Quantitative MRI in the Local Staging of Prostate Cancer: A Systematic Review and Meta-Analysis.

BACKGROUND: Local staging of prostate cancer (PCa) is important for treatment planning. Radiologist interpretation using qualitative criteria is variable with high specificity but low sensitivity. Quantitative methods may be useful in the diagnosis of extracapsular extension (ECE). PURPOSE: To assess the performance of quantitative MRI markers for detecting ECE. STUDY TYPE: Systematic review and meta-analysis. SUBJECTS: 4800 patients from 28 studies with histopathologically confirmed PCa on radical prostatectomy were pooled for meta-analysis. Patients from 46 studies were included for systematic review. FIELD STRENGTH/SEQUENCE: Diffusion-weighted, T2-weighted, and dynamic contrast-enhanced MRI at 1.5 T or 3 T. ASSESSMENT: PubMed, Embase, Web of Science, Scopus, and Cochrane databases were searched to identify studies on diagnostic test accuracy or association of any quantitative MRI markers with ECE. Results extracted by two independent reviewers for tumor contact length (TCL) and mean apparent diffusion coefficient (ADC-mean) were pooled for meta-analysis, but not for other quantitative markers including radiomics due to low number of studies available. STATISTICAL TESTS: Hierarchical summary receiver operating characteristic (HSROC) curves were computed for both TCL and ADC-mean, but summary operating points were computed for TCL only. Heterogeneity was investigated by meta-regression. Results were significant if P ≤ 0.05. RESULTS: At the 10 mm threshold for TCL, summary sensitivity and specificity were 0.76 [95% confidence interval (CI) 0.71-0.81] and 0.68 [95% CI 0.63-0.73], respectively. At the 15 mm threshold, summary sensitivity and specificity were 0.70 [95% CI 0.53-0.83] and 0.74 [95% CI 0.60-0.84] respectively. The area under the HSROC curves for TCL and ADC-mean were 0.79 and 0.78, respectively. Significant sources of heterogeneity for TCL included timing of MRI relative to biopsy. DATA CONCLUSION: Both 10 mm and 15 mm thresholds for TCL may be reasonable for clinical use. From comparison of the HSROC curves, ADC-mean may be superior to TCL at higher sensitivities. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 2.

Deep learning applied to automatic disease detection using chest X-rays.

Deep learning (DL) has shown rapid advancement and considerable promise when applied to the automatic detection of diseases using CXRs. This is important given the widespread use of CXRs across the world in diagnosing significant pathologies, and the lack of trained radiologists to report them. This review article introduces the basic concepts of DL as applied to CXR image analysis including basic deep neural network (DNN) structure, the use of transfer learning and the application of data augmentation. It then reviews the current literature on how DNN models have been applied to the detection of common CXR abnormalities (e.g. lung nodules, pneumonia, tuberculosis and pneumothorax) over the last few years. This includes DL approaches employed for the classification of multiple different diseases (multi-class classification). Performance of different techniques and models and their comparison with human observers are presented. Some of the challenges facing DNN models, including their future implementation and relationships to radiologists, are also discussed.

Adaptive tutorials versus web-based resources in radiology: a mixed methods analysis in junior doctors of efficacy and engagement.

BACKGROUND: Radiology education is limited in undergraduate Medicine programs. Junior doctors might not have the necessary background to effectively order and interpret diagnostic imaging investigations. Furthermore, junior doctors are often time-poor, balancing clinical commitments with ongoing learning, leadership and teaching responsibilities. Previous studies have demonstrated the efficacy of radiology-themed online adaptive tutorials for senior medical students. Such adaptive tutorials might therefore be an efficient and effective form of radiology education for junior doctors. METHODS: A randomised controlled crossover trial was performed to evaluate the impact of adaptive tutorials on learning the indications for, and interpretation of, basic imaging studies, compared with peer-reviewed web-based resources. Ninety-one volunteer junior doctors, comprising 53 postgraduate year 1 (PGY 1) and 38 postgraduate year 2 (PGY 2), were randomly allocated into two groups. In the first phase of the trial, focusing on head CT, one group accessed adaptive tutorials while the other received web-based resources. In the second phase of the trial, focusing on chest CT, the groups crossed over. Following each phase of the trial, participants completed exam-style online assessments. At the conclusion of the study, participants also completed an online questionnaire regarding perceived engagement and efficacy of each type of educational resource. RESULTS: Junior doctors completed the adaptive tutorials significantly faster than the relevant web-based resources for both head CT and chest CT (p = 0.03 and < 0.01 respectively). Mean quiz scores were higher in the groups receiving adaptive tutorials on head CT and chest CT (86.4% vs 83.5 and 77.7% vs 75% respectively). However, in contrast to previous studies in senior medical students, these differences were not statistically significant. Participants reported higher engagement and perceived value of adaptive tutorials, compared with web-based resources. CONCLUSIONS: Adaptive tutorials are more time-efficient than existing web-based resources for learning radiology by junior doctors, while both types of resources were equally effective for learning in this cohort. Junior doctors found the adaptive tutorials more engaging and were more likely to recommend these resources to their colleagues.

