Crowdsourcing pneumothorax annotations using machine learning annotations on the NIH chest X-ray dataset.

Pneumothorax is a potentially life-threatening condition that requires prompt recognition and often urgent intervention. In the ICU setting, large numbers of chest radiographs are performed and must be interpreted on a daily basis which may delay diagnosis of this entity. Development of artificial intelligence (AI) techniques to detect pneumothorax could help expedite detection as well as localize and potentially quantify pneumothorax. Open image analysis competitions are useful in advancing state-of-the art AI algorithms but generally require large expert annotated datasets. We have annotated and adjudicated a large dataset of chest radiographs to be made public with the goal of sparking innovation in this space. Because of the cumbersome and time-consuming nature of image labeling, we explored the value of using AI models to generate annotations for review. Utilization of this machine learning annotation (MLA) technique appeared to expedite our annotation process with relatively high sensitivity at the expense of specificity. Further research is required to confirm and better characterize the value of MLAs. Our adjudicated dataset is now available for public consumption in the form of a challenge.

Integrating Natural Language Processing and Machine Learning Algorithms to Categorize Oncologic Response in Radiology Reports.

A significant volume of medical data remains unstructured. Natural language processing (NLP) and machine learning (ML) techniques have shown to successfully extract insights from radiology reports. However, the codependent effects of NLP and ML in this context have not been well-studied. Between April 1, 2015 and November 1, 2016, 9418 cross-sectional abdomen/pelvis CT and MR examinations containing our internal structured reporting element for cancer were separated into four categories: Progression, Stable Disease, Improvement, or No Cancer. We combined each of three NLP techniques with five ML algorithms to predict the assigned label using the unstructured report text and compared the performance of each combination. The three NLP algorithms included term frequency-inverse document frequency (TF-IDF), term frequency weighting (TF), and 16-bit feature hashing. The ML algorithms included logistic regression (LR), random decision forest (RDF), one-vs-all support vector machine (SVM), one-vs-all Bayes point machine (BPM), and fully connected neural network (NN). The best-performing NLP model consisted of tokenized unigrams and bigrams with TF-IDF. Increasing N-gram length yielded little to no added benefit for most ML algorithms. With all parameters optimized, SVM had the best performance on the test dataset, with 90.6 average accuracy and F score of 0.813. The interplay between ML and NLP algorithms and their effect on interpretation accuracy is complex. The best accuracy is achieved when both algorithms are optimized concurrently.

Cardiac CT angiography in the emergency department.

OBJECTIVE. Nearly 8 million patients present annually to emergency departments (EDs) in the United States with acute chest pain. Identifying those with a sufficiently low risk of acute coronary syndrome (ACS) remains challenging. Early imaging is important for risk stratification of these individuals. The objective of this article is to discuss the role of cardiac CT angiography (CTA) as a safe, efficient, and cost-effective tool in this setting and review state-of-the-art technology, protocols, advantages, and limitations from the perspective of our institution's 10-year experience. CONCLUSION. Early utilization of cardiac CTA in patients presenting to the ED with chest pain and a low to intermediate risk of ACS quickly identifies a group of particularly low-risk patients (< 1% risk of adverse events within 30 days) and allows safe and expedited discharge. By preventing unnecessary admissions and prolonged lengths of stay, a strategy based on early cardiac CTA has been shown to be efficient, although potential overutilization and other issues require long-term study.

Reproducibility of volume and densitometric measures of emphysema on repeat computed tomography with an interval of 1 week.

OBJECTIVES: The reproducibilities of CT lung volume and densitometric measures of emphysema were assessed over 1 week. The influence of breathhold on reproducibility was assessed. METHODS: HRCT was performed on 44 subjects at inspiration on two visits with a 7-day interval. CT lung volume, relative area below -950HU (RA950-raw), and 15th percentile density (PD15-raw) were computed. Volume correction was used to obtain RA950-adj and PD15-adj. Reproducibilities between visits were assessed using concordance correlation coefficient (CCC) and repeatability coefficient (RC). Reproducibilities were compared between raw and adjusted measures. Differences between visits were computed for volume and density measures. Correlations were computed for density differences versus volume difference. Subgroup analysis was performed using a 0.25 L volume difference threshold. RESULTS: High CCC were observed for all measures in full group (CCC > 0.97). Reproducibilities of volume (RC = 0.67 L), RA950-raw (RC = 2.3%), and PD15-raw (RC = 10.6HU) were observed. Volume correction significantly improved PD15 (RC = 3.6HU) but not RA950 (RC = 1.7%). RA950-raw and PD15-raw had significantly better RC in <0.25 L subgroup than ≥0.25 L. Significant correlations with volume were observed for RA950-raw and PD15-raw (R (2) > 0.71), but not RA950-adj or PD15-adj (R (2) < 0.11). CONCLUSIONS: Good breathhold and RA950 reproducibilities were achieved. PD15 was less reproducible but improved with volume correction or superior breathhold reproduction. KEY POINTS: • Good breath-hold reproducibility is achievable between multiple CT examinations. • Reproducibility of densitometric measures may be improved by statistical volume correction. • Volume correction may result in decreased signal. • Densitometric reproducibility may also be improved by achieving good breath-hold reproduction. • Careful consideration of signal and noise is necessary in reproducibility assessment.

