The University of California San Francisco Brain Metastases Stereotactic Radiosurgery (UCSF-BMSR) MRI Dataset.

Supplemental material is available for this article.

Altered Iron and Microstructure in Huntington's Disease Subcortical Nuclei: Insight From 7T MRI.

BACKGROUND: Pathophysiological changes of Huntington's disease (HD) can precede symptom onset by decades. Robust imaging biomarkers are needed to monitor HD progression, especially before the clinical onset. PURPOSE: To investigate iron dysregulation and microstructure alterations in subcortical regions as HD imaging biomarkers, and to associate such alterations with motor and cognitive impairments. STUDY TYPE: Prospective. POPULATION: Fourteen individuals with premanifest HD (38.0 ± 11.0 years, 9 females; far-from-onset N = 6, near-onset N = 8), 21 manifest HD patients (49.1 ± 12.1 years, 11 females), and 33 age-matched healthy controls (43.9 ± 12.2 years, 17 females). FIELD STRENGTH/SEQUENCE: 7 T, T1 -weighted imaging, quantitative susceptibility mapping, and diffusion tensor imaging. ASSESSMENT: Volume, susceptibility, fractional anisotropy (FA), and mean diffusivity (MD) within subcortical brain structures were compared across groups, used to establish HD classification models, and correlated to clinical measures and cognitive assessments. STATISTICAL TESTS: Generalized linear model, multivariate logistic regression, receiver operating characteristics with the area under the curve (AUC), and likelihood ratio test comparing a volumetric model to one that also includes susceptibility and diffusion metrics, Wilcoxon paired signed-rank test, and Pearson's correlation. A P-value <0.05 after Benjamini-Hochberg correction was considered statistically significant. RESULTS: Significantly higher striatal susceptibility and FA were found in premanifest and manifest HD preceding atrophy, even in far-from-onset premanifest HD compared to controls (putamen susceptibility: 0.027 ± 0.022 vs. 0.018 ± 0.013 ppm; FA: 0.358 ± 0.048 vs. 0.313 ± 0.039). The model with additional susceptibility, FA, and MD features showed higher AUC compared to volume features alone when differentiating premanifest HD from HC (0.83 vs. 0.66), and manifest from premanifest HD (0.94 vs. 0.83). Higher striatal susceptibility significantly correlated with cognitive deterioration in HD (executive function: r = -0.600; socioemotional function: r = -0.486). DATA CONCLUSION: 7 T MRI revealed iron dysregulation and microstructure alterations with HD progression, which could precede volume loss, provide added value to HD differentiation, and might be associated with cognitive changes. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 2.

Multicenter Study of Long-Term Outcomes and Quality of Life in PHACE Syndrome after Age 10.

OBJECTIVE: To characterize long-term outcomes of PHACE syndrome. STUDY DESIGN: Multicenter study with cross-sectional interviews and chart review of individuals with definite PHACE syndrome ≥10 years of age. Data from charts were collected across multiple PHACE-related topics. Data not available in charts were collected from patients directly. Likert scales were used to assess the impact of specific findings. Patient-Reported Outcomes Measurement Information System (PROMIS) scales were used to assess quality of life domains. RESULTS: A total of 104/153 (68%) individuals contacted participated in the study at a median of 14 years of age (range 10-77 years). There were infantile hemangioma (IH) residua in 94.1%. Approximately one-half had received laser treatment for residual IH, and the majority (89.5%) of participants were satisfied or very satisfied with the appearance. Neurocognitive manifestations were common including headaches/migraines (72.1%), participant-reported learning differences (45.1%), and need for individualized education plans (39.4%). Cerebrovascular arteriopathy was present in 91.3%, with progression identified in 20/68 (29.4%) of those with available follow-up imaging reports. Among these, 6/68 (8.8%) developed moyamoya vasculopathy or progressive stenoocclusion, leading to isolated circulation at or above the level of the circle of Willis. Despite the prevalence of cerebrovascular arteriopathy, the proportion of those with ischemic stroke was low (2/104; 1.9%). PROMIS global health scores were lower than population norms by at least 1 SD. CONCLUSIONS: PHACE syndrome is associated with long-term, mild to severe morbidities including IH residua, headaches, learning differences, and progressive arteriopathy. Primary and specialty follow-up care is critical for PHACE patients into adulthood.

Correlating Quantitative MRI-based Apparent Diffusion Coefficient Metrics with 24-month Neurodevelopmental Outcomes in Neonates from the HEAL Trial.

