Assessment of Radiology Artificial Intelligence Software: A Validation and Evaluation Framework.

Artificial intelligence (AI) software in radiology is becoming increasingly prevalent and performance is improving rapidly with new applications for given use cases being developed continuously, oftentimes with development and validation occurring in parallel. Several guidelines have provided reporting standards for publications of AI-based research in medicine and radiology. Yet, there is an unmet need for recommendations on the assessment of AI software before adoption and after commercialization. As the radiology AI ecosystem continues to grow and mature, a formalization of system assessment and evaluation is paramount to ensure patient safety, relevance and support to clinical workflows, and optimal allocation of limited AI development and validation resources before broader implementation into clinical practice. To fulfil these needs, we provide a glossary for AI software types, use cases and roles within the clinical workflow; list healthcare needs, key performance indicators and required information about software prior to assessment; and lay out examples of software performance metrics per software category. This conceptual framework is intended to streamline communication with the AI software industry and provide healthcare decision makers and radiologists with tools to assess the potential use of these software. The proposed software evaluation framework lays the foundation for a radiologist-led prospective validation network of radiology AI software. Learning Points: The rapid expansion of AI applications in radiology requires standardization of AI software specification, classification, and evaluation. The Canadian Association of Radiologists' AI Tech & Apps Working Group Proposes an AI Specification document format and supports the implementation of a clinical expert evaluation process for Radiology AI software.

Transabdominal ultrasound of pancreatic ductal adenocarcinoma: A multi-centered population-based study in sensitivity, associated diagnostic intervals, and survival.

OBJECTIVES: To determine the sensitivity of ultrasound (US) in detecting pancreatic ductal adenocarcinoma in our region, to identify factors associated with US test result, and assess the impact on the diagnostic interval and survival. METHODS: Patients diagnosed between January 1, 2014 and December 31, 2015 in Nova Scotia, Canada were identified by a cancer registry. US performed prior to diagnosis were retrospectively graded as true positive (TP), indeterminate or false negative (FN). Amongst US results, differences in age, weight and tumor size were assessed [one-way analysis of variance (ANOVA)]. Associations between result and sex, tumor location (proximal/distal), clinical suspicion of malignancy, and visualization of the pancreas, tumor, secondary signs and liver metastases were assessed (Chi-square). Mean follow-up imaging, diagnostic, and survival intervals were assessed (one-way ANOVA). RESULTS: One hundred thirteen US of 107 patients (54 women; mean 70 ± 13 years) were graded as follows: 48/113 (42.5%) TPs; 42/113 (37.2%) indeterminates; and 23/113 (20.4%) FNs. Sensitivity was 48/71(67.6%). There was no difference in age, weight or tumor size amongst US result (P > 0.5). FNs had proportionally more men (P = 0.011) and lacked clinical suspicion of malignancy (P = 0.0006); TPs had proportionally more proximal tumors (P = 0.017). US result was associated with visualization of the pancreas, tumor, secondary signs and liver metastases (P < 0.005). FNs had longer mean follow-up imaging (P < 0.0001) and diagnostic (P = 0.0007) intervals, and worse mean survival (P = 0.034). CONCLUSIONS: In our region, the sensitivity of US in detecting pancreatic ductal adenocarcinoma is 67.6%. A false negative US is associated with delayed diagnostic work-up and worse mean survival.

Prospective Evaluation of Virtual MR Elastography With Diffusion-Weighted Imaging in Subjects With Nonalcoholic Fatty Liver Disease.

BACKGROUND: Non-alcoholic fatty liver disease (NAFLD) is increasingly common worldwide and can lead to the development of cirrhosis, liver failure and cancer. Virtual magnetic resonance elastography (VMRE), which is based on a shifted apparent diffusion coefficient (sADC), is a potential noninvasive method to assess liver fibrosis without the specialized hardware and expertise required to implement traditional MR elastography (MRE). Although hepatic steatosis is known to confound ADC measurements, previous studies using VMRE have not corrected for hepatic fat fraction. PURPOSE: To compare VMRE, corrected for the confounding effects of unsuppressed fat signal, to MRE and biopsy in subjects with suspected NAFLD. STUDY TYPE: Prospective, cross-sectional. POPULATION: A total of 49 adult subjects with suspected NAFLD (18 male; median age 55 years, range 33-74 years) who underwent liver biopsy. FIELD STRENGTH/SEQUENCE: 3T, diffusion-weighted spin echo planar, chemical-shift encoded (IDEAL IQ) and MRE sequences. ASSESSMENT: Two observers drew regions of interest on sADC, proton density fat fraction and MRE-derived stiffness maps. Fat-corrected sADC values were used to calculate the diffusion-based shear modulus according to the VMRE method. Predicted fibrosis stage for MRE and VMRE was determined using previously published cut-off values. STATISTICAL TESTS: The relationship between VMRE and MRE was assessed with least-squares linear regression (coefficient of determination, R2 ). Agreement between MRE and VMRE-predicted fibrosis stage was evaluated with a kappa coefficient and accuracy compared using McNemar's test. A one-way ANOVA determined if the fat-corrected sADC (VMRE) and MRE differed by fibrosis stage. A P value < 0.05 was considered statistically significant. RESULTS: Least squares regression of VMRE vs. MRE revealed R2  = 0.046 and a slope that was not significantly different from zero (P = 0.14). There was no agreement between MRE and VMRE-predicted fibrosis stage (kappa = -0.01). The proportion of correctly predicted fibrosis stage was significantly higher for MRE compared to VMRE. MRE was significantly associated with fibrosis stage, but fat-corrected sADC was not (P = 0.24). DATA CONCLUSION: Fat-corrected VMRE was not associated with fibrosis stage in NAFLD. Further investigation is required if VMRE is to be considered in subjects with NAFLD. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 2.

