Junior Radiologists Research Articles

Deep learning-based classification of primary bone tumors on radiographs: A preliminary study

BackgroundTo develop a deep learning model to classify primary bone tumors from preoperative radiographs and compare performance with radiologists.MethodsA total of 1356 patients (2899 images) with histologically confirmed primary bone tumors and pre-operative radiographs were identified from five institutions’ pathology databases. Manual cropping was performed by radiologists to label the lesions. Binary discriminatory capacity (benign versus not-benign and malignant versus not-malignant) and three-way classification (benign versus intermediate versus malignant) performance of our model were evaluated. The generalizability of our model was investigated on data from external test set. Final model performance was compared with interpretation from five radiologists of varying level of experience using the Permutations tests.FindingsFor benign vs. not benign, model achieved area under curve (AUC) of 0•894 and 0•877 on cross-validation and external testing, respectively. For malignant vs. not malignant, model achieved AUC of 0•907 and 0•916 on cross-validation and external testing, respectively. For three-way classification, model achieved 72•1% accuracy vs. 74•6% and 72•1% for the two subspecialists on cross-validation (p = 0•03 and p = 0•52, respectively). On external testing, model achieved 73•4% accuracy vs. 69•3%, 73•4%, 73•1%, 67•9%, and 63•4% for the two subspecialists and three junior radiologists (p = 0•14, p = 0•89, p = 0•93, p = 0•02, p < 0•01 for radiologists 1–5, respectively).InterpretationDeep learning can classify primary bone tumors using conventional radiographs in a multi-institutional dataset with similar accuracy compared to subspecialists, and better performance than junior radiologists.FundingThe project described was supported by RSNA Research & Education Foundation, through grant number RSCH2004 to Harrison X. Bai.

Radiomics-Assisted Presurgical Prediction for Surgical Portal Vein-Superior Mesenteric Vein Invasion in Pancreatic Ductal Adenocarcinoma.

ObjectivesTo develop a radiomics signature for predicting surgical portal vein-superior mesenteric vein (PV-SMV) in patients with pancreatic ductal adenocarcinoma (PDAC) and measure the effect of providing the predictions of radiomics signature to radiologists with different diagnostic experiences during imaging interpretation.MethodsBetween February 2008 and June 2020, 146 patients with PDAC in pancreatic head or uncinate process from two institutions were retrospectively included and randomly split into a training (n = 88) and a validation (n =58) cohort. Intraoperative vascular exploration findings were used to identify surgical PV-SMV invasion. Radiomics features were extracted from the portal venous phase CT images. Radiomics signature was built with a linear elastic-net regression model. Area under receiver operating characteristic curve (AUC) of the radiomics signature was calculated. A senior and a junior radiologist independently review CT scans and made the diagnosis for PV-SMV invasion both with and without radiomics score (Radscore) assistance. A 2-sided Pearson’s chi-squared test was conducted to evaluate whether there was a difference in sensitivity, specificity, and accuracy between the radiomics signature and the unassisted radiologists. To assess the incremental value of providing Radscore predictions to the radiologists, we compared the performance between unassisted evaluation and Radscore-assisted evaluation by using the McNemar test.ResultsNumbers of patients identified as presence of surgical PV-SMV invasion were 33 (37.5%) and 19 (32.8%) in the training and validation cohort, respectively. The radiomics signature achieved an AUC of 0.848 (95% confidence interval, 0.724–0.971) in the validation cohort and had a comparable sensitivity, specificity, and accuracy as the senior radiologist in predicting PV-SMV invasion (all p-values > 0.05). Providing predictions of radiomics signature increased both radiologists’ sensitivity in identifying PV-SMV invasion, while only the increase of the junior radiologist was significant (63.2 vs 89.5%, p-value = 0.025) instead of the senior radiologist (73.7 vs 89.5%, p-value = 0.08). Both radiologists’ accuracy had no significant increase when provided radiomics signature assistance (both p-values > 0.05).ConclusionsThe radiomics signature can predict surgical PV-SMV invasion in patients with PDAC and may have incremental value to the diagnostic performance of radiologists during imaging interpretation.