Background Multiple qualitative scoring systems have been created to capture the imaging severity of hypoxic ischemic brain injury. Purpose To evaluate quantitative volumes of acute brain injury at MRI in neonates with hypoxic ischemic brain injury and correlate these findings with 24-month neurodevelopmental outcomes and qualitative brain injury scoring by radiologists. Materials and Methods In this secondary analysis, brain diffusion-weighted MRI data from neonates in the High-dose Erythropoietin for Asphyxia and Encephalopathy trial, which recruited participants between January 2017 and October 2019, were analyzed. Volume of acute brain injury, defined as brain with apparent diffusion coefficient (ADC) less than 800 × 10-6 mm2/sec, was automatically computed across the whole brain and within the thalami and white matter. Outcomes of death and neurodevelopmental impairment (NDI) were recorded at 24-month follow-up. Associations between the presence and volume (in milliliters) of acute brain injury with 24-month outcomes were evaluated using multiple logistic regression. The correlation between quantitative acute brain injury volume and qualitative MRI scores was assessed using the Kendall tau-b test. Results A total of 416 neonates had available MRI data (mean gestational age, 39.1 weeks ± 1.4 [SD]; 235 male) and 113 (27%) showed evidence of acute brain injury at MRI. Of the 387 participants with 24-month follow-up data, 185 (48%) died or had any NDI. Volume of acute injury greater than 1 mL (odds ratio [OR], 13.9 [95% CI: 5.93, 32.45]; P < .001) and presence of any acute injury in the brain (OR, 4.5 [95% CI: 2.6, 7.8]; P < .001) were associated with increased odds of death or any NDI. Quantitative whole-brain acute injury volume was strongly associated with radiologists' qualitative scoring of diffusion-weighted images (Kendall tau-b = 0.56; P < .001). Conclusion Automated quantitative volume of brain injury is associated with death, moderate to severe NDI, and cerebral palsy in neonates with hypoxic ischemic encephalopathy and correlated well with qualitative MRI scoring of acute brain injury. Clinical trial registration no. NCT02811263 © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Huisman in this issue.

Implementation and prospective real-time evaluation of a generalized system for in-clinic deployment and validation of machine learning models in radiology.

The medical imaging community has embraced Machine Learning (ML) as evidenced by the rapid increase in the number of ML models being developed, but validating and deploying these models in the clinic remains a challenge. The engineering involved in integrating and assessing the efficacy of ML models within the clinical workflow is complex. This paper presents a general-purpose, end-to-end, clinically integrated ML model deployment and validation system implemented at UCSF. Engineering and usability challenges and results from 3 use cases are presented. A generalized validation system based on free, open-source software (OSS) was implemented, connecting clinical imaging modalities, the Picture Archiving and Communication System (PACS), and an ML inference server. ML pipelines were implemented in NVIDIA's Clara Deploy framework with results and clinician feedback stored in a customized XNAT instance, separate from the clinical record but linked from within PACS. Prospective clinical validation studies of 3 ML models were conducted, with data routed from multiple clinical imaging modalities and PACS. Completed validation studies provided expert clinical feedback on model performance and usability, plus system reliability and performance metrics. Clinical validation of ML models entails assessing model performance, impact on clinical infrastructure, robustness, and usability. Study results must be easily accessible to participating clinicians but remain outside the clinical record. Building a system that generalizes and scales across multiple ML models takes the concerted effort of software engineers, clinicians, data scientists, and system administrators, and benefits from the use of modular OSS. The present work provides a template for institutions looking to translate and clinically validate ML models in the clinic, together with required resources and expected challenges.

Ecodesign and Operational Strategies to Reduce the Carbon Footprint of MRI for Energy Cost Savings.