Editorial for "Prostate Cancer Risk Stratification in Men With a Clinical Suspicion of Prostate Cancer Using a Unique Biparametric MRI and Expression of 11 Genes in Apparently Benign Tissue: Evaluation Using Machine-Learning Techniques".

LEVEL OF EVIDENCE: 5 TECHNICAL EFFICACY STAGE: 3 J. Magn. Reson. Imaging 2020;51:1554-1555.

Adaptive Tutorials Versus Web-Based Resources in Radiology: A Mixed Methods Analysis of Efficacy and Engagement in Senior Medical Students.

RATIONALE AND OBJECTIVES: Radiology education is suited to delivery via e-learning which may be used to fill gaps in knowledge and help prepare medical students for internship. There is limited evidence of effectiveness of adaptive tutorials, a form of e-learning in a senior medical student cohort. MATERIALS AND METHODS: A randomized mixed methods crossover trial was performed to assess effectiveness of adaptive tutorials on engagement and understanding of appropriate use and interpretation of basic imaging studies. Eighty-one volunteer medical students from years 5 and 6 of a 6-year program were randomly allocated to one of two groups. In the first phase of the trial on head CT, one group received access to adaptive tutorials and the other to peer-reviewed web-based resources. A cross over was performed and the second phase of the trial addressing chest CT commenced. Examination style assessments were completed at the end of each phase. At the trial's conclusion, an online questionnaire was provided to evaluate student perceptions of engagement and efficacy of each educational resource. RESULTS: Adaptive tutorial groups in both phases achieved higher mean scores than controls which were statistically significant in the first phase only. Students reported higher engagement and overall perceived value of the adaptive tutorials than controls. CONCLUSION: Adaptive tutorials are overwhelmingly supported by senior medical students. Questionnaire responses suggest the engaging nature of the tutorials efficiently aids participation and knowledge retention which is in principle supported by test results.

Peripheral bronchial identification on chest CT using unsupervised machine learning.

PURPOSE: To automatically identify small- to medium-diameter bronchial segments distributed throughout the lungs. METHODS: We segment the peripheral pulmonary vascular tree and construct cross-sectional images perpendicular to the lung vasculature. The bronchi running with pulmonary arteries appear as concentric rings, and potential center points that lie within the bronchi are identified by looking for circles (using the circular Hough transform) and rings (using a novel variable ring filter). The number of candidate bronchial center points are further reduced by using agglomerative hierarchical clustering applied to the points represented with 18 features pertaining to their 3D position, orientation and appearance of the surrounding cross-sectional image. Resulting clusters corresponded to bronchial segments. Parameters of the algorithm are varied and applied to two experimental data sets to find the best values for bronchial identification. The optimized algorithm was then applied to a further 21 CT studies obtained using two different CT vendors. RESULTS: The parameters that result in the most number of true positive bronchial center points with > 95% precision are a tolerance of 0.15 for the hierarchical clustering algorithm and a threshold of 75 HU with 10 spokes for the ring filter. Overall, the performance on all 21 test data sets from CT scans from both vendors demonstrates a mean number of 563 bronchial points detected per CT study, with a mean precision of 96%. The detected points across this group of test data sets are relatively uniformly distributed spatially with respect to spherical coordinates with the origin at the center of the test imaging data sets. CONCLUSION: We have constructed a robust algorithm for automatic detection of small- to medium-diameter bronchial segments throughout the lungs using a combination of knowledge-based approaches and unsupervised machine learning. It appears robust over two different CT vendors with similar acquisition parameters.