Improving the Diagnostic Accuracy of Breast BI-RADS 4 Microcalcification-Only Lesions Using Contrast-Enhanced Mammography

BackgroundContrast-enhanced mammography (CEM) is a novel breast imaging technique that can provide additional information of breast tissue blood supply. This study aimed to test the possibility of CEM in improving the diagnostic accuracy of Breast Imaging Reporting and Data System (BI-RADS) 4 calcification-only lesions with consideration of morphology and distribution. Patients and MethodsData of patients with suspicious malignant calcification-only lesions (BI-RADS 4) on low-energy CEM and proved pathologic diagnoses were retrospectively collected. Two junior radiologists independently reviewed the two sets of CEM images, low-energy images (LE) to describe the calcifications by morphology and distribution type, and recombined images (CE) to record the presence of enhancement. Low-risk and high-risk groups were divided by calcification morphology, distribution, and both, respectively. Positive predictive values and misdiagnosis rates (MDR) were compared between LE-only reading and CE reading. Diagnostic performance was also tested using machine learning method. ResultsThe study included 74 lesions (26 malignant and 48 benign). Positive predictive values were significantly higher and MDRs were significantly lower using CE images than using LE alone for both the low-risk morphology type and low-risk distribution type (P < .05). MDRs were significantly lower when using CE images (18.18%-24.00%) than using LE images alone in low-risk group (76.36%-80.00%) (P < .05). Using a machine learning method, significant improvements in the area under the receiver operating characteristic curve were observed in both low-risk and high-risk groups. ConclusionCEM has the potential to aid in the diagnosis of BI-RADS 4 calcification-only lesions; in particular, those presented as low risk in morphology and/or distribution may benefit more.

Diagnostic Performance of Ultrasound Computer-Aided Diagnosis Software Compared with That of Radiologists with Different Levels of Expertise for Thyroid Malignancy: A Multicenter Prospective Study

The aim of the work described here was to evaluate the diagnostic performance of ultrasound thyroid computer-aided diagnosis (CAD) software. This multicenter prospective study included 494 patients (565 thyroid nodules) who underwent surgery or biopsy after ultrasonography at four hospitals from January 2019 to September 2019. The diagnostic performance metrics of different readers were calculated and compared with the pathologic results. The sensitivity of CAD was outstanding and was equivalent to that of a senior radiologist (90.51% vs. 88.47%, p > 0.05). The area under the curve of CAD was equivalent to that of a junior radiologist (0.748 vs. 0.739, p > 0.05). However, the specificity was only 49.63%, which was lower than those of the three radiologists (75.56%, 85.93% and 90.37% for the junior, intermediate and senior radiologists, respectively). The diagnostic performance of the junior radiologist was significantly improved with the aid of CAD (junior + CAD). The sensitivity and area under the curve of junior + CAD were improved from 72.20% to 89.93% and from 0.739 to 0.816, respectively (both p values <0.05), and the positive predictive value, negative predictive value and κ coefficient improved from 76.3% to 78.6%, 82.0% to 86.8% and 0.394 to 0.511, respectively. Though specificity slightly decreased from 75.56% to 73.33%, the difference was not statistically significant (p > 0.05). In general, the clinical application value of CAD is promising, and its instrumental value for junior radiologists is significant.

Evaluation of a convolutional neural network for ovarian tumor differentiation based on magnetic resonance imaging.

There currently lacks a noninvasive and accurate method to distinguish benign and malignant ovarian lesion prior to treatment. This study developed a deep learning algorithm that distinguishes benign from malignant ovarian lesion by applying a convolutional neural network on routine MR imaging. Five hundred forty-five lesions (379 benign and 166 malignant) from 451 patients from a single institution were divided into training, validation, and testing set in a 7:2:1 ratio. Model performance was compared with four junior and three senior radiologists on the test set. Compared with junior radiologists averaged, the final ensemble model combining MR imaging and clinical variables had a higher test accuracy (0.87 vs 0.64, p < 0.001) and specificity (0.92 vs 0.64, p < 0.001) with comparable sensitivity (0.75 vs 0.63, p = 0.407). Against the senior radiologists averaged, the final ensemble model also had a higher test accuracy (0.87 vs 0.74, p = 0.033) and specificity (0.92 vs 0.70, p < 0.001) with comparable sensitivity (0.75 vs 0.83, p = 0.557). Assisted by the model's probabilities, the junior radiologists achieved a higher average test accuracy (0.77 vs 0.64, Δ = 0.13, p < 0.001) and specificity (0.81 vs 0.64, Δ = 0.17, p < 0.001) with unchanged sensitivity (0.69 vs 0.63, Δ = 0.06, p = 0.302). With the AI probabilities, the junior radiologists had higher specificity (0.81 vs 0.70, Δ = 0.11, p = 0.005) but similar accuracy (0.77 vs 0.74, Δ = 0.03, p = 0.409) and sensitivity (0.69 vs 0.83, Δ = -0.146, p = 0.097) when compared with the senior radiologists. These results demonstrate that artificial intelligence based on deep learning can assist radiologists in assessing the nature of ovarian lesions and improve their performance. • Artificial Intelligence based on deep learning can assess the nature of ovarian lesions on routine MRI with higher accuracy and specificity than radiologists. • Assisted by the deep learning model's probabilities, junior radiologists achieved better performance that matched those of senior radiologists.

