RATIONALE AND OBJECTIVES:Intra-procedural patient motion is common in clinical MRI and degrades image quality. We evaluated whether certain clinical characteristics and patient sociodemographic factors are associated with patient motion during MRI. MATERIALS AND METHODS:We retrospectively reviewed consecutive MRI reports from 2022 at two U.S. urban health systems (university and safety net). Exams were acquired using standardized protocols and reported by a single academic radiology group. Motion was defined by any mention of "motion" in the report body or impression. Clinical and sociodemographic characteristics were extracted from the electronic health record. Multivariable logistic regression assessed associations between patient characteristics and motion, adjusting for care setting. RESULTS:Among 68,517 MRIs, emergency and inpatient studies had higher odds (OR) [95%CI] of motion compared with outpatient exams (emergency OR 2.55; [2.38,2.74]; inpatient OR 3.26; [3.06,3.48]). After adjustment, older age remained associated with motion, (≥85 years OR 1.91; [1.64,2.23], vs. 19-34 years). Male gender had higher odds than female (OR 1.14; [1.08,1.20]). Black race had higher odds than White (OR 1.18; [1.10,1.27]). Increasing obesity was associated with greater odds (class II OR 1.11; [1.01,1.22]; class III OR 1.48; [1.34,1.64]). Ethnicity, preferred language, and health system were not significantly associated with motion. CONCLUSION:Motion on MRI disproportionately affects elderly patients, men, individuals with obesity, and Black patients, independent of care setting. Accounting for motion-risk factors into scheduling, pre-scan counseling, and positioning protocols may reduce motion-limited studies. Addressing these patterns through workflow design is important for promoting equitable, high-quality MRI across diverse patient populations.

PURPOSE:The COVID-19 pandemic disrupted normal mechanisms of health care delivery and facilitated the rapid and widespread implementation of telehealth technology. As a result, the effectiveness of virtual health care visits in diverse populations represents an important consideration. We used lung cancer screening as a prototype to determine whether subsequent adherence differs between virtual and in-person encounters in an urban, safety-net health care system. METHODS:We conducted a retrospective analysis of initial low-dose computed tomography (LDCT) ordered for lung cancer screening from March 2020 through February 2023 within Parkland Health, the integrated safety-net provider for Dallas County, TX. We collected data on patient characteristics, visit type, and LDCT completion from the electronic medical record. Associations among these variables were assessed using the chi-square test. We also performed interaction analyses according to visit type. RESULTS:Initial LDCT orders were placed for a total of 1,887 patients, of whom 43% were female, 45% were Black, and 17% were Hispanic. Among these orders, 343 (18%) were placed during virtual health care visits. From March to August 2020, 79 of 163 (48%) LDCT orders were placed during virtual visits; after that time, 264 of 1,724 (15%) LDCT orders were placed during virtual visits. No patient characteristics were significantly associated with visit type (in-person v virtual) or LDCT completion. Rates of LDCT completion were 95% after in-person visits and 97% after virtual visits (P = .13). CONCLUSION:In a safety-net lung cancer screening population, patients were as likely to complete postvisit initial LDCT when ordered in a virtual encounter as in an in-person encounter.

The work relative value unit (wRVU) measures the physician's work involved in performing a service and is commonly used to quantify physician productivity. A critical component factored in wRVUs is the time required to perform a service. In musculoskeletal radiology, this time correlates directly with the number of images produced per radiograph. The purpose of this project was to evaluate whether the actual number of acquired images matches the number of views indicated in musculoskeletal radiographs CPT code descriptions.A query of our internal database returned 76,204 musculoskeletal radiograph reports. 440 random radiographs were reviewed to evaluate variability in the number of images obtained. This sample consisted of ten studies from each of the forty-four musculoskeletal codes. We recorded the number of actual images obtained. 242 studies from the safety net health care system and 198 studies from the university associated hospitals and clinics were evaluated.Seventy-five studies (31%) were found to have mismatched number of images among the 242 studies from the safety net health care system. Sixty-six studies (33%) were found to have mismatched number of images among the 198 studies sample from university associated tertiary care system. There was significant difference between the extra images obtained at two different health care systems (p<0.001). There were more studies with extra images in the safety net system compared to the university hospital.The commonly used wRVU metric has broad variability in the assessment of work productivity for musculoskeletal radiographs given the variance in the number of images obtained.

