Purpose To examine the role of sex in abnormal white matter microstructure after soccer heading as identified by using the diffusion-tensor imaging (DTI) metric fractional anisotropy (FA). Materials and Methods In this prospective cross-sectional study, 98 individuals who were enrolled in a larger prospective study of amateur soccer players (from 2013 to 2016) were matched 1:1 for age and history of soccer heading in the prior 12 months. Among the subjects, 49 men (mean age, 25.7 years; range, 18-50 years) and 49 women (mean age, 25.8 years; range, 18-50 years) with median total soccer headings per year of 487 and 469, respectively, underwent 3.0-T DTI. Images were registered to the Johns Hopkins University template. A voxelwise linear regression was fitted for FA with terms for the number of headings during the previous 12 months and its interaction with sex after controlling for the following potential confounders: age, years of education, number of lifetime concussions, and handedness. In the resulting statistical maps, P < .01 indicated a statistically significant difference, with a threshold cluster size larger than 100 mm3. Results Among men, three regions were identified in which greater heading exposure was associated with lower FA; eight such regions were identified among women (>100 contiguous voxels, P < .01). In seven of the eight regions identified in women, the association between heading and FA was stronger in women than in men. There was no significant difference of heading with FA between the sexes for any region in which heading was associated with FA among men (P > .01, <100 contiguous voxels). Conclusion With similar exposure to heading, women exhibit more widespread evidence of microstructural white matter alteration than do men, suggesting preliminary support for a biologic divergence of brain response to repetitive trauma.
Purpose To develop a machine learning model that allows high-risk breast lesions (HRLs) diagnosed with image-guided needle biopsy that require surgical excision to be distinguished from HRLs that are at low risk for upgrade to cancer at surgery and thus could be surveilled. Materials and Methods Consecutive patients with biopsy-proven HRLs who underwent surgery or at least 2 years of imaging follow-up from June 2006 to April 2015 were identified. A random forest machine learning model was developed to identify HRLs at low risk for upgrade to cancer. Traditional features such as age and HRL histologic results were used in the model, as were text features from the biopsy pathologic report. Results One thousand six HRLs were identified, with a cancer upgrade rate of 11.4% (115 of 1006). A machine learning random forest model was developed with 671 HRLs and tested with an independent set of 335 HRLs. Among the most important traditional features were age and HRL histologic results (eg, atypical ductal hyperplasia). An important text feature from the pathologic reports was “severely atypical.” Instead of surgical excision of all HRLs, if those categorized with the model to be at low risk for upgrade were surveilled and the remainder were excised, then 97.4% (37 of 38) of malignancies would have been diagnosed at surgery, and 30.6% (91 of 297) of surgeries of benign lesions could have been avoided. Conclusion This study provides proof of concept that a machine learning model can be applied to predict the risk of upgrade of HRLs to cancer. Use of this model could decrease unnecessary surgery by nearly one-third and could help guide clinical decision making with regard to surveillance versus surgical excision of HRLs. (©) RSNA, 2017.
Purpose To determine if patient survival and mechanisms of right ventricular failure in pulmonary hypertension could be predicted by using supervised machine learning of three-dimensional patterns of systolic cardiac motion. Materials and Methods The study was approved by a research ethics committee, and participants gave written informed consent. Two hundred fifty-six patients (143 women; mean age ± standard deviation, 63 years ± 17) with newly diagnosed pulmonary hypertension underwent cardiac magnetic resonance (MR) imaging, right-sided heart catheterization, and 6-minute walk testing with a median follow-up of 4.0 years. Semiautomated segmentation of short-axis cine images was used to create a three-dimensional model of right ventricular motion. Supervised principal components analysis was used to identify patterns of systolic motion that were most strongly predictive of survival. Survival prediction was assessed by using difference in median survival time and area under the curve with time-dependent receiver operating characteristic analysis for 1-year survival. Results At the end of follow-up, 36% of patients (93 of 256) died, and one underwent lung transplantation. Poor outcome was predicted by a loss of effective contraction in the septum and free wall, coupled with reduced basal longitudinal motion. When added to conventional imaging and hemodynamic, functional, and clinical markers, three-dimensional cardiac motion improved survival prediction (area under the receiver operating characteristic curve, 0.73 vs 0.60, respectively; P < .001) and provided greater differentiation according to difference in median survival time between high- and low-risk groups (13.8 vs 10.7 years, respectively; P < .001). Conclusion A machine-learning survival model that uses three-dimensional cardiac motion predicts outcome independent of conventional risk factors in patients with newly diagnosed pulmonary hypertension. Online supplemental material is available for this article.