Transitioning from trainee to breast radiologist: A guide for a successful first year

The transition from trainee to newly minted breast radiologist is exciting and daunting in equal measure. The early years in practice are pivotal to long-term success in breast imaging whether entering academic or nonacademic practice. Yet a paucity of literature exists to guide junior radiologists in their early career transition. New breast radiologists can successfully navigate the start of a prosperous and enriching career by implementing strategies adapted from the business world and collective wisdom from the radiology world. This article provides an outline of tips and habits for new radiologists to incorporate in their work lives as attendings to ensure that they will thrive in breast imaging for years to come.

MRI-based Bosniak Classification of Cystic Renal Masses, Version 2019: Interobserver Agreement, Impact of Readers' Experience, and Diagnostic Performance.

Background The 2019 Bosniak classification (version 2019) of cystic renal masses (CRMs) provides a systematic update to the currently used 2005 Bosniak classification (version 2005). Further validation is required before widespread application. Purpose To evaluate the interobserver agreement of MRI criteria, the impact of readers' experience, and the diagnostic performance between version 2019 and version 2005. Materials and Methods From January 2009 to December 2018, consecutive patients with CRM who had undergone renal MRI and surgical-pathologic examination were included in this retrospective study. On the basis of version 2019 and version 2005, all CRMs were independently classified by eight radiologists with different levels of experience. By using multirater κ statistics, interobserver agreement was evaluated with comparisons between classifications and between senior and junior radiologists. Diagnostic performance between classifications by dichotomizing classes I-IV into lower (I-IIF) and higher (III-IV) classes was compared by using the McNemar test. P < .05 was considered to indicate a statistically significant difference. Results A total of 207 patients (mean age ± standard deviation, 49 years ± 12; 139 male and 68 female patients) with CRMs were included. Overall, interobserver agreement was higher with version 2019 than version 2005 (weighted κ = 0.64 vs 0.50, respectively; P < .001). Interobserver agreement between senior and junior radiologists did not differ between version 2019 (weighted κ = 0.65 vs 0.64, respectively; P = .71) and version 2005 (weighted κ = 0.54 vs 0.46; P < .001). Diagnostic specificity for malignancy was higher with version 2019 than with version 2005 (83% [92 of 111] vs 68% [75 of 111], respectively; P < .001), without any difference in sensitivity (89% [85 of 96] vs 84% [81 of 96]; P = .34). Conclusion In the updated Bosniak classification, interobserver agreement improved and was unaffected by observers' experience. The diagnostic performance with version 2019 was superior to that with version 2005, with higher specificity. Published under a CC BY 4.0 license. Online supplemental material is available for this article. See also the editorial by Choyke in this issue.

Comparison of PI-RADS version 2.1 and PI-RADS version 2 regarding interreader variability and diagnostic accuracy for transition zone prostate cancer.

To compare the diagnostic performance of PI-RADS version 2.1 (PI-RADS v2.1) and PI-RADS v2 for transition zone prostate cancer (TZPC), and analyse its performance for readers with different experience levels. Eighty-five patients with suspected prostate cancer who underwent biopsy after MRI scan between January and December 2017 were retrospectively enrolled. One junior radiologist (reader 1, 1year of experience in using PI-RADS v2) and one senior radiologist (reader 2, 6years of experience) independently reviewed and assigned a score for each lesion according to PI-RADS v2.1 and v2. The template-guided transperineal prostate biopsy was used for standard of reference. To compare the diagnostic performance of the two methods, the AUC was calculated. The sensitivity, specificity, and accuracy were calculated at predefined positive values (PI-RADS ≥ 3). The interreader agreement and frequency of prostate cancer for each PI-RADS category were also calculated. Among the 85 patients, 27 had prostate cancers, and 25 were clinically significant prostate cancer (csPCa). The AUC values for diagnosing clinically significant prostate cancer significantly increased with PI-RADS v2.1 for reader 2 (0.766 vs. 0.902, P = 0.009). The specificity and accuracy for both readers also increased with PI-RADS v2.1 (specificity: reader 1, 41.7% vs. 78.3% and reader 2, 33.3% vs. 81.7%; accuracy: reader 1, 52.9% vs. 76.5% and reader 2, 48.2% vs. 83.5%, all P < 0.05). The interreader agreement was good for both versions. The percentage of prostate cancer decreased in lower PI-RADS categories (PI-RADS 2) and increased in higher PI-RADS categories (PI-RADS 3 ~ 4). Compared with PI-RADS v2, PI-RADS v2.1 may improve radiologists' diagnostic performance for TZPC.