A method for the automatic quantification of the completeness of pulmonary fissures: evaluation in a database of subjects with severe emphysema.

OBJECTIVES: To propose and evaluate a technique for automatic quantification of fissural completeness from chest computed tomography (CT) in a database of subjects with severe emphysema. METHODS: Ninety-six CT studies of patients with severe emphysema were included. The lungs, fissures and lobes were automatically segmented. The completeness of the fissures was calculated as the percentage of the lobar border defined by a fissure. The completeness score of the automatic method was compared with a visual consensus read by three radiologists using boxplots, rank sum tests and ROC analysis. RESULTS: The consensus read found 49% (47/96), 15% (14/96) and 67% (64/96) of the right major, right minor and left major fissures to be complete. For all fissures visually assessed as being complete the automatic method resulted in significantly higher completeness scores (mean 92.78%) than for those assessed as being partial or absent (mean 77.16%; all p values <0.001). The areas under the curves for the automatic fissural completeness were 0.88, 0.91 and 0.83 for the right major, right minor and left major fissures respectively. CONCLUSIONS: An automatic method is able to quantify fissural completeness in a cohort of subjects with severe emphysema consistent with a visual consensus read of three radiologists. KEY POINTS: • Lobar fissures are important for assessing the extent and distribution of lung disease • Modern CT allows automatic lobar segmentation and assessment of the fissures • This segmentation can also assess the completeness of the fissures. • Such assessment is important for decisions about novel therapies (eg for emphysema).

Emphysema lung lobe volume reduction: effects on the ipsilateral and contralateral lobes.

OBJECTIVES: To investigate volumetric and density changes in the ipsilateral and contralateral lobes following volume reduction of an emphysematous target lobe. METHODS: The study included 289 subjects with heterogeneous emphysema, who underwent bronchoscopic volume reduction of the most diseased lobe with endobronchial valves and 132 untreated controls. Lobar volume and low-attenuation relative area (RA) changes post-procedure were measured from computed tomography images. Regression analysis (Spearman's rho) was performed to test the association between change in the target lobe volume and changes in volume and density variables in the other lobes. RESULTS: The target lobe volume at full inspiration in the treatment group had a mean reduction of -0.45 L (SE = 0.034, P < 0.0001), and was associated with volume increases in the ipsilateral lobe (rho = -0.68, P < 0.0001) and contralateral lung (rho = -0.16, P = 0.006), and overall reductions in expiratory RA (rho = 0.31, P < 0.0001) and residual volume (RV)/total lung capacity (TLC) (rho = 0.13, P = 0.03). CONCLUSIONS: When the volume of an emphysematous target lobe is reduced, the volume is redistributed primarily to the ipsilateral lobe, with an overall reduction. Image-based changes in lobar volumes and densities indicate that target lobe volume reduction is associated with statistically significant overall reductions in air trapping, consistent with expansion of the healthier lung. KEY POINTS: Computed tomography allows assessment of the treatment of emphysema with endobronchial valves. • Endobronchial valves can reduce the volume of an emphysematous lung lobe. • Compensatory expansion is greater in ipsilateral lobes than in the contralateral lung. • Reduced air trapping is measurable by RV/TLC and smaller low attenuation area.

A combined pulmonary-radiology workshop for visual evaluation of COPD: study design, chest CT findings and concordance with quantitative evaluation.