Background Radiology is a major contributor to health care's climate footprint due to energy-intensive devices, particularly MRI, which uses the most energy. Purpose To determine the energy, cost, and carbon savings that could be achieved through different scanner power management strategies. Materials and Methods In this retrospective evaluation, four outpatient MRI scanners from three vendors were individually equipped with power meters (1-Hz sampling rate). Power measurement logs were extracted for 39 days. Data were segmented into off, idle, prepared-to-scan, scan, or power-save modes for each scanner. Energy, cost (assuming a mean cost of $0.14 per kilowatt hour), and carbon savings were calculated for the lowest scanner activity modes. Data were summarized using descriptive statistics and 95% CIs. Results Projected annual energy consumption per scanner ranged from 82.7 to 171.1 MW-hours, with 72%-91% defined as nonproductive. Power draws for each mode were measured as 6.4 kW ± 0.1 (SD; power-save mode), 7.3 kW ± 0.6 to 9.7 kW ± 0.2 (off), 9.5 kW ± 0.9 to 14.5 kW ± 0.5 (idle), 17.3 kW ± 0.5 to 25.6 kW ± 0.6 (prepared-to-scan mode), and 28.6 kW ± 8.6 to 48.3 kW ± 11.8 (scan mode). Switching MRI units from idle to off mode for 12 hours overnight reduced power consumption by 25%-33%, translating to a potential annual savings of 12.3-21.0 MW-hours, $1717-$2943, and 8.7-14.9 metric tons of carbon dioxide (CO2) equivalent (MTCO2eq). The power-save mode further reduced consumption by 22%-28% compared with off mode, potentially saving an additional 8.8-11.4 MW-hours, $1226-$1594, and 6.2-8.1 MTCO2eq per year for 12 hours overnight. Implementation of a power-save mode for 12 hours overnight in all outpatient MRI units in the United States could save U.S. health care 58 863.2-76 288.2 MW-hours, $8.2-$10.7 million, and 41 606.4-54 088.3 MTCO2eq. Conclusion Powering down MRI units made radiology departments more energy efficient and showed substantial sustainability and cost benefits. © RSNA, 2023 Supplemental material is available for this article. See also the article by Vosshenrich and Heye in this issue.

What referring clinicians value most: Accuracy of radiology results and personal interactions with radiologists.

PURPOSE: We sought to identify which aspects of the referring clinician experience are most strongly correlated with overall satisfaction, and hence of greatest relevant importance to referring clinicians. METHODS: A survey instrument assessing referring clinician satisfaction throughout 11 domains of the radiology process map was distributed 2720 clinicians. The survey contained sections assessing each process map domain, with each section including a question about satisfaction overall in that domain and multiple more granular questions. The final question on the survey was overall satisfaction with the department. Univariate logistic regression and multivariate logistic regression were performed to assess the association between individual survey questions and overall satisfaction with the department. RESULTS: 729 referring clinicians (27%) completed the survey. Using univariate logistic regression nearly every question was associated with overall satisfaction. Amongst the 11 domains of the radiology process map multivariate logistic regression identified the following as mostly strongly associated with overall satisfaction: results/reporting overall (odds ratio 4.71; 95% confidence interval 2.15-10.23), section with which work most closely overall (3.39; 1.28-8.64), and inpatient radiology overall (2.39; 1.08-5.08). Other survey questions associated with overall satisfaction on multivariate logistic regression were attending radiologist interactions (odds ratio 3.71; 95% confidence interval 1.54-8.69), timeliness of inpatient radiology results (2.91; 1.01-8.09), technologist interactions (2.15; 0.99-4.40), appointment availability for urgent outpatient studies (2.01; 1.08-3.64), and guidance for selecting correct imaging study (1.88; 1.04-3.34). CONCLUSION: Referring clinicians value most the accuracy of the radiology report and their interactions with attending radiologists, particularly within the section they work most closely.

DeepSWI: Using Deep Learning to Enhance Susceptibility Contrast on T2*-Weighted MRI.

BACKGROUND: Although susceptibility-weighted imaging (SWI) is the gold standard for visualizing cerebral microbleeds (CMBs) in the brain, the required phase data are not always available clinically. Having a postprocessing tool for generating SWI contrast from T2*-weighted magnitude images is therefore advantageous. PURPOSE: To create synthetic SWI images from clinical T2*-weighted magnitude images using deep learning and evaluate the resulting images in terms of similarity to conventional SWI images and ability to detect radiation-associated CMBs. STUDY TYPE: Retrospective. POPULATION: A total of 145 adults (87 males/58 females; 43.9 years old) with radiation-associated CMBs were used to train (16,093 patches/121 patients), validate (484 patches/4 patients), and test (2420 patches/20 patients) our networks. FIELD STRENGTH/SEQUENCE: 3D T2*-weighted, gradient-echo acquired at 3 T. ASSESSMENT: Structural similarity index (SSIM), peak signal-to-noise-ratio (PSNR), normalized mean-squared-error (nMSE), CMB counts, and line profiles were compared among magnitude, original SWI, and synthetic SWI images. Three blinded raters (J.E.V.M., M.A.M., B.B. with 8-, 6-, and 4-years of experience, respectively) independently rated and classified test-set images. STATISTICAL TESTS: Kruskall-Wallis and Wilcoxon signed-rank tests were used to compare SSIM, PSNR, nMSE, and CMB counts among magnitude, original SWI, and predicted synthetic SWI images. Intraclass correlation assessed interrater variability. P values <0.005 were considered statistically significant. RESULTS: SSIM values of the predicted vs. original SWI (0.972, 0.995, 0.9864) were statistically significantly higher than that of the magnitude vs. original SWI (0.970, 0.994, 0.9861) for whole brain, vascular structures, and brain tissue regions, respectively; 67% (19/28) CMBs detected on original SWI images were also detected on the predicted SWI, whereas only 10 (36%) were detected on magnitude images. Overall image quality was similar between the synthetic and original SWI images, with less artifacts on the former. CONCLUSIONS: This study demonstrated that deep learning can increase the susceptibility contrast present in neurovasculature and CMBs on T2*-weighted magnitude images, without residual susceptibility-induced artifacts. This may be useful for more accurately estimating CMB burden from magnitude images alone. EVIDENCE LEVEL: 3. TECHNICAL EFFICACY: Stage 2.