Purpose To document the imaging findings associated with congenital Zika virus infection as found in the Instituto de Pesquisa in Campina Grande State Paraiba (IPESQ) in northeastern Brazil, where the congenital infection has been particularly severe. Materials and Methods From June 2015 to May 2016, 438 patients were referred to the IPESQ for rash occurring during pregnancy or for suspected fetal central nervous system abnormality. Patients who underwent imaging at IPESQ were included, as well as those with documented Zika virus infection in fluid or tissue (n = 17, confirmed infection cohort) or those with brain findings suspicious for Zika virus infection, with intracranial calcifications (n = 28, presumed infection cohort). Imaging examinations included 12 fetal magnetic resonance (MR) examinations, 42 postnatal brain computed tomographic examinations, and 11 postnatal brain MR examinations. Images were reviewed by four radiologists, with final opinion achieved by means of consensus. Results Brain abnormalities seen in confirmed (n = 17) and presumed (n = 28) congenital Zika virus infections were similar, with ventriculomegaly in 16 of 17 (94%) and 27 of 28 (96%) infections, respectively; abnormalities of the corpus callosum in 16 of 17 (94%) and 22 of 28 (78%) infections, respectively; and cortical migrational abnormalities in 16 of 17 (94%) and 28 of 28 (100%) infections, respectively. Although most fetuses underwent at least one examination that showed head circumference below the 5th percentile, head circumference could be normal in the presence of severe ventriculomegaly (seen in three fetuses). Intracranial calcifications were most commonly seen at the gray matter-white matter junction, in 15 of 17 (88%) and 28 of 28 (100%) confirmed and presumed infections, respectively. The basal ganglia and/or thalamus were also commonly involved with calcifications in 11 of 17 (65%) and 18 of 28 (64%) infections, respectively. The skull frequently had a collapsed appearance with overlapping sutures and redundant skin folds and, occasionally, intracranial herniation of orbital fat and clot in the confluence of sinuses. Conclusion The spectrum of findings associated with congenital Zika virus infection in the IPESQ in northeastern Brazil is illustrated to aid the radiologist in identifying Zika virus infection at imaging. (©) RSNA, 2016 Online supplemental material is available for this article.
Imaging research laboratories are rapidly creating machine learning systems that achieve expert human performance using open-source methods and tools. These artificial intelligence systems are being developed to improve medical image reconstruction, noise reduction, quality assurance, triage, segmentation, computer-aided detection, computer-aided classification, and radiogenomics. In August 2018, a meeting was held in Bethesda, Maryland, at the National Institutes of Health to discuss the current state of the art and knowledge gaps and to develop a roadmap for future research initiatives. Key research priorities include: 1, new image reconstruction methods that efficiently produce images suitable for human interpretation from source data; 2, automated image labeling and annotation methods, including information extraction from the imaging report, electronic phenotyping, and prospective structured image reporting; 3, new machine learning methods for clinical imaging data, such as tailored, pretrained model architectures, and federated machine learning methods; 4, machine learning methods that can explain the advice they provide to human users (so-called explainable artificial intelligence); and 5, validated methods for image de-identification and data sharing to facilitate wide availability of clinical imaging data sets. This research roadmap is intended to identify and prioritize these needs for academic research laboratories, funding agencies, professional societies, and industry.