A Clinical Assessment of an Ultrasound Computer-Aided Diagnosis System in Differentiating Thyroid Nodules With Radiologists of Different Diagnostic Experience.

IntroductionThis study aimed to assess the diagnostic performance and the added value to radiologists of different levels of a computer-aided diagnosis (CAD) system for the detection of thyroid cancers.Methods303 patients who underwent thyroidectomy from October 2018 to July 2019 were retrospectively reviewed. The diagnostic performance of the senior radiologist, the junior radiologist, and the CAD system were compared. The added value of the CAD system was assessed and subgroup analyses were performed according to the size of thyroid nodules.ResultsIn total, 186 malignant thyroid nodules, and 179 benign thyroid nodules were included; 168 were papillary thyroid carcinoma (PTC), 7 were medullary thyroid carcinoma (MTC), 11 were follicular carcinoma (FTC), 127 were follicular adenoma (FA) and 52 were nodular goiters. The CAD system showed a comparable specificity as the senior radiologist (86.0% vs. 86.0%, p > 0.99), but a lower sensitivity and a lower area under the receiver operating characteristic (AUROC) curve (sensitivity: 71.5% vs. 95.2%, p < 0.001; AUROC: 0.788 vs. 0.906, p < 0.001). The CAD system improved the diagnostic sensitivities of both the senior and the junior radiologists (97.8% vs. 95.2%, p = 0.063; 88.2% vs. 75.3%, p < 0.001).ConclusionThe use of the CAD system using artificial intelligence is a potential tool to distinguish malignant thyroid nodules and is preferable to serve as a second opinion for less experienced radiologists to improve their diagnosis performance.

Performance of morphologic criteria for the diagnosis of cirrhosis in patients with non-alcoholic steatohepatitis compared to other etiologies of chronic liver disease: effect of level of training and experience.

To compare the diagnostic performance of morphologic criteria for detection of cirrhosis in patients with alcoholic liver disease (ALD), hepatitis C (HCV), and non-alcoholic steatohepatitis (NASH). One hundred patients (53 male) with different etiologies of chronic liver disease (NASH, n=41; HCV, n=39; and ALD, n=20) and with different degrees of fibrosis on histopathologic examination (74 with cirrhosis) were retrospectively evaluated. Four readers (R1: fellowship-trained abdominal radiologist, R2: community attending radiologist, R3: senior radiology resident/research fellow, R4: junior radiology resident) analyzed the contrast-enhanced CTs for presence of commonly accepted morphologic changes of cirrhosis and portal hypertension. Each reader assigned an overall score (using a 5-point scale) for possibility of cirrhosis based on liver morphology and features of portal hypertension. Diagnostic performance, sensitivity, and specificity for the diagnosis of cirrhosis were calculated and compared between different etiologies of chronic liver disease. Performance of readers was affected by their level of training. Less experienced readers had overall lower sensitivity for diagnosis of cirrhosis in NASH group (R3: 81.5%, R4: 63.0% compared to 96.3% for both R1 and R2). Sensitivities for detection of NASH cirrhosis significantly decreased for less experienced readers in the absence of ascites (R3: 75.0%, R4: 62.0%) or other features of portal hypertension (R3: 50.0%; R4: 0%). The specificity was consistently high among different etiologies and for all readers (85.7-100%). Inter-reader agreement for morphologic features ranged widely from fair to almost perfect (k: 0.23-0.85). Cirrhotic changes in NASH are subtler and can be underestimated by less experienced readers.

Observer agreement and clinical significance of chest CT reporting in patients suspected of COVID-19.