PURPOSE:First-line imaging for diagnosis of deep vein thrombosis (DVT) is ultrasound. Although the Wells criteria and laboratory testing help inform best ordering practices, this investigation evaluates the influence of various patient and nonpatient factors on examination results. METHODS:The authors analyzed structured reports from all inpatient and emergency department ultrasound examinations across two institutions from 2015 to 2022. Examination, provider, and patient factors were compared with examination results for effect on examination results and intergroup variance. RESULTS:The overall rate of acute or new noncalf DVT was 10.4%. Rates of acute or new noncalf DVT were found to be higher in inpatients (versus emergency department patients), in upper extremities (versus lower extremities), in patients with lower body mass index values, among physicians (versus advanced practice providers), and when examinations were ordered by providers with fewer overall DVT ultrasound examinations ordered. Mixed results were observed for number of limbs examined and provider supervision status. Examination day of week and time of day were not significant. CONCLUSIONS:DVT ultrasound results varied according to examination-, facility-, patient-, and provider-level factors, which can inform institutional quality monitoring programs, resource utilization, and future investigations into factors that may influence diagnostic testing results.

Background: Although low-dose, CT -based lung cancer screening (LCS) can decrease lung cancer mortality in high-risk individuals, the process may be complex and pose challenges to patients, particularly those from minority underinsured and uninsured populations. We conducted a randomized controlled trial of telephone-based navigation for LCS within an integrated, urban, safety-net health care system. Patients and Methods: Patients eligible for LCS were randomized (1:1) to usual care with or without navigation at Parkland Health in Dallas, Texas. The primary endpoint was completion of the first 3 consecutive steps in a patient 's LCS process. We explored differences in completion of LCS steps between navigation and usual care groups, controlling for patient characteristics using the chi-square test. Results: Patients (N = 447) were randomized to either navigation (n = 225) or usual care (n = 222). Mean patient age was 62 years, 46% were female, and 69% were racial/ethnic minorities. There was no difference in completion of the first 3 steps of the LCS algorithm between arms (12% vs 9%, respectively; P = .30). For ordered LCS steps, completion rates were higher among patients who received navigation (86% vs 79%; P = .03). The primary reason for step noncompletion was lack of order placement. Conclusions: In this study, lack of order placement was a key reason for incomplete LCS steps. When orders were placed, patients who received navigation had higher rates of completion. Clinical team education and enhanced electronic health record processes to simplify order placement, coupled with patient navigation, may improve LCS in safety-net health care systems.

Background: Recent modifications to low-dose CT (LDCT)-based lung cancer screening guidelines increase the number of eligible individuals, particularly among racial and ethnic minorities. Because these populations disproportionately live in metropolitan areas, we analyzed the association between travel time and initial LDCT completion within an integrated, urban safety-net health care system. Methods: Using Esri's StreetMap Premium, OpenStreetMap, and the r5r package in R, we determined projected private vehicle and public transportation travel times between patient residence and the screening facility for LDCT ordered in March 2017 through December 2022 at Parkland Memorial Hospital in Dallas, Texas. We characterized associations between travel time and LDCT completion in univariable and multivariable analyses. We tested these associations in a simulation of 10,000 permutations of private vehicle and public transportation distribution. Results: A total of 2,287 patients were included in the analysis, of whom 1,553 (68%) completed the initial ordered LDCT. Mean age was 63 years, and 73% were underrepresented minorities. Median travel time from patient residence to the LDCT screening facility was 17 minutes by private vehicle and 67 minutes by public transportation. There was a small difference in travel time to the LDCT screening facility by public transportation for patients who completed LDCT versus those who did not (67 vs 66 min, respectively; P =.04) but no difference in travel time by private vehicle for these patients (17 min for both; P=.67). In multivariable analysis, LDCT completion was not associated with projected travel time to the LDCT facility by private vehicle (odds ratio, 1.01; 95% CI, 0.82-1.25) or public transportation (odds ratio, 1.14; 95% CI, 0.89-1.44). Similar results were noted across travel-type permutations. Black individuals were 29% less likely to complete LDCT screening compared with White individuals. Conclusions: In an urban population comprising predominantly underrepresented minorities, projected travel time is not associated with initial LDCT completion in an integrated health care system. Other reasons for differences in LDCT completion warrant investigation.

