC5222783

Abstract

Most fluorine-18 fluorodeoxyglucose (FDG)-positron emission tomography with computed tomography (PET/CT) studies are performed on cancer patients. These patients are at increased risk of pulmonary embolism (PE). In this retrospective review, we determined the rate of PE, and the prevalence of associated FDG-PET findings on intravenous (IV) contrast-enhanced PET/CT.We identified all PET/CT studies performed at our institution with a reported finding of PE between January 2005 and October 2012. The medical record was reviewed for symptoms, which were identified after the diagnosis of PE, and whether the patients received treatment. The prevalence of associated FDG-PET findings was determined.A total of 65 total cases of PE (of 182,72 total PET/CT examinations) were identified of which 59 were previously unknown. This gives an incidental PE (IPE) rate of 0.32%. Of the patients where sufficient clinical information was available, 34 of 36 (94%) were treated either with therapeutic anticoagulation or inferior vena cava filter, and 30 of 36 (83%) were asymptomatic in retrospect. Of the patients with IPE, we found nine (15.2%) with associated focal pulmonary artery hypermetabolism, three (5.1%) with hypermetabolic pulmonary infarction, and one with increased isolated right ventricular FDG uptake (1.7%). One case of chronic PE demonstrated a focal hypometabolic filling defect in a pulmonary artery on PET.We found IPE in 0.32% of PET/CT scans. Focal pulmonary artery hypermetabolism or hypometabolism, and hypermetabolic pulmonary artery infarction with the "rim sign" were uncommonly associated with PE. These findings could raise the possibility of IPE in non-IV contrast-enhanced PET/CT studies.

View details for DOI 10.1016/j.acra.2014.02.013

View details for Web of Science ID 000336414100004

View details for PubMedID 24809314

View details for Web of Science ID 000361438102290

Abstract

The transcription factor Foxp3 is critical to the suppressive phenotype of CD4+ regulatory T cells. Studies have clearly shown that numerous autoimmune diseases are marked by the presence of activated CD4+ T cells within the setting of chronic inflammation. Therefore, drugs capable of inducing Foxp3 expression in activated CD4+ T cells could be of great therapeutic interest. We have previously shown that the small molecule G-1, an agonist directed against the membrane-bound G protein-coupled estrogen receptor, can induce IL10 expression in naive CD4+ T cells. In addition, we and others have demonstrated that G-1 attenuates disease in an animal model of experimental autoimmune encephalomyelitis. Using ex vivo cultures of purified CD4+ T cells, we show that G-1 can elicit Foxp3 expression under TH17-polarizing conditions, which mimic the in situ inflammatory milieu of several autoimmune diseases. These findings build upon previous results demonstrating the immunosuppressive properties of the novel estrogenic small molecule G-1.

View details for DOI 10.1097/CJI.0b013e31828d8e3b

View details for Web of Science ID 000316528400004

View details for PubMedID 23502766

View details for PubMedCentralID PMC3635139

Abstract

Interleukin-10 (IL-10) is a potent suppressor of the immune system, commonly produced by CD4(+) T cells to limit ongoing inflammatory responses minimizing host damage. Many autoimmune diseases are marked by large populations of activated CD4(+) T cells within the setting of chronic inflammation; therefore, drugs capable of inducing IL-10 production in CD4(+) T cells would be of great therapeutic value. Previous reports have shown that the small molecule G-1, an agonist of the membrane-bound G-protein-coupled estrogen receptor GPER, attenuates disease in an animal model of autoimmune encephalomyelitis. However, the direct effects of G-1 on CD4(+) T-cell populations remain unknown. Using ex vivo cultures of purified CD4(+) T cells, we show that G-1 elicits IL-10 expression in T helper type 17 (Th17) -polarized cells, increasing the number of IL-10(+) and IL-10(+) IL-17A(+) cells via de novo induction of IL-10. T-cell cultures differentiated in the presence of G-1 secreted threefold more IL-10, with no change in IL-17A, tumour necrosis factor-, or interferon-. Moreover, inhibition of extracellular signal-regulated kinase (but not p38 or Jun N-terminal kinase) signalling blocked the response, while analysis of Foxp3 and RORt expression demonstrated increased numbers of IL-10(+) cells in both the Th17 (RORt(+)) and Foxp3(+) RORt(+) hybrid T-cell compartments. Our findings translated in vivo as systemic treatment of male mice with G-1 led to increased IL-10 secretion from splenocytes following T-cell receptor cross-linking. These results demonstrate that G-1 acts directly on CD4(+) T cells, and to our knowledge provide the first example of a synthetic small molecule capable of eliciting IL-10 expression in Th17 or hybrid T-cell populations.

View details for DOI 10.1111/j.1365-2567.2011.03471.x

View details for Web of Science ID 000293696200010

View details for PubMedID 21722102

View details for PubMedCentralID PMC3173698

View details for Web of Science ID 000209758304100

Abstract

The unfolded protein response (UPR) signaling pathway regulates the functional capacity of the endoplasmic reticulum for protein folding. Beyond a role for UPR signaling during terminal differentiation of mature B cells to antibody-secreting plasma cells, the status or importance of UPR signaling during hematopoiesis has not been explored, due in part to difficulties in isolating sufficient quantities of cells at developmentally intermediate stages required for biochemical analysis. Following reconstitution of irradiated mice with hematopoietic cells carrying a fluorescent UPR reporter construct, we found that IRE1 nuclease activity for XBP1 splicing is active at early stages of T- and B-lymphocyte differentiation: in bone marrow pro-B cells and in CD4(+)CD8(+) double positive thymic T cells. IRE1 was not active in B cells at later stages. In T cells, IRE activity was not detected in the more mature CD4(+) T-cell population but was active in the CD8(+) cytotoxic T-cell population. Multiple signals are likely to be involved in activating IRE1 during lymphocyte differentiation, including rearrangement of antigen receptor genes. Our results show that reporter-transduced hematopoietic stem cells provide a quick and easy means to identify UPR signaling component activation in physiological settings.

View details for DOI 10.1074/jbc.M801395200

View details for Web of Science ID 000256949200024

View details for PubMedID 18375386