Electrocardiographic measurement of infarct size compared to cardiac MRI in reperfused first time ST-segment elevation myocardial infarction.

BACKGROUND: Myocardial infarct size (IS) following ST-segment elevation myocardial infarction (STEMI) is an important prognostic factor. We assessed the Selvester 32-point QRS score from the 12-lead ECG for measurement of IS in STEMI patients receiving reperfusion therapy compared to cardiac magnetic resonance imaging (CMRI). Furthermore we sought to explore the impact of microvascular obstruction (MVO) on the scoring system, and determine factors contributing to discrepancies between CMRI IS and Selvester score. METHODS: We examined 70 patients (55 men, 15 women), mean age 57±10years with a first time STEMI (46 anterior, 24 non-anterior). QRS scores were calculated early and at follow-up (mean 2±1 and 59±14days post-STEMI). Myocardial core scar size (5SD) was measured at 5.3±3.3 and 57.8±13.5days post-infarction by CMRI. MVO was determined on initial MRI. Logistic regression analysis was performed to determine factors contributing to discordant scores, defined as a difference between CMRI and Selvester IS of >6% myocardium. RESULTS: QRS scoring of anterior infarcts correlated with CMRI IS both early (r=0.734, p<0.0001) and at follow-up (r=0.716, p<0.0001); however no correlation was seen among non-anterior infarcts. QRS scoring overestimated IS at all time points. There was better agreement between ECG and CMRI measured IS in patients without MVO at both time points. Anterior infarction was inversely predictive of discordant IS estimation acutely, and larger Selvester scores were predictive of inaccurate scoring at both time periods. CONCLUSIONS: Selvester QRS score correlates well with CMRI IS for anterior infarcts. MVO did not independently affect the score.

High-sensitivity troponin T predicts infarct scar characteristics and adverse left ventricular function by cardiac magnetic resonance imaging early after reperfused acute myocardial infarction.

BACKGROUND: Late gadolinium enhancement cardiac magnetic resonance imaging (CMRI) is the current standard for evaluation of myocardial infarct scar size and characteristics. Because post-ST-segment elevation myocardial infarction (STEMI) troponin levels correlate with clinical outcomes, we sought to determine the sampling period for high-sensitivity troponin T (hs-TnT) that would best predict CMRI-measured infarct scar characteristics and left ventricular (LV) function. METHODS AND RESULTS: Among 201 patients with first presentation with STEMI who were prospectively recruited, we measured serial hs-TnT levels at admission, peak, 24 hours, 48 hours, and 72 hours after STEMI. Indexed LV volumes, LV ejection fraction (LVEF) and infarct scar characteristics (scar size, scar heterogeneity, myocardial salvage index, and microvascular obstruction) were evaluated by CMRI at a median of 4 days post-STEMI. Peak and serial hs-TnT levels correlated positively with early indexed LV volumes and infarct scar characteristics, and negatively correlated with myocardial salvage index and LVEF. Both 48- and 72-hour hs-TnT levels similarly predicted "large" total infarct scar size (odds ratios [ORs] 3.08 and 3.53, both P < .001), myocardial salvage index (ORs 1.68 and 2.30, both P < .001), and LVEF <40% (ORs 2.16 and 2.17, both P < .001) on univariate analyses. On multivariate analyses, 48- and 72-hour hs-TnT levels independently predicted large infarct scar size (ORs 2.05 and 2.31, both P < .001), reduced myocardial salvage index (OR 1.39 [P = .031] and OR 1.55 [P = .009]), and LVEF <40% (OR 1.47 [P = .018] and OR 1.43 [P = .026]). All measured hs-TnT levels had a modest association and similar capacity to predict microvascular obstruction. CONCLUSIONS: Levels of hs-TnT at 48 and 72 hours, measured during the "plateau phase" post-STEMI, predicted infarct scar size, poor myocardial salvage, and LVEF. These levels also correlated with scar heterogeneity and microvascular obstruction post-STEMI. Since ascertaining peak levels after STEMI is challenging in routine practice, based on the biphasic kinetics of hs-TnT, a measurement at 48 to 72 hours (during the plateau phase) provides a useful and simple method for early evaluation of LV function and infarct scar characteristics.