10577 Background: Although low-dose computed tomography (LDCT)-based lung cancer screening (LCS) can decrease lung cancer mortality in high-risk individuals, it may be a complex and time-intensive process that poses challenges to patients, particularly those from minority, under- and uninsured populations. We conducted a pragmatic randomized controlled trial of telephone-based navigation for LCS within an integrated, urban safety-net healthcare system (Parkland Health, Dallas, Texas). Methods: Patients eligible for LCS based on United States Preventive Services Task Force 2013 Guidelines were randomized (1:1) to usual care with or without phone-based navigation. Following a structured protocol, navigators made systemic contact with patients to provide appointment reminders, share information and resources, assess barriers to LCS, and address smoking cessation. The primary endpoint was completion of the first three consecutive steps (eg, 3 annual LDCTs or LDCT, other imaging, biopsy) in a patient’s LCS process. We also explored differences in completion of LCS steps in navigation and usual care groups according to patient characteristics using chi-square test. Results: Patients (n=447) were randomized to phone-based navigation (n=225) or usual care (n=222) between February 2017 and February 2019. Mean patient age was 62 years, 46% were female, and 69% were racial-ethnic minorities. There was no significant difference in completion of the first three steps of the LCS algorithm (12 vs. 9%; P=0.3) between arms. Completion of LCS steps was not impacted by navigation among different subgroups, including age (<65 vs. ≥65 years), gender (male vs. female), race/ethnicity (non-hispanic white vs. racial minorities), and comorbidity status (Charlson Comorbidity Index <5 vs. ≥5). A key reason for step non-completion was lack of order placement. In the navigation arm, 368 of expected 675 steps (55%) were ordered, and in the usual care arm, 344 of expected 666 steps (52%) were ordered. Despite low three-step completion rates overall, the majority (>75%) of LCS algorithm steps were completed when ordered by their clinical team(Table). In exploratory univariable analysis, completion rates for ordered steps were 86% for navigation and 79% for usual care ( P=0.03). Conclusions: In this study, lack of order placement was a key reason for incomplete LCS steps. When orders were placed, navigated patients had higher rates of completion. Clinical team education and enhanced EHR processes to simplify order placement, coupled with patient navigation, may increase LCS uptake and completion in safety-net healthcare systems. Clinical trial information: NCT02758054 . [Table: see text]

In this article, we demonstrate the use of a software-based radiologist reporting tool for the implementation of American College of Radiology Thyroid Imaging, Reporting and Data System thyroid nodule risk-stratification. The technical details are described with emphasis on addressing the information security and patient privacy issues while allowing it to integrate with the electronic health record and radiology reporting dictation software. Its practical implementation is assessed in a quality improvement project in which guideline adherence and recommendation congruence were measured pre and post implementation. The descriptions of our solution and the release of the open-sourced codes may be helpful in future implementation of similar web-based calculators.

BACKGROUND. Artificial intelligence (AI) algorithms have shown strong performance for detection of pulmonary embolism (PE) on CT examinations performed using a dedicated protocol for PE detection. AI performance is less well studied for detecting PE on examinations ordered for reasons other than suspected PE (i.e., incidental PE [iPE]). OBJECTIVE. The purpose of this study was to assess the diagnostic performance of an AI algorithm for detection of iPE on conventional contrast-enhanced chest CT examinations. METHODS. This retrospective study included 2555 patients (mean age, 53.2 ± 14.5 [SD] years; 1340 women, 1215 men) who underwent 3003 conventional contrast-enhanced chest CT examinations (i.e., not using pulmonary CTA protocols) between September 2019 and February 2020. A commercial AI algorithm was applied to the images to detect acute iPE. A vendor-supplied natural language processing (NLP) algorithm was applied to the clinical reports to identify examinations interpreted as positive for iPE. For all examinations that were positive by the AI-based image review or by NLP-based report review, a multireader adjudication process was implemented to establish a reference standard for iPE. Images were also reviewed to identify explanations of AI misclassifications. RESULTS. On the basis of the adjudication process, the frequency of iPE was 1.3% (40/3003). AI detected four iPEs missed by clinical reports, and clinical reports detected seven iPEs missed by AI. AI, compared with clinical reports, exhibited significantly lower PPV (86.8% vs 97.3%, p = .03) and specificity (99.8% vs 100.0%, p = .045). Differences in sensitivity (82.5% vs 90.0%, p = .37) and NPV (99.8% vs 99.9%, p = .36) were not significant. For AI, neither sensitivity nor specificity varied significantly in association with age, sex, patient status, or cancer-related clinical scenario (all p > .05). Explanations of false-positives by AI included metastatic lymph nodes and pulmonary venous filling defect, and explanations of false-negatives by AI included surgically altered anatomy and small-caliber subsegmental vessels. CONCLUSION. AI had high NPV and moderate PPV for iPE detection, detecting some iPEs missed by radiologists. CLINICAL IMPACT. Potential applications of the AI tool include serving as a second reader to help detect additional iPEs or as a worklist triage tool to allow earlier iPE detection and intervention. Various explanations of AI misclassifications may provide targets for model improvement.