UNLABELLED: The purposes of this study were: to describe chest CT findings in normal non-smoking controls and cigarette smokers with and without COPD; to compare the prevalence of CT abnormalities with severity of COPD; and to evaluate concordance between visual and quantitative chest CT (QCT) scoring. METHODS: Volumetric inspiratory and expiratory CT scans of 294 subjects, including normal non-smokers, smokers without COPD, and smokers with GOLD Stage I-IV COPD, were scored at a multi-reader workshop using a standardized worksheet. There were 58 observers (33 pulmonologists, 25 radiologists); each scan was scored by 9-11 observers. Interobserver agreement was calculated using kappa statistic. Median score of visual observations was compared with QCT measurements. RESULTS: Interobserver agreement was moderate for the presence or absence of emphysema and for the presence of panlobular emphysema; fair for the presence of centrilobular, paraseptal, and bullous emphysema subtypes and for the presence of bronchial wall thickening; and poor for gas trapping, centrilobular nodularity, mosaic attenuation, and bronchial dilation. Agreement was similar for radiologists and pulmonologists. The prevalence on CT readings of most abnormalities (e.g. emphysema, bronchial wall thickening, mosaic attenuation, expiratory gas trapping) increased significantly with greater COPD severity, while the prevalence of centrilobular nodularity decreased. Concordances between visual scoring and quantitative scoring of emphysema, gas trapping and airway wall thickening were 75%, 87% and 65%, respectively. CONCLUSIONS: Despite substantial inter-observer variation, visual assessment of chest CT scans in cigarette smokers provides information regarding lung disease severity; visual scoring may be complementary to quantitative evaluation.

CT Features of Lymphatic Plastic Bronchitis in Adults: Correlation with Multimodality Lymphatic Imaging.

Purpose: To distinguish CT patterns of lymphatic and nonlymphatic causes of plastic bronchitis (PB) through comparison with lymphatic imaging. Materials and Methods: In this retrospective study, chest CT images acquired prior to lymphatic workup were assessed in 44 patients with PB from January 2014 to August 2020. The location and extent of ground-glass opacity (GGO) was compared with symptoms and lymphatic imaging. Statistical analysis was performed using descriptive statistics, logistic regression, Pearson correlation coefficient, and unweighted κ coefficient for interobserver agreement. Sensitivity and specificity of GGO for lymphatic PB were calculated. Results: Lymphatic imaging was performed in 44 patients (median age, 52 years ± 21 [IQR]; 23 women): 35 with lymphatic PB and nine with nonlymphatic PB. GGO was more frequently observed in patients with lymphatic PB than in those with nonlymphatic PB (91% [32 of 35] vs 33% [three of nine]; P < .001). Univariate logistic regression confirmed this result by showing that GGO was a significant predictor of lymphatic PB (odds ratio, 21 (95% CI: 3.8, 159.7). The model areas under the receiver operating characteristic curve (AUCs) of GGO unadjusted and adjusted for demographics were 0.79 and 0.86, respectively. The location of GGO correlated with lymphatic imaging and bronchoscopic findings. Overall sensitivity and specificity of GGO for lymphatic PB were 91% (32 of 35; 95% CI: 76, 98) and 67% (six of nine; 95% CI: 30, 93), respectively. Conclusion: Patients with lymphatic PB predominantly had multifocal GGO with or without a "crazy paving" pattern; identification of GGO should prompt lymphatic workup in this frequently misdiagnosed condition.Keywords: Lymphography, Lymphatic, CT, Tracheobronchial Tree, Thorax© RSNA, 2022See also commentary by Kligerman and White in this issue.

Multimodality imaging of Surgicel®, an important mimic of post-operative complication in the thorax.

Absorbable hemostatic agents such as Surgicel are hemostatic materials composed of an oxidized cellulose polymer used to control post-surgical bleeding and cause coagulation. This material is sometimes purposefully left in situ where it slowly degrades over time and can produce an imaging appearance that mimics serious post-operative complications such as gangrenous infections and anastomotic leaks as well as potentially mimicking disease recurrence in later stages. In this article, we review the multimodality imaging appearance of this material in situ longitudinally in the range of post-operative settings, in order to promote awareness of this entity when interpreting post-operative imaging. We present this as a pictorial review focusing primarily but not exclusively on the chest noting that the thoracic imaging appearance of Surgicel® is less well reported in the published literature. An understanding of this entity may help to minimize erroneous diagnosis of a postoperative complication leading to unnecessary interventions.

A 53-Year-Old Man Presents to the ED With Shortness of Breath, Cough, and Fever.