ObjectivesTo assess interobserver agreement and clinical significance of chest CT reporting in patients suspected of COVID-19.MethodsFrom 16 to 24 March 2020, 241 consecutive patients addressed to hospital for COVID-19 suspicion had both chest CT and SARS-CoV-2 RT-PCR. Eight observers (2 thoracic and 2 general senior radiologists, 2 junior radiologists, and 2 emergency physicians) retrospectively categorized each CT into one out of 4 categories (evocative, compatible for COVID-19 pneumonia, not evocative, and normal). Observer agreement for categorization between all readers and pairs of readers with similar experience was evaluated with the Kappa coefficient. The results of a consensus categorization were correlated to RT-PCR.ResultsObserver agreement across the 4 categories was good between all readers (κ value 0.61 95% CI 0.60–0.63) and moderate to good between pairs of readers (0.54–0.75). It was very good (κ 0.81 95% CI 0.79–0.83), fair (κ 0.32 95% CI 0.29–0.34), moderate (κ 0.56 95% CI 0.54–0.58), and moderate (0.58 95% CI 0.56–0.61) for the categories evocative, compatible, not evocative, and normal, respectively. RT-PCR was positive in 97%, 50%, 31%, and 11% of cases in the respective categories. Observer agreement was lower (p < 0.001) and RT-PCR positive cases less frequently categorized evocative in the presence of an underlying pulmonary disease (p < 0.001).ConclusionInterobserver agreement for chest CT reporting using categorization of findings is good in patients suspected of COVID-19. Among patients considered for hospitalization in an epidemic context, CT categorized evocative is highly predictive of COVID-19, whereas the predictive value of CT decreases between the categories compatible and not evocative.Key Points • In patients suspected of COVID-19, interobserver agreement for chest CT reporting into categories is good, and very good to categorize CT “evocative.” • Chest CT can participate in estimating the likelihood of COVID-19 in patients presenting to hospital during the outbreak, CT categorized “evocative” being highly predictive of the disease whereas almost a third of patients with CT “not evocative” had a positive RT-PCR in our study. • Observer agreement is lower and CTs of positive RT-PCR cases less frequently “evocative” in presence of an underlying pulmonary disease.

Comparison of breast density assessment between human eye and automated software on digital and synthetic mammography: Impact on breast cancer risk

PurposeTo evaluate the agreement between automatic assessment software of breast density based on artificial intelligence (AI) and visual assessment by a senior and a junior radiologist, as well as the impact on the assessment of breast cancer risk (BCR) at 5 years. Materials and methodsWe retrospectively included 311 consecutive women (mean age, 55.6±8.5 [SD]; range: 40–74 years) without a personal history of breast cancer who underwent routine mammography between January 1, 2019 and February 28, 2019. Mammographic breast density (MBD) was independently evaluated by a junior and a senior reader on digital mammography (DM) and synthetic mammography (SM) using BI-RADS (5th edition) and by an AI software. For each MBD, BCR at 5 years was estimated per woman by the AI software. Interobserver agreement for MBD between the two readers and the AI software were evaluated by quadratic κ coefficients. Reproducibility of BCR was assessed by intraclass correlation coefficient (ICC). ResultsAgreement for MBD assessment on DM and SM was almost perfect between senior and junior radiologists (κ=0.88 [95% CI: 0.84–0.92] and κ=0.86 [95% CI: 0.82–0.90], respectively) and substantial between the senior radiologist and AI (κ=0.79; 95% CI: 0.73–0.84). There was substantial agreement between DM and SM for the senior radiologist (κ=0.79; 95% CI: 0.74–0.84). BCR evaluation at 5 years was highly reproducible between the two radiologists on DM and SM (ICC=0.98 [95% CI: 0.97–0.98] for both), between BCR evaluation based on DM and SM evaluated by the senior (ICC=0.96; 95% CI: 0.95–0.97) or junior radiologist (ICC=0.97; 95% CI: 0.96–0.98) and between the senior radiologist and AI (ICC=0.96; 95% CI: 0.95–0.97). ConclusionThis preliminary study demonstrates a very good agreement for BCR evaluation based on the evaluation of MBD by a senior radiologist, junior radiologist and AI software.