121 Background: As part of a CPRIT prevention services grant led by UTSW, Parkland Health adapted a patient navigation model previously piloted in 2017 to optimize low-dose computed tomography (LDCT) imaging for lung cancer screening and evidence-based tobacco cessation services. Methods: Non-clinical patient navigators (PN) contacted patients with a LDCT order to discuss their lung scan and smoking cessation options via telephone encounters. The PN has three planned touchpoints to assess patient knowledge, identify barriers, and refer to appropriate resources. Program evaluation data has been collected monthly for enrolled patients (n = 460) from August 2021-June 2022 via REDCap and EPIC. Navigation excluded patients who were incarcerated, had no phone number listed in medical record, or patients opted out of the program on intake. Results: Of the LDCT-eligible patients, PNs were able to complete two or more navigation calls with 55.62% of patients assigned. 77% of patients completed their first LDCT, compared to 80% in the initial trial. The most common barriers to screening completion included transportation (n = 38), insurance coverage (n = 22), and cost (n = 9). Of the navigated patients who were active smokers (75%), 33% scheduled at least one smoking cessation-related visit. Some patients unresponsive to conversations about quitting, were responsive to language around “reducing” tobacco use. Conclusions: Disproportionately under- and uninsured, Parkland Health patients face a variety of barriers to screening. As a result, PN staff must balance reaching all patients with a LDCT order and spending additional time navigating patients with complex barriers. To address these barriers, the PN program facilitates coordination of patients between primary care providers, radiology operations, smoking cessation clinics, patient financial services, lung diagnostic clinic, and social work. For example, patients can schedule a smoking cessation visit directly with our PN instead of being referred. As Parkland has a decentralized LCS program model, having embedded non-clinical PNs within existing roles in the cancer center, like medical practice assistants, increases continuity of care. These individuals have the benefit from day-to-day knowledge of downstream services that bolster their ability to connect patients to external services to address barriers.

Purpose To develop and evaluate domain-specific and pretrained bidirectional encoder representations from transformers (BERT) models in a transfer learning task on varying training dataset sizes to annotate a larger overall dataset. Materials and Methods The authors retrospectively reviewed 69 095 anonymized adult chest radiograph reports (reports dated April 2020-March 2021). From the overall cohort, 1004 reports were randomly selected and labeled for the presence or absence of each of the following devices: endotracheal tube (ETT), enterogastric tube (NGT, or Dobhoff tube), central venous catheter (CVC), and Swan-Ganz catheter (SGC). Pretrained transformer models (BERT, PubMedBERT, DistilBERT, RoBERTa, and DeBERTa) were trained, validated, and tested on 60%, 20%, and 20%, respectively, of these reports through fivefold cross-validation. Additional training involved varying dataset sizes with 5%, 10%, 15%, 20%, and 40% of the 1004 reports. The best-performing epochs were used to assess area under the receiver operating characteristic curve (AUC) and determine run time on the overall dataset. Results The highest average AUCs from fivefold cross-validation were 0.996 for ETT (RoBERTa), 0.994 for NGT (RoBERTa), 0.991 for CVC (PubMedBERT), and 0.98 for SGC (PubMedBERT). DeBERTa demonstrated the highest AUC for each support device trained on 5% of the training set. PubMedBERT showed a higher AUC with a decreasing training set size compared with BERT. Training and validation time was shortest for DistilBERT at 3 minutes 39 seconds on the annotated cohort. Conclusion Pretrained and domain-specific transformer models required small training datasets and short training times to create a highly accurate final model that expedites autonomous annotation of large datasets.Keywords: Informatics, Named Entity Recognition, Transfer Learning Supplemental material is available for this article. ©RSNA, 2022See also the commentary by Zech in this issue.