CASE PRESENTATION: A 53-year-old man presented to the ED at a time of low severe acute respiratory syndrome coronavirus 2, also known as coronavirus disease 2019 (COVID-19), prevalence and reported 2 weeks of progressive shortness of breath, dry cough, headache, myalgias, diarrhea, and recurrent low-grade fevers to 39°C for 1 week with several days of recorded peripheral capillary oxygen saturation of 80% to 90% (room air) on home pulse oximeter. Five days earlier, he had visited an urgent care center where a routine respiratory viral panel was reportedly negative. A COVID-19 reverse transcriptase polymerase chain reaction test result was pending at the time of ED visit. He reported a past medical history of gastroesophageal reflux disease that was treated with famotidine. Travel history included an out-of-state trip 3 weeks earlier, but no recent international travel.

Preliminary Assessment of an Optical Flow Method (OFM) for Nonrigid Registration and Temporal Subtraction (TS) of Serial CT Examinations to Facilitate Evaluation of Interval Change in Metastatic Lung Nodules.

RATIONALE AND OBJECTIVES: Accurate assessment of size change of lung nodules on chest computed tomography (CT) is important for diagnosis and response assessment. However, manual methods are time-consuming and error-prone. We therefore assessed whether an optical flow method (OFM) with temporal subtraction (TS) can facilitate detection and quantification of lung nodule change on serial CT datasets. MATERIALS AND METHODS: Serial chest CT examinations were selected from 12 patients with multiple pulmonary metastases. Lung nodules were evaluated for change in size using: (1) OFM with TS and (2) reference standard visual and manual assessment. Average time required to assess interval change using both methods was recorded and compared. Concordance of agreement between OFM with TS and reference standard assessment for nodule change was examined. RESULTS: 285 solid pulmonary nodules were evaluated. The average time per nodule to assess interval change in nodule size by OFM with TS (mean 1.15 + 0.5 minutes) was significantly less (P = 0.02) than that the reference standard approach (mean 1.56 + 0.5 minutes). Agreement between OFM with TS and reference standard occurred for 63.2% of nodules overall (kappa = 0.50, standard error 0.35, P< 0.00001), and significantly increased with larger nodule size (kappa = 0.48 for nodules <5 mm; kappa = 0.94 for nodules >20 mm, P < 0.0001). CONCLUSIONS: This preliminary study demonstrates the feasibility of an OFM with TS to assess for interval change in metastatic lung nodules on serial CT examinations with significantly improved reading speed and moderate agreement relative to reference standard assessment. Agreement improved with larger nodule size.

Imaging of Cardiac Infections: A Comprehensive Review and Investigation Flowchart for Diagnostic Workup.

Infections of the cardiovascular system may present with nonspecific symptoms, and it is common for patients to undergo multiple investigations to arrive at the diagnosis. Echocardiography is central to the diagnosis of endocarditis and pericarditis. However, cardiac computed tomography (CT) and magnetic resonance imaging also play an additive role in these diagnoses; in fact, magnetic resonance imaging is central to the diagnosis of myocarditis. Functional imaging (fluorine-18 fluorodeoxyglucose-positron emission tomography/CT and radiolabeled white blood cell single-photon emission computed tomography/CT) is useful in the diagnosis in prosthesis-related and disseminated infection. This pictorial review will detail the most commonly encountered cardiovascular bacterial and viral infections, including coronavirus disease-2019, in clinical practice and provide an evidence basis for the selection of each imaging modality in the investigation of native tissues and common prostheses.

Integration of Risk Survival Measures Estimated From Pre- and Posttreatment Computed Tomography Scans Improves Stratification of Patients With Early-Stage Non-small Cell Lung Cancer Treated With Stereotactic Body Radiation Therapy.