Impact of Simulation Training on Radiology Resident Performance in Neonatal Head Ultrasound

The aim of this study was to determine whether resident performance in head ultrasound on neonates improves following brain phantom simulation training. Ten junior radiology residents with at least one year of radiology training were divided into two equal groups. Both groups received a detailed head ultrasound protocol sheet and observed a technologist perform a head ultrasound on a neonatal patient at the beginning of their first pediatric radiology rotation. Both groups of residents also received teaching with a brain phantom model. Group A residents independently performed one head ultrasound exam, subsequently received phantom simulation training, and then performed a post-training head ultrasound exam. Group B residents received phantom simulation training prior to their first head ultrasound exam. Three pediatric radiologists independently and blindly reviewed the ultrasound images of each head ultrasound exam for proficiency of image acquisition using a validated scoring system. Scores of Group A residents prior to phantom training were compared to their scores after phantom training as well as to scores of Group B residents using simple linear regression. There was a statistically significant improvement in the performance of head ultrasound on neonates when comparing the same residents pre- and postphantom training (p = 0.003). Residents who initially trained with the phantom performed significantly better on their first head ultrasound examination on a neonate than those residents who did not (p = 0.005). Our novel head ultrasound phantom training model significantly improves radiology resident performance of head ultrasound on neonates and may, therefore, be beneficial for residency education.

Estimation of malignancy of pulmonary nodules at CT scans: Effect of computer-aided diagnosis on diagnostic performance of radiologists.

To develop a computer-aided diagnosis (CAD) system for distinguishing malignant from benign pulmonary nodules on computed tomography (CT) scans, and to assess whether the diagnostic performance of radiologists with different experiences can be improved with the assistant of CAD. A total of 857 malignant nodules from 601 patients and 426 benign nodules from 278 patients were retrospectively collected from four hospitals. In this study, we exploited convolutional neural network in the framework of deep learning to classify whether a nodule was benign or malignant. A total of 745 malignant nodules and 370 benign nodules were used as the training data of our CAD system. The remaining 112 malignant nodules and 56 benign nodules were used as the test data. The participants were two senior chest radiologists, two secondary chest radiologists, and two junior radiology residents. The readers estimated the likelihood of malignancy of pulmonary nodules first without and then with CAD output. Receiver-operating characteristic (ROC) curve was used to evaluate readers' diagnostic performance. When a threshold level of 58% was used to estimate the likelihood of malignancy, the sensitivity, specificity, and diagnostic accuracy values of our CAD scheme alone were 93.8%, 83.9%, and 90.5%, respectively. For all six readers, the mean area under the ROC curve (Az ) values without and with CAD system were 0.913 and 0.938, respectively. For each reader, there is a large difference in Az values that assessed without and with CAD system. With CAD output, the readers made correct changes an average of 15.7 times and incorrect changes an average of 2 times. Our CAD system significantly improved the diagnostic performance of readers regardless of their experience levels for assessment of the likelihood of malignancy of pulmonary nodules.

Diagnostic Accuracy of North America Expert Consensus Statement on Reporting CT Findings in Patients Suspected of Having COVID-19 Infection: An Italian Single-Center Experience.

PurposeTo evaluate the diagnostic accuracy of the four standardized categories for CT reporting proposed by the Radiological Society of North America (RSNA) to support a faster triage compared with real-time reverse-transcription polymerase chain reaction (RT-PCR), which is the reference standard for suspected coronavirus disease 2019 (COVID-19), but has long reporting time (6–48 hours).Materials and MethodsA retrospective analysis of 569 thin-section CT examinations performed for patients suspected of having COVID-19 from February 27 to March 27, 2020 (peak of infection in Italy) was conducted. The imaging pattern was classified according to the statement by the RSNA as “typical,” “indeterminate,” “atypical,” and “negative” and compared with RT-PCR for 460 patients. Interobserver variability in reporting between a senior and a junior radiologist was evaluated. Use of the vascular enlargement sign in indeterminate cases was also assessed.ResultsThe diagnosis of COVID-19 was made in 45.9% (211/460) of patients. The “typical” pattern (n = 172) showed a sensitivity of 71.6%, a specificity of 91.6%, and a positive predictive value of 87.8% for COVID-19. The “atypical” (n = 67) and “negative” (n = 123) pattern demonstrated a positive predictive value of 89.6% and 86.2% for non–COVID-19, respectively. The “indeterminate” (n = 98) pattern was nonspecific, but vascular enlargement was most frequently found in patients with COVID-19 (86.1%; P < .001). Interobserver agreement was good for the “typical” and “negative” pattern and fair for “indeterminate” and “atypical” (κ = 0.5; P = .002).ConclusionIn an epidemic setting, the application of the four categories proposed by the RSNA provides a standardized diagnostic hypothesis, strongly linked to the RT-PCR results for the “typical,” “atypical,” and “negative” pattern. In the “indeterminate” pattern, the analysis of the vascular enlargement sign could facilitate the interpretation of imaging features.© RSNA, 2020

Effectiveness of a 12-Month Emergency Radiology Curriculum for Improving Self-Confidence and Competence of Postgraduate Year 1 Radiology Residents-A Canadian Study.