Purpose To examine the effects of subconcussive impacts resulting from a single season of youth (age range, 8-13 years) football on changes in specific white matter (WM) tracts as detected with diffusion-tensor imaging in the absence of clinically diagnosed concussions. Materials and Methods Head impact data were recorded by using the Head Impact Telemetry system and quantified as the combined-probability risk-weighted cumulative exposure (RWECP). Twenty-five male participants were evaluated for seasonal fractional anisotropy (FA) changes in specific WM tracts: the inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus, and superior longitudinal fasciculus (SLF). Fiber tracts were segmented into a central core and two fiber terminals. The relationship between seasonal FA change in the whole fiber, central core, and the fiber terminals with RWECP was also investigated. Linear regression analysis was conducted to determine the association between RWECP and change in fiber tract FA during the season. Results There were statistically significant linear relationships between RWEcp and decreased FA in the whole (R(2) = 0.433; P = .003), core (R(2) = 0.3649; P = .007), and terminals (R(2) = 0.5666; P < .001) of left IFOF. A trend toward statistical significance (P = .08) in right SLF was observed. A statistically significant correlation between decrease in FA of the right SLF terminal and RWECP was also observed (R(2) = 0.2893; P = .028). Conclusion This study found a statistically significant relationship between head impact exposure and change of FA fractional anisotropy value of whole, core, and terminals of left IFOF and right SLF's terminals where WM and gray matter intersect, in the absence of a clinically diagnosed concussion. (©) RSNA, 2016.
Background Bariatric embolization is a new endovascular procedure to treat patients with obesity. However, the safety and efficacy of bariatric embolization are unknown. Purpose To evaluate the safety and efficacy of bariatric embolization in severely obese adults at up to 12 months after the procedure. Materials and Methods For this prospective study (NCT0216512 on ClinicalTrials.gov), 20 participants (16 women) aged 27-68 years (mean ± standard deviation, 44 years ± 11) with mean body mass index of 45 ± 4.1 were enrolled at two institutions from June 2014 to February 2018. Transarterial embolization of the gastric fundus was performed using 300- to 500-µm embolic microspheres. Primary end points were 30-day adverse events and weight loss at up to 12 months. Secondary end points at up to 12 months included technical feasibility, health-related quality of life (Short Form-36 Health Survey ([SF-36]), impact of weight on quality of life (IWQOL-Lite), and hunger or appetite using a visual assessment scale. Analysis of outcomes was performed by using one-sample t tests and other exploratory statistics. Results Bariatric embolization was performed successfully for all participants with no major adverse events. Eight participants had a total of 11 minor adverse events. Mean excess weight loss was 8.2% (95% confidence interval [CI]: 6.3%, 10%; P < .001) at 1 month, 11.5% (95% CI: 8.7%, 14%; P < .001) at 3 months, 12.8% (95% CI: 8.3%, 17%; P < .001) at 6 months, and 11.5% (95% CI: 6.8%, 16%; P < .001) at 12 months. From baseline to 12 months, mean SF-36 scores increased (mental component summary, from 46 ± 11 to 50 ± 10, P = .44; physical component summary, from 46 ± 8.0 to 50 ± 9.3, P = .15) and mean IWQOL-Lite scores increased from 57 ± 18 to 77 ± 18 (P < .001). Hunger or appetite decreased for 4 weeks after embolization and increased thereafter, without reaching pre-embolization levels. Conclusion Bariatric embolization is well tolerated in severely obese adults, inducing appetite suppression and weight loss for up to 12 months. Published under a CC BY-NC-ND 4.0 license. Online supplemental material is available for this article.
Although computer-aided diagnosis (CAD) is widely used in mammography, conventional CAD programs that use prompts to indicate potential cancers on the mammograms have not led to an improvement in diagnostic accuracy. Because of the advances in machine learning, especially with use of deep (multilayered) convolutional neural networks, artificial intelligence has undergone a transformation that has improved the quality of the predictions of the models. Recently, such deep learning algorithms have been applied to mammography and digital breast tomosynthesis (DBT). In this review, the authors explain how deep learning works in the context of mammography and DBT and define the important technical challenges. Subsequently, they discuss the current status and future perspectives of artificial intelligence-based clinical applications for mammography, DBT, and radiomics. Available algorithms are advanced and approach the performance of radiologists-especially for cancer detection and risk prediction at mammography. However, clinical validation is largely lacking, and it is not clear how the power of deep learning should be used to optimize practice. Further development of deep learning models is necessary for DBT, and this requires collection of larger databases. It is expected that deep learning will eventually have an important role in DBT, including the generation of synthetic images.