Multi-tasking deep network for tinnitus classification and severity prediction from multimodal structural MR images.

Objective:Subjective tinnitus is an auditory phantom perceptual disorder without an objective biomarker. Fast and efficient diagnostic tools will advance clinical practice by detecting or confirming the condition, tracking change in severity, and monitoring treatment response. Motivated by evidence of subtle anatomical, morphological, or functional information in magnetic resonance images of the brain, we examine data-driven machine learning methods for joint tinnitus classification (tinnitus or no tinnitus) and tinnitus severity prediction.Approach:We propose a deep multi-task multimodal framework for tinnitus classification and severity prediction using structural MRI (sMRI) data. To leverage complementary information multimodal neuroimaging data, we integrate two modalities of three-dimensional sMRI-T1 weighted (T1w) and T2 weighted (T2w) images. To explore the key components in the MR images that drove task performance, we segment both T1w and T2w images into three different components-cerebrospinal fluid, grey matter and white matter, and evaluate performance of each segmented image.Main results:Results demonstrate that our multimodal framework capitalizes on the information across both modalities (T1w and T2w) for the joint task of tinnitus classification and severity prediction.Significance:Our model outperforms existing learning-based and conventional methods in terms of accuracy, sensitivity, specificity, and negative predictive value.

Comparison of quantitative susceptibility mapping methods for iron-sensitive susceptibility imaging at 7T: An evaluation in healthy subjects and patients with Huntington's disease.

Quantitative susceptibility mapping (QSM) is a promising tool for investigating iron dysregulation in neurodegenerative diseases, including Huntington's disease (HD). Many diverse methods have been proposed to generate accurate and robust QSM images. In this study, we evaluated the performance of different dipole inversion algorithms for iron-sensitive susceptibility imaging at 7T on healthy subjects of a large age range and patients with HD. We compared an iterative least-squares-based method (iLSQR), iterative methods that use regularization, single-step approaches, and deep learning-based techniques. Their performance was evaluated by comparing: (1) deviations from a multiple-orientation QSM reference; (2) visual appearance of QSM maps and the presence of artifacts; (3) susceptibility in subcortical brain regions with age; (4) regional brain susceptibility with published postmortem brain iron quantification; and (5) susceptibility in HD-affected basal ganglia regions between HD subjects and healthy controls. We found that single-step QSM methods with either total variation or total generalized variation constraints (SSTV/SSTGV) and the single-step deep learning method iQSM generally provided the best performance in terms of correlation with iron deposition and were better at differentiating between healthy controls and premanifest HD individuals, while deep learning QSM methods trained with multiple-orientation susceptibility data created QSM maps that were most similar to the multiple orientation reference and with the best visual scores.

The University of California San Francisco Preoperative Diffuse Glioma MRI Dataset.

Supplemental material is available for this article. Keywords: Informatics, MR Diffusion Tensor Imaging, MR Perfusion, MR Imaging, Neuro-Oncology, CNS, Brain/Brain Stem, Oncology, Radiogenomics, Radiology-Pathology Integration © RSNA, 2022.

Longitudinal Assessment of Posttreatment Diffuse Glioma Tissue Volumes with Three-dimensional Convolutional Neural Networks.