Limited radiology curriculum during postgraduate year 1 of radiology residency combined with increasing workloads during emergency radiology call have contributed to heightened anxiety and responsibilities for junior residents. This study aimed to evaluate the effectiveness of a 12-month emergency radiology curriculum on self-rated confidence and general competence of Canadian postgraduate year 1 radiology residents. A cohort of Canadian postgraduate year 1 Diagnostic Radiology residents voluntarily enrolled in a 12-month self-directed online emergency radiology curriculum (9 modules). Participants completed pretest and posttest surveys and examinations to gauge their self-rated competence on module material and knowledge acquisition, respectively. Average pretest and posttest scores were compared using Student 2-tailed unpaired t test, and Likert data from self-reported confidence were compared using a Mann Whitney U test. Statistical significance was defined as P < .05. Sixty-six trainees completed at least 1 module, and 15 trainees completed all 9 modules. Both self-rated confidence and posttest scores were statistically higher after module completion (P < .001) for all 9 learning modules. The greatest improvement in test scores was seen in the female genitourinary module (28.12 ± 3.018; difference between pretest and posttest means ± SEM). Our study demonstrates learning benefits for junior radiology trainees who participated in a self-directed online emergency radiology curriculum during postgraduate year 1. In the face of ever-increasing demands for imaging in on-call settings across Canada, inclusion of a self-directed online curriculum may become more important for upcoming competency-based medical education as it encourages a learner-driven and non-time-based method of education.

Cystatin C and Sarcopenia Predict Acute on Chronic Liver Failure Development and Mortality in Patients on the Liver Transplant Waiting List.

Cystatin C (CysC) is an early biomarker of renal dysfunction scarcely studied in patients awaiting liver transplantation (LT). Sarcopenia is frequent in cirrhosis and impacts prognosis. We aimed to assess the capability of these factors to predict survival and acute-on-chronic liver failure (ACLF) in patients awaiting LT, as well as early post-LT outcomes. Single-center study that included all cirrhotic patients listed for LT between 2014 and 2017. Competing risk regression analysis was used to evaluate the capability of liver-, kidney-, and global status-related variables at waitlist (WL) inclusion to predict WL mortality and ACLF. Variables associated with post-LT outcomes were evaluated with logistic regression analysis. One-hundred-and-eighty patients were included. Fifty-six (31%) patients developed ACLF, 54 (30%) underwent LT and 35 (19%) died. In the adjusted competing risk regression analysis, CysC ≥ 1.5 mg/L, sarcopenia and MELD-Na were independent predictors of ACLF in the WL, while CysC ≥ 1.5 mg/L, sarcopenia and albumin were independent predictors of mortality. The cumulative incidence of ACLF and mortality at 12 months were 50% and 34% in patients with sarcopenia and CysC ≥1.5 mg/L. An estimated glomerular filtration rate by chronic kidney disease (CKD)-EPI-CysC-creatinine <60 mL/min/1.73 m at WL inclusion was an independent predictor of the need for renal replacement therapy (RRT) in the first month post-LT. Higher levels of CysC and sarcopenia are strongly associated with the ACLF and mortality in WL. The assessment of both risk factors may improve the prognostic evaluation and allow identifying a group of patients with a very high risk of poor outcomes while awaiting LT.

Malignancy risk stratification and FNA recommendations for thyroid nodules: A comparison of ACR TI-RADS, AACE/ACE/AME and ATA guidelines