BackgroundLung cancer screening trials generally enroll motivated, relatively healthy, and adherent populations. We therefore evaluated the prevalence and effects of comorbidities in a real-world population undergoing low-dose computed tomography (LDCT) scans.Patients and MethodsWe calculated the Charlson Comorbidity Index (CCI) of patients for whom an initial low-dose computed tomography (LDCT) for lung cancer screening was ordered between February 2017 and February 2019 in an integrated safety-net healthcare system. We examined the association between CCI and initial LDCT completion using multivariable logistic regression, assessed the association between specific medical comorbidity and LDCT completion using Chi-square test or Fisher's exact test as appropriate, and examined the association between CCI and LDCT Lung-RADS results using Fisher's exact test.ResultsA total of 1358 patients were included in the analysis. Mean age was 63 years, 57% were women, and 50% were Black. Patients had moderate comorbidity burden (median CCI 3) with chronic pulmonary disease the most common comorbidity. Overall, 943 LDCT (70%) were completed. There was no difference in 30-day, 90-day, or 1-year completion rates of initial LDCT according to CCI. However, 30-day LDCT completion rates did increase over time (P < .001). Lung-RADS scores were not associated with CCI.ConclusionIn a real-world setting, patients undergoing lung cancer screening have moderate comorbidity burden. The degree and type of medical comorbidity are not associated with initial screening completion or results. Timeliness of LDCT completion may improve as program experience increases.

BACKGROUND. Postoperative prolonged mechanical ventilation is associated with increased morbidity and mortality. Reliable predictors of the need for postoperative mechanical ventilation after abdominal or pelvic surgeries are lacking. OBJECTIVE. The purpose of this study was to explore associations between preoperative thoracic CT findings and the need for postoperative mechanical ventilation after major abdominal or pelvic surgeries. METHODS. This retrospective case-control study included patients who underwent abdominal or pelvic surgeries during the period from January 1, 2014, through December 31, 2018, and had undergone preoperative thoracic CT. Case patients were patients who required postoperative mechanical ventilation. Control patients and case patients were matched at a 3:1 ratio on the basis of age, sex, body mass index, chronic obstructive pulmonary disease, smoking status, and surgery type. Two radiologists (readers 1 and 2) reviewed the CT images. Findings were compared between groups. RESULTS. The study included 165 patients (70 women, 95 men; mean age, 67.0 ± 9.7 [SD] years; 42 case patients and 123 matched control patients). Bronchial wall thickening and pericardial effusion were more frequent in case patients than control patients for reader 2 (10% vs 2%, p = .03; 17% vs 5%, p = .01) but not for reader 1. Pulmonary artery diameter (mean ± SD) was greater in case patients than control patients for reader 2 (2.9 ± 0.5 cm vs 2.8 ± 0.5 cm, p = .045) but not reader 1. Right lung height was lower in case patients than control patients for reader 1 (18.4 ± 2.9 cm vs 19.9 ± 2.7 cm, p = .01) and reader 2 (18.3 ± 2.9 cm vs 19.8 ± 2.7 cm, p = .01). Left lung height was lower in case patients than control patients for reader 1 (19.5 ± 3.1 cm vs 21.1 ± 2.6 cm, p = .01) and reader 2 (19.6 ± 2.4 cm vs 20.9 ± 2.6 cm, p = .01). Anteroposterior (AP) chest diameter was greater for case patients than control patients for reader 1 (14.0 ± 2.3 cm vs 12.9 ± 3.7 cm, p = .02) and reader 2 (14.2 ± 2.2 cm vs 13.2 ± 3.6 cm, p = .04). In a multivariable regression model using pooled reader data, bronchial wall thickening exhibited an odds ratio (OR) of 4.6 (95% CI, 1.3-16.5; p = .02); pericardial effusion, an OR of 5.1 (95% CI, 1.7-15.5; p = .004); pulmonary artery diameter, an OR of 1.4 per 1-cm increase (95% CI, 0.7-3.0; p = .32); mean lung height, an OR of 0.8 per 1-cm increase (95% CI, 0.7-1.001; p = .05); and AP chest diameter, an OR of 1.2 per 1-cm increase (95% CI, 1.013-1.4; p = .03). CONCLUSION. CT features are associated with the need for postoperative mechanical ventilation after abdominal or pelvic surgery. CLINICAL IMPACT. Many patients undergo thoracic CT before abdominal or pelvic surgery; the CT findings may complement preoperative clinical risk factors.