Neural networks were trained for segmentation and longitudinal assessment of posttreatment diffuse glioma. A retrospective cohort (from January 2018 to December 2019) of 298 patients with diffuse glioma (mean age, 52 years ± 14 [SD]; 177 men; 152 patients with glioblastoma, 72 patients with astrocytoma, and 74 patients with oligodendroglioma) who underwent two consecutive multimodal MRI examinations were randomly selected into training (n = 198) and testing (n = 100) samples. A posttreatment tumor segmentation three-dimensional nnU-Net convolutional neural network with multichannel inputs (T1, T2, and T1 postcontrast and fluid-attenuated inversion recovery [FLAIR]) was trained to segment three multiclass tissue types (peritumoral edematous, infiltrated, or treatment-changed tissue [ED]; active tumor or enhancing tissue [AT]; and necrotic core). Separate longitudinal change nnU-Nets were trained on registered and subtracted FLAIR and T1 postlongitudinal images to localize and better quantify and classify changes in ED and AT. Segmentation Dice scores, volume similarities, and 95th percentile Hausdorff distances ranged from 0.72 to 0.89, 0.90 to 0.96, and 2.5 to 3.6 mm, respectively. Accuracy rates of the posttreatment tumor segmentation and longitudinal change networks being able to classify longitudinal changes in ED and AT as increased, decreased, or unchanged were 76%-79% and 90%-91%, respectively. The accuracy levels of the longitudinal change networks were not significantly different from those of three neuroradiologists (accuracy, 90%-92%; κ, 0.58-0.63; P > .05). The results of this study support the potential clinical value of artificial intelligence-based automated longitudinal assessment of posttreatment diffuse glioma. Keywords: MR Imaging, Neuro-Oncology, Neural Networks, CNS, Brain/Brain Stem, Segmentation, Quantification, Convolutional Neural Network (CNN) Supplemental material is available for this article. © RSNA, 2022.

Functional network alterations in young brain tumor patients with radiotherapy-induced memory impairments and vascular injury.

Background: Cognitive impairment and cerebral microbleeds (CMBs) are long-term side-effects of cranial radiation therapy (RT). Previously we showed that memory function is disrupted in young patients and that the rate of cognitive decline correlates with CMB development. However, vascular injury alone cannot explain RT-induced cognitive decline. Here we use resting-state functional MRI (rsfMRI) to further investigate the complex mechanisms underlying memory impairment after RT. Methods: Nineteen young patients previously treated with or without focal or whole-brain RT for a brain tumor underwent cognitive testing followed by 7T rsfMRI and susceptibility-weighted imaging for CMB detection. Global brain modularity and efficiency, and rsfMRI signal variability within the dorsal attention, salience, and frontoparietal networks were computed. We evaluated whether MR metrics could distinguish age- and sex-matched controls (N = 19) from patients and differentiate patients based on RT exposure and aggressiveness. We also related MR metrics with memory performance, CMB burden, and risk factors for cognitive decline after RT. Results: Compared to controls, patients exhibited widespread hyperconnectivity, similar modularity, and significantly increased efficiency (p < 0.001) and network variability (p < 0.001). The most abnormal values were detected in patients treated with high dose whole-brain RT, having supratentorial tumors, and who did not undergo RT but had hydrocephalus. MR metrics and memory performance were correlated (R = 0.34-0.53), though MR metrics were more strongly related to risk factors for cognitive worsening and CMB burden with evidence of functional recovery. Conclusions: MR metrics describing brain connectivity and variability represent promising candidate imaging biomarkers for monitoring of long-term cognitive side-effects after RT.

Deep Learning to Predict Neonatal and Infant Brain Age from Myelination on Brain MRI Scans.