The scapuloacromial angle: a 3D analysis of acromial slope and its relationship with shoulder impingement.

PURPOSE: To quantify acromial slope (the scapuloacromial [SA] angle). MATERIALS AND METHODS: We used shoulder MR studies in 64 subjects who were stratified surgically into impingement (with and without tear) and nonimpingement (shoulder instability) groups. We measure, in three dimensions, the angle between unit normal vectors to planes of best fit to the undersurface of the acromion and the medial surface of the blade of the scapula. We examine the projection of this SA angle onto the transverse, coronal, and sagittal planes to gauge some measure of rotation, lateral tilt, and anterior tilt with respect to the patient. Orientation of the acromion and scapular blade in relation to the MR scanner and therefore to the patient was also calculated. RESULTS: We found no significant differences between these two groups when examining age and side of pathology. A statistically significant difference was seen between males and females. CONCLUSION: Our study does not support the idea that a flatter acromial slope causes impingement and is an etiological basis for rotator cuff tears. We believe that acromial slope, regardless of imaging plane, is not useful in classifying patients who present with shoulder pain and should not be considered as a source of pathologic changes.

Shoulder impingement: objective 3D shape analysis of acromial morphologic features.

PURPOSE: To retrospectively and quantitatively analyze the acromial undersurface in three dimensions and to determine its association with impingement syndrome and rotator cuff tears. MATERIALS AND METHODS: Institutional review board approval was received with exemption of informed consent for this retrospective HIPAA-compliant study. Magnetic resonance images were evaluated in 84 patients (63 males, 21 females; mean age, 42.6 years; age range, 15-74 years). On the basis of surgical results, patients were separated into three groups: those without shoulder impingement or rotator cuff tears (31 patients), those with shoulder impingement (22 patients), and those with rotator cuff tears (31 patients). To quantitate the acromial undersurface, the structure was manually plotted, and a mathematic model was created by using splines. The undersurface was divided into a 20 x 20 grid. For each patient, a shape index (SI) data set and local undersurface angulation (LUA) data set were determined. Regression analyses were used to identify differences between groups, and prediction models were constructed on the basis of correspondent areas. RESULTS: Analysis of both data sets yielded no apparent progression between groups and demonstrated a lack of similarity between the impingement and rotator cuff tear groups. Discrimination between groups could be demonstrated by dividing the lateral portion of the acromial undersurface into contiguous blocks. The highest overall diagnostic accuracy of our prediction models was 58.3% (49 of 84) by using 10 blocks of the SI data set and 73.8% (62 of 84) by using five blocks of the LUA data set. CONCLUSION: Three-dimensional modeling yields objective data about the acromial undersurface. On the basis of this method, osseous impingement by the acromion is not a primary cause of shoulder impingement syndrome or rotator cuff tears.

Quantification of the curvature and shape of the interventricular septum.

The interventricular septum (IVS) occupies a unique position within the heart, lying between the left (LV) and right (RV) ventricular cavities. Changes in its normal geometry may signify not only abnormalities of the septal myocardium, but also abnormal pressure differences between the LV and RV. Flattening of the IVS has been noted with cross-sectional imaging in association with pulmonary hypertension, but the septal curvature and shape have not previously been measured in three dimensions. This paper describes a method to model the RV surface of the IVS from spatially registered cross-sectional images for measurements of curvature. A smoothing 2D spline surface is constructed through the RV septal surface at regular times during the cardiac cycle, and the principal curvatures, as well as the Gaussian and mean curvatures, shape index, and curvedness, are calculated. Vector and color surface maps and graphs of average curvature and shape indices are constructed. Consistent curvature patterns were observed in four normal subjects. This method of measuring septal geometry can provide potentially useful new information on the effects of RV disease. We examine the problem of describing septal motion, and describe a simple measure of septal curvature that may be of clinical value.