PURPOSE: To predict overall survival of patients receiving stereotactic body radiation therapy (SBRT) for early-stage non-small cell lung cancer (ES-NSCLC), we developed a radiomic model that integrates risk of death estimates and changes based on pre- and posttreatment computed tomography (CT) scans. We hypothesize this innovation will improve our ability to stratify patients into various oncologic outcomes with greater accuracy. METHODS AND MATERIALS: Two cohorts of patients with ES-NSCLC uniformly treated with SBRT (a median dose of 50 Gy in 4-5 fractions) were studied. Prediction models were built on a discovery cohort of 100 patients with treatment planning CT scans, and then were applied to a separate validation cohort of 60 patients with pre- and posttreatment CT scans for evaluating their performance. RESULTS: Prediction models achieved a c-index up to 0.734 in predicting survival outcomes of the validation cohort. The integration of the pretreatment risk of survival measures (risk-high vs risk-low) and changes (risk-increase vs risk-decrease) in risk of survival measures between the pretreatment and posttreatment scans further stratified the patients into 4 subgroups (risk: high, increase; risk: high, decrease; risk: low, increase; risk: low, decrease) with significant difference (χ2 = 18.549, P = .0003, log-rank test). There was also a significant difference between the risk-increase and risk-decrease groups (χ2 = 6.80, P = .0091, log-rank test). In addition, a significant difference (χ2 = 7.493, P = .0062, log-rank test) was observed between the risk-high and risk-low groups obtained based on the pretreatment risk of survival measures. CONCLUSION: The integration of risk of survival measures estimated from pre- and posttreatment CT scans can help differentiate patients with good expected survival from those who will do more poorly following SBRT. The analysis of these radiomics-based longitudinal risk measures may help identify patients with early-stage NSCLC who will benefit from adjuvant treatment after lung SBRT, such as immunotherapy.

Impact of Interobserver Variability in Manual Segmentation of Non-Small Cell Lung Cancer (NSCLC) Applying Low-Rank Radiomic Representation on Computed Tomography.

This study tackles interobserver variability with respect to specialty training in manual segmentation of non-small cell lung cancer (NSCLC). Four readers included for segmentation are: a data scientist (BY), a medical student (LS), a radiology trainee (MH), and a specialty-trained radiologist (SK) for a total of 293 patients from two publicly available databases. Sørensen-Dice (SD) coefficients and low rank Pearson correlation coefficients (CC) of 429 radiomics were calculated to assess interobserver variability. Cox proportional hazard (CPH) models and Kaplan-Meier (KM) curves of overall survival (OS) prediction for each dataset were also generated. SD and CC for segmentations demonstrated high similarities, yielding, SD: 0.79 and CC: 0.92 (BY-SK), SD: 0.81 and CC: 0.83 (LS-SK), and SD: 0.84 and CC: 0.91 (MH-SK) in average for both databases, respectively. OS through the maximal CPH model for the two datasets yielded c-statistics of 0.7 (95% CI) and 0.69 (95% CI), while adding radiomic and clinical variables (sex, stage/morphological status, and histology) together. KM curves also showed significant discrimination between high- and low-risk patients (p-value < 0.005). This supports that readers' level of training and clinical experience may not significantly influence the ability to extract accurate radiomic features for NSCLC on CT. This potentially allows flexibility in the training required to produce robust prognostic imaging biomarkers for potential clinical translation.

Delayed-Phase Enhancement for Evaluation of Malignant Pleural Mesothelioma on Computed Tomography: A Prospective Cohort Study.

BACKGROUND: Radiologic assessment of malignant pleural mesothelioma (MPM) on computed tomography (CT) imaging can be limited by similar attenuations of MPM and adjacent tissues. This can result in inaccuracies in defining the presence and extent of pleural tumor burden. We hypothesized that increasing the time delay for pleural enhancement will optimize discrimination between MPM and noncancerous tissues on CT. Here we conduct a prospective observational study to determine the optimal time delay for imaging MPM on CT. PATIENTS AND METHODS: Adult MPM patients (n = 15) were enrolled in this prospective exploratory imaging trial. Patients with < 1 cm MPM thickness, prior pleurectomy, pleurodesis, pleural radiotherapy, or antiangiogenic therapy were excluded. All patients underwent a dynamically-enhanced CT with multiple time delays (0 - 10 minutes) after intravenous contrast administration. Tumor tissue attenuation was measured at each phase of enhancement. A qualitative assessment of tumor enhancement kinetics was also performed. The optimal phase of enhancement based on qualitative lesion conspicuity and quantitative tumor enhancement was then compared. RESULTS: MPM tumor enhancement was quantitatively and qualitatively increased at time delays beyond the conventional time delay for thoracic CT imaging (40-60 seconds). Patient tumor enhancement kinetics, displayed as the fraction of maximal tumor tissue attenuation as a function of time, revealed an optimal time delay of 230 to 300 seconds after intravenous contrast administration. There was an association between degree of tumor enhancement and subjective lesion conspicuity. CONCLUSION: Optimal MPM contrast enhancement occurs at a later phase than typically acquired with conventional thoracic CT imaging.

18F-Fluciclovine Uptake in Acute Appendicitis.