To compare diagnostic performance and malignancy risk stratification among guidelines set forth by the American Thyroid Association (ATA) in 2015, the American Association of Clinical Endocrinologists (AACE), the American College of Endocrinology (ACE) and the Association Medici Endocrinologi (AME) in 2016, and the American College of Radiology (ACR) in 2017. The retrospective study was approved by the hospital ethics committee, and the informed consent requirement was waived. From October 2015 to March 2016, a total of 230 patients with 230 consecutive thyroid nodules were enrolled in this study. Each nodule was classified by one junior and one senior radiologist separately according to ACR TI-RADS, AACE/ACE/AME and ATA guidelines. The malignancy diagnostic performance and the number of FNA recommendations were pairwise compared among three guidelines using chi-square tests. Of the 230 thyroid nodules, 137 were malignant, and 93 were benign. However, 19.6% of the nodules (45 of 230) did not match any pattern using the ATA guidelines but with a high risk of malignancy (68.9%). The ACR TI-RADS derived the highest diagnostic performance, from both junior radiologist (AUC 0.815) and senior radiologist (AUC 0.864). The ACR guidelines also showed the greatest level of sensitivity (junior: 86.1%, senior: 94.9%), compared with AACE/ACE/AME and ATA guidelines. The number of thyroid nodules recommended to fine-needle aspiration (FNA) was the lowest (37.8%, 40.4%) by ACR TI-RADS, and meanwhile, the malignant detection rate within these nodules was highest (64.4%, 68.8%). The ACR guidelines present a higher level of diagnostic indicators and may offer a meaningful reduction in FNA recommendations with a higher malignancy detection rate.

Gross tumour volume delineation in anal cancer on T2-weighted and diffusion-weighted MRI – Reproducibility between radiologists and radiation oncologists and impact of reader experience level and DWI image quality

PurposeTo assess how gross tumour volume (GTV) delineation in anal cancer is affected by interobserver variations between radiologists and radiation oncologists, expertise level, and use of T2-weighted MRI (T2W-MRI) vs. diffusion-weighted imaging (DWI), and to explore effects of DWI quality. Methods and materialsWe retrospectively analyzed the MRIs (T2W-MRI and b800-DWI) of 25 anal cancer patients. Four readers (Senior and Junior Radiologist; Senior and Junior Radiation Oncologist) independently delineated GTVs, first on T2W-MRI only and then on DWI (with reference to T2W-MRI). Maximum Tumour Diameter (MTD) was calculated from each GTV. Mean GTVs/MTDs were compared between readers and between T2W-MRI vs. DWI. Interobserver agreement was calculated as Intraclass Correlation Coefficient (ICC), Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD). DWI image quality was assessed using a 5-point artefact scale. ResultsInterobserver agreement between radiologists vs. radiation oncologists and between junior vs. senior readers was good–excellent, with similar agreement for T2W-MRI and DWI (e.g. ICCs 0.72–0.94 for T2W-MRI and 0.68–0.89 for DWI). There was a trend towards smaller GTVs on DWI, but only for the radiologists (P = 0.03–0.07). Moderate-severe DWI-artefacts were observed in 11/25 (44%) cases. Agreement tended to be lower in these cases. ConclusionOverall interobserver agreement for anal cancer GTV delineation on MRI is good for both radiologists and radiation oncologists, regardless of experience level. Use of DWI did not improve agreement. DWI artefacts affecting GTV delineation occurred in almost half of the patients, which may severely limit the use of DWI for radiotherapy planning if no steps are undertaken to avoid them.

Deep learning-based fully automated detection and segmentation of lymph nodes on multiparametric-mri for rectal cancer: A multicentre study.

BackgroundAccurate lymph nodes (LNs) assessment is important for rectal cancer (RC) staging in multiparametric magnetic resonance imaging (mpMRI). However, it is incredibly time-consumming to identify all the LNs in scan region. This study aims to develop and validate a deep-learning-based, fully-automated lymph node detection and segmentation (auto-LNDS) model based on mpMRI.MethodsIn total, 5789 annotated LNs (diameter ≥ 3 mm) in mpMRI from 293 patients with RC in a single center were enrolled. Fused T2-weighted images (T2WI) and diffusion-weighted images (DWI) provided input for the deep learning framework Mask R-CNN through transfer learning to generate the auto-LNDS model. The model was then validated both on the internal and external datasets consisting of 935 LNs and 1198 LNs, respectively. The performance for LNs detection was evaluated using sensitivity, positive predictive value (PPV), and false positive rate per case (FP/vol), and segmentation performance was evaluated using the Dice similarity coefficient (DSC).FindingsFor LNs detection, auto-LNDS achieved sensitivity, PPV, and FP/vol of 80.0%, 73.5% and 8.6 in internal testing, and 62.6%, 64.5%, and 8.2 in external testing, respectively, significantly better than the performance of junior radiologists. The time taken for model detection and segmentation was 1.3 s/case, compared with 200 s/case for the radiologists. For LNs segmentation, the DSC of the model was in the range of 0.81–0.82.InterpretationThis deep learning–based auto-LNDS model can achieve pelvic LNseffectively based on mpMRI for RC, and holds great potential for facilitating N-staging in clinical practice.