Introduction: ESWL is underutilised in ureteric stone management. The GIRFT report showed just four units nationally treated >10% of acute ureteric stones with ESWL. Despite guideline recommendations as a first-line treatment option, few large volume studies have been published. We present our experience of ESWL in 530 ureteric stone cases in the largest series we are aware of to date. Methods: Retrospective review of prospectively collected data between December 2012-February 2020 was performed. Data relating to patient demographics, stone characteristics, skin-to-stone distance, and treatment failure were collected. Cost analysis was conducted by the hospital trust's Head of Finance. Chi-squared analysis for statistical significance was performed. Results: A success rate of 67.9% with a mean treatment number of 1.7 sessions was observed (n=530). Statistically significant outcomes were observed for stone size (p=0.0001), stone density (Hounsfield units) (p=0.006) and skin-to-stone distance (p=0.03). Stone position was not statistically significant (p=0.54). However, the small number of stones treated >13mm or >1250HU had an approximate 50% chance of successful treatment. In our practice acute ureteric ESWL was found to be less costly than acute ureterorenoscopy, consistent with findings from previous NHS studies. Conclusion: Acute ESWL is a safe, reliable, and financially viable treatment option for a wider spectrum of patients than reflected in international guidelines based on our large, heterogenous series. In the COVID-19 era, with theatre access reduced and concerns over aerosol generating procedures, acute ESWL remains an attractive first-line treatment option.

BACKGROUND. When performing ultrasound (US) for hepatocellular carcinoma (HCC) screening, numerous factors may impair hepatic visualization, potentially lowering sensitivity. US LI-RADS includes a visualization score as a technical adequacy measure. OBJECTIVE. The purpose of this article is to identify associations between examination, sonographer, and radiologist factors and the visualization score in liver US HCC screening. METHODS. This retrospective study included 6598 patients (3979 men, 2619 women; mean age, 58 years) at risk for HCC who underwent a total of 10,589 liver US examinations performed by 91 sonographers and interpreted by 50 radiologists. Visualization scores (A, no or minimal limitations; B, moderate limitations; C, severe limitations) were extracted from clinical reports. Patient location (emergency department [ED], inpatient, outpatient), sonographer and radiologist liver US volumes during the study period (< 50, 50-500, > 500 examinations), and radiologist practice pattern (US, abdominal, community, interventional) were recorded. Associations with visualization scores were explored. RESULTS. Frequencies of visualization scores were 71.5%, 24.2%, and 4.2% for A, B, and C, respectively. Scores varied significantly (p <.001) between examinations performed in ED patients (49.8%, 40.1%, and 10.2%), inpatients (58.8%, 33.9%, and 7.3%), and outpatients (76.7%, 20.3%, and 2.9%). Scores also varied significantly (p <.001) by sonographer volume (< 50 examinations: 58.4%, 33.7%, and 7.9%; > 500 examinations: 72.9%, 22.5%, and 4.6%); reader volume (< 50 examinations: 62.9%, 29.9%, and 7.1%; > 500 examinations: 67.3%, 28.0%, and 4.7%); and reader practice pattern (US: 74.5%, 21.3%, and 4.3%; abdominal: 67.0%, 28.1%, and 4.8%; community: 75.2%, 21.9%, and 2.9%; interventional: 68.5%, 24.1%, and 7.4%). In multivariable analysis, independent predictors of score C were patient location (ED/inpatient: odds ratio [OR], 2.62; p <.001) and sonographer volume (< 50: OR, 1.55; p =.01). Among sonographers performing 50 or more examinations, the percentage of outpatient examinations with score C ranged from 0.8% to 5.4%; 9/33 were above the upper 95% CI of 3.2%. CONCLUSION. The US LI-RADS visualization score may identify factors affecting quality of HCC screening examinations and identify outlier sonographers in terms of poor examination quality. The approach also highlights potential systematic biases among radiologists in their quality assessment process. CLINICAL IMPACT. These findings may be applied to guide targeted quality improvement efforts and establish best practices and performance standards for screening programs.