Background Previous studies showed that nicotinized electronic cigarettes (hereafter, e-cigarettes) elicit systemic oxidative stress and inflammation. However, the effect of the aerosol alone on endothelial function is not fully understood. Purpose To quantify surrogate markers of endothelial function in nonsmokers after inhalation of aerosol from nicotine-free e-cigarettes. Materials and Methods In this prospective study (from May to September 2018), nonsmokers underwent 3.0-T MRI before and after inhaling nicotine-free e-cigarette aerosol. Peripheral vascular reactivity to cuff-induced ischemia was quantified by temporally resolving blood flow velocity and oxygenation (SvO2) in superficial femoral artery and vein, respectively, along with artery luminal flow-mediated dilation. Precuff occlusion, resistivity index, baseline blood flow velocity, and SvO2 were evaluated. During reactive hyperemia, blood flow velocity yielded peak velocity, time to peak, and acceleration rate (hyperemic index); SvO2 yielded washout time of oxygen-depleted blood, rate of resaturation, and maximum SvO2 increase (overshoot). Cerebrovascular reactivity was assessed in the superior sagittal sinus, evaluating the breath-hold index. Central arterial stiffness was measured via aortic pulse wave velocity. Differences before versus after e-cigarette vaping were tested with Hotelling T2 test. Results Thirty-one healthy never-smokers (mean age, 24.3 years ± 4.3; 14 women) were evaluated. After e-cigarette vaping, resistivity index was higher (0.03 of 1.30 [2.3%]; P < .05), luminal flow-mediated dilation severely blunted (-3.2% of 9.4% [-34%]; P < .001), along with reduced peak velocity (-9.9 of 56.6 cm/sec [-17.5%]; P < .001), hyperemic index (-3.9 of 15.1 cm/sec2 [-25.8%]; P < .001), and delayed time to peak (2.1 of 7.1 sec [29.6%]; P = .005); baseline SvO2 was lower (-13 of 65 %HbO2 [-20%]; P < .001) and overshoot higher (10 of 19 %HbO2 [52.6%]; P < .001); and aortic pulse wave velocity marginally increased (0.19 of 6.05 m/sec [3%]; P = .05). Remaining parameters did not change after aerosol inhalation. Conclusion Inhaling nicotine-free electronic cigarette aerosol transiently impacted endothelial function in healthy nonsmokers. Further studies are needed to address the potentially adverse long-term effects on vascular health. © RSNA, 2019 Online supplemental material is available for this article.
Purpose To evaluate the efficacy of deep convolutional neural networks (DCNNs) for detecting tuberculosis (TB) on chest radiographs. Materials and Methods Four deidentified HIPAA-compliant datasets were used in this study that were exempted from review by the institutional review board, which consisted of 1007 posteroanterior chest radiographs. The datasets were split into training (68.0%), validation (17.1%), and test (14.9%). Two different DCNNs, AlexNet and GoogLeNet, were used to classify the images as having manifestations of pulmonary TB or as healthy. Both untrained and pretrained networks on ImageNet were used, and augmentation with multiple preprocessing techniques. Ensembles were performed on the best-performing algorithms. For cases where the classifiers were in disagreement, an independent board-certified cardiothoracic radiologist blindly interpreted the images to evaluate a potential radiologist-augmented workflow. Receiver operating characteristic curves and areas under the curve (AUCs) were used to assess model performance by using the DeLong method for statistical comparison of receiver operating characteristic curves. Results The best-performing classifier had an AUC of 0.99, which was an ensemble of the AlexNet and GoogLeNet DCNNs. The AUCs of the pretrained models were greater than that of the untrained models (P < .001). Augmenting the dataset further increased accuracy (P values for AlexNet and GoogLeNet were .03 and .02, respectively). The DCNNs had disagreement in 13 of the 150 test cases, which were blindly reviewed by a cardiothoracic radiologist, who correctly interpreted all 13 cases (100%). This radiologist-augmented approach resulted in a sensitivity of 97.3% and specificity 100%. Conclusion Deep learning with DCNNs can accurately classify TB at chest radiography with an AUC of 0.99. A radiologist-augmented approach for cases where there was disagreement among the classifiers further improved accuracy. (©) RSNA, 2017.