Background Assessment of appropriate brain myelination on T1- and T2-weighted MRI scans is based on gestationally corrected age (GCA) and requires subjective visual inspection of the brain with knowledge of normal myelination milestones. Purpose To develop a convolutional neural network (CNN) capable of estimating neonatal and infant GCA based on brain myelination on MRI scans. Materials and methods In this retrospective study from one academic medical center, brain MRI scans of patients aged 0-25 months with reported normal myelination were consecutively collected between January 1995 and June 2019. The GCA at MRI was manually calculated. After exclusion criteria were applied, T1- and T2-weighted MRI scans were preprocessed with skull stripping, linear registration, z scoring for normalization, and downsampling. A three-dimensional regression CNN was trained to predict GCA using mean absolute error (MAE) as its loss function. Attention maps were calculated using layer-wise relevance propagation. Models were validated on an external test set from the National Institutes of Health (NIH). Model MAEs were compared using Kruskal-Wallis and Mann-Whitney tests. Results A total of 518 neonates and infants (mean GCA, 67 weeks ± 33 [SD], 56% male) was included, comprising 469 T1-, 438 T2-, and 389 T1- and T2-weighted studies. Across 10 runs, MAEs of T1-, T2-, and T1- and T2-weighted networks were 9.8 ± 2.3, 9.1 ± 1.9, and 7.7 ± 1.7 weeks, respectively. Attention map analysis demonstrated increased network attention to the cerebellum, posterior white matter, and basal ganglia signal in neonates with GCA of less than 40 weeks and the anterior white matter signal in infants with GCA of more than 120 weeks, corresponding to the known progression of myelination. The T1- and T2-weighted network tested on the external NIH test set had an MAE of 9.1 weeks, which was reduced to 5.9 weeks with further training using half the external test set (P < .001). Conclusion A three-dimensional convolutional neural network can predict the gestationally corrected age of neonates and infants aged 0-25 months based on brain myelination patterns on T1- and T2-weighted MRI scans. © RSNA, 2022 Online supplemental material is available for this article.

Interinstitutional Portability of a Deep Learning Brain MRI Lesion Segmentation Algorithm.

PURPOSE: To assess how well a brain MRI lesion segmentation algorithm trained at one institution performed at another institution, and to assess the effect of multi-institutional training datasets for mitigating performance loss. MATERIALS AND METHODS: In this retrospective study, a three-dimensional U-Net for brain MRI abnormality segmentation was trained on data from 293 patients from one institution (IN1) (median age, 54 years; 165 women; patients treated between 2008 and 2018) and tested on data from 51 patients from a second institution (IN2) (median age, 46 years; 27 women; patients treated between 2003 and 2019). The model was then trained on additional data from various sources: (a) 285 multi-institution brain tumor segmentations, (b) 198 IN2 brain tumor segmentations, and (c) 34 IN2 lesion segmentations from various brain pathologic conditions. All trained models were tested on IN1 and external IN2 test datasets, assessing segmentation performance using Dice coefficients. RESULTS: The U-Net accurately segmented brain MRI lesions across various pathologic conditions. Performance was lower when tested at an external institution (median Dice score, 0.70 [IN2] vs 0.76 [IN1]). Addition of 483 training cases of a single pathologic condition, including from IN2, did not raise performance (median Dice score, 0.72; P = .10). Addition of IN2 training data with heterogeneous pathologic features, representing only 10% (34 of 329) of total training data, increased performance to baseline (Dice score, 0.77; P < .001). This final model produced total lesion volumes with a high correlation to the reference standard (Spearman r = 0.98). CONCLUSION: For brain MRI lesion segmentation, adding a modest amount of relevant training data from an external institution to a previously trained model supported successful application of the model to this external institution.Keywords: Neural Networks, Brain/Brain Stem, Segmentation Supplemental material is available for this article. © RSNA, 2021.

Clinical language search algorithm from free-text: facilitating appropriate imaging.

BACKGROUND: The comprehensiveness and maintenance of the American College of Radiology (ACR) Appropriateness Criteria (AC) makes it a unique resource for evidence-based clinical imaging decision support, but it is underutilized by clinicians. To facilitate the use of imaging recommendations, we develop a natural language processing (NLP) search algorithm that automatically matches clinical indications that physicians write into imaging orders to appropriate AC imaging recommendations. METHODS: We apply a hybrid model of semantic similarity from a sent2vec model trained on 223 million scientific sentences, combined with term frequency inverse document frequency features. AC documents are ranked based on their embeddings' cosine distance to query. For model testing, we compiled a dataset of simulated simple and complex indications for each AC document (n = 410) and another with clinical indications from randomly sampled radiology reports (n = 100). We compare our algorithm to a custom google search engine. RESULTS: On the simulated indications, our algorithm ranked ground truth documents as top 3 for 98% of simple queries and 85% of complex queries. Similarly, on the randomly sampled radiology report dataset, the algorithm ranked 86% of indications with a single match as top 3. Vague and distracting phrases present in the free-text indications were main sources of errors. Our algorithm provides more relevant results than a custom Google search engine, especially for complex queries. CONCLUSIONS: We have developed and evaluated an NLP algorithm that matches clinical indications to appropriate AC guidelines. This approach can be integrated into imaging ordering systems for automated access to guidelines.