An 85-year-old man with biochemical recurrence of prostate cancer after prostatectomy was imaged with F-fluciclovine PET/CT. Images incidentally revealed F-fluciclovine uptake in a dilated appendix with associated fat stranding, suggestive of acute appendicitis. The patient was then questioned about abdominal symptoms, and he reported severe right lower quadrant pain. He then underwent laparoscopic appendectomy with pathology confirming acute appendicitis.

First-Year Radiology Residents Teaching Anatomy to First-Year Medical Students: A Symbiotic Relationship.

OBJECTIVES: Our institution has developed an educational program in which first-year radiology residents teach first-year medical students during gross anatomy laboratory sessions. The purpose of this study is to assess the impact of this program on medical student knowledge and perceptions of radiology, and on resident attitudes toward teaching. MATERIALS AND METHODS: First-year resident pairs taught small groups of medical students during weekly 15-minute interactive sessions, and were evaluated on teaching skills by senior residents. A survey about attitudes toward radiology and a knowledge quiz were sent to the medical students, and a survey about attitudes toward teaching was sent to the first-year radiology residents, both pre-course and post-course. RESULTS: Students' radiology knowledge significantly increased between the pre-course and post-course survey across all categories tested (P < 0.001). Additionally, there were significant improvements in terms of students' confidence in radiologic anatomy skills, perceived importance of radiology for medical training, familiarity with the field of radiology, and perception that radiologists are friendly (P < 0.001). Radiology residents felt more confident in their teaching proficiency (P < 0.001) by the conclusion of the course. CONCLUSIONS: Resident-led small-group teaching sessions during anatomy laboratory are mutually beneficial for medical students and radiology residents. The program also allows radiology residents to be exposed early on in residency to teaching and academic medicine.

The Role of Imaging in the Management of Suspected or Known COVID-19 Pneumonia: A Multidisciplinary Perspective.

COVID-19 is an illness caused by a novel coronavirus that has rapidly escalated into a global pandemic leading to an urgent medical effort to better characterize this disease biologically, clinically and by imaging. In this review, we present the current approach to imaging of COVID-19 pneumonia. We focus on the appropriate utilization of thoracic imaging modalities to guide clinical management. We will also describe radiologic findings that are considered typical, atypical and generally not compatible with of COVID-19 infection. Further, we review imaging examples of COVID-19 imaging mimics, such as organizing pneumonia, eosinophilic pneumonia and other viral infections.

The Role of Imaging in the Management of Suspected or Known COVID-19 Pneumonia. A Multidisciplinary Perspective.

Coronavirus disease (COVID-19) is an illness caused by a novel coronavirus that has rapidly escalated into a global pandemic leading to an urgent medical effort to better characterize this disease biologically, clinically, and by imaging. In this review, we present the current approach to imaging of COVID-19 pneumonia. We focus on the appropriate use of thoracic imaging modalities to guide clinical management. We also describe radiologic findings that are considered typical, atypical, and generally not compatible with COVID-19. Furthermore, we review imaging examples of COVID-19 imaging mimics, such as organizing pneumonia, eosinophilic pneumonia, and other viral infections.

Impact of Early Radiology Research Experiences on Medical Student Perceptions of Radiology and Research.

RATIONALE AND OBJECTIVES: To promote opportunities for medical students to gain early exposure to radiology and research, our institution has initiated programs which fund summer radiology research projects for rising second-year medical students. This study assesses the impact of these faculty-mentored summer research experiences on medical student perceptions of radiology and research, in terms of both knowledge and interest. MATERIALS AND METHODS: A voluntary, anonymous survey was administered to students both before and after the summer research period. Both the pre-program survey and post-program survey included 7-point Likert-scale questions (1 = strongly disagree; 7 = strongly agree) to evaluate students' perceptions about research and students' perceptions about radiology as a specialty. Faculty mentors were sent an analogous post-program survey that included an evaluation of their student's research skills. RESULTS: The surveys were completed by 9 of 11 students and 10 of 11 mentors. Students' perceived knowledge of radiology as a specialty improved (P = 0.02) between the pre-program survey and post-program survey. Similarly, there was an increase in students' perceived knowledge of research skills (P = 0.02) between the pre-program survey and post-program survey, with student ratings of research skills consistent with those of mentors. High student interest in both radiology and research was maintained over the course of the program. CONCLUSION: Our pilot study suggests that summer research experiences can improve knowledge of radiology and research among medical students. Continued evaluation of this annual program will allow us to enhance the benefit to medical students and thereby bolster interest in academic radiology.