US LI-RADS Visualization Score: Interobserver Variability and Association With Cause of Liver Disease, Sex, and Body Mass Index.

OBJECTIVES: To evaluate the interobserver agreement between radiologists using the Ultrasound Liver Reporting And Data System (US LI-RADS) visualization score and assess association between visualization score and cause of liver disease, sex, and body mass index (BMI). METHODS: This retrospective, single institution, cross-sectional study evaluated 237 consecutive hepatocellular carcinoma surveillance US examinations between March 4, 2017 and September 4, 2017. Five abdominal radiologists independently assigned a US LI-RADS visualization score (A, no or minimal limitations; B, moderate limitations; C, severe limitations). Interobserver agreement was assessed with a weighted Kappa statistic. Association between US visualization score (A vs B or C) and cause of liver disease, sex, and BMI (< or ≥ 25 kg/m2) was evaluated using univariate and multivariate analyses. RESULTS: The average weighted Kappa statistic for all raters was 0.51. A score of either B or C was assigned by the majority of radiologists in 148/237 cases and was significantly associated with cause of liver disease (P = 0.014) and elevated BMI (P < 0.001). Subjects with viral liver disease were 3.32 times (95% CI: 1.44-8.38) more likely to have a score of A than those with non-alcoholic steatohepatitis (P = 0.007). The adjusted odds ratio of visualization score A was 0.249 (95% CI: 0.13-0.48) among those whose BMI was ≥25 kg/m2 vs. BMI < 25 kg/m2. CONCLUSION: Interobserver agreement between radiologists using US LI-RADS score was moderate. The majority of US examinations were scored as having moderate or severe limitations, and this was significantly associated with non-alcoholic steatohepatitis and increased BMI.

Diagnostic performance of US for suspected appendicitis: Does multi-categorical reporting provide better estimates of disease in adults, and what factors are associated with false or indeterminate results?

PURPOSE: To identify factors associated with false or indeterminate US result for suspected appendicitis, and assess whether multi-categorical reporting of US yields more precise estimates regarding the probability of appendicitis. METHODS: 562 US examinations for suspected appendicitis between May 2013-April 2015 were categorized as true (77/562 true positives or true negatives) or false/indeterminate (485/562 false negatives, false positives or indeterminates) based on results from a prior study. Of 541 examinations with images available retrospectively, a category of A-E was assigned as follows: non-visualized appendix with secondary findings (A) absent or (B) present; appendix visualized and considered (C) negative, (D) equivocal, or (E) positive for appendicitis. The following factors were recorded: age; sex; scan time (daytime vs. off-hours); resident/fellow involvement; abdominal subspecialty radiologist; radiologist experience (>5 years or not); and tenderness on interrogation. Associations between factors and US result were assessed (t-tests, Fisher's exact test and multivariate logistic regression). RESULTS: The true group had proportionally more males (18/77 (23.4%) vs. 66/485 (13.6%), p = 0.04) and patients with sonographic tenderness (43/77 (55.8%) vs. 132/353 (27.3%), p < 0.0001). There was no significant difference or association with other factors. On multivariate logistic regression, false/indeterminate results were 1.9 times (95% CIs 1.0-3.5) more likely among females and 3.8 times more likely in the absence of tenderness (95% CIs 2.3-6.4). The proportion of patients with appendicitis in categories A-E was 34/410 (8.3%), 24/44 (54.5%), 0/18 (0%), 0/3 (0%) and 61/66 (92.4%), respectively. CONCLUSIONS: Females and absence of tenderness were associated with a false/indeterminate US. Categorical reporting provides more granular estimates of the post-test probability of appendicitis.

Toward individualized prediction of seizure recurrence: Hippocampal neuroimaging features in a cohort of patients from a first seizure clinic.

PURPOSE: We performed an exploratory analysis of electroencephalography (EEG) and neuroimaging data from a cohort of 51 patients with first seizure (FS) and new-onset epilepsy (NOE) to identify variables, or combinations of variables, that might discriminate between clinical trajectories over a one-year period and yield potential biomarkers of epileptogenesis. METHODS: Patients underwent EEG, hippocampal and whole brain structural magnetic resonance imaging (MRI), diffusion tensor imaging (DTI), and magnetic resonance spectroscopy (MRS) within six weeks of the index seizure, and repeat neuroimaging one year later. We classified patients with FS as having had a single seizure (FS-SS) or having converted to epilepsy (FS-CON) after one year and performed logistic regression to identify combinations of variables that might discriminate between FS-SS and FS-CON, and between FS-SS and the combined group FS-CON + NOE. We performed paired t-tests to assess changes in quantitative variables over time. RESULTS: Several combinations of variables derived from hippocampal structural MRI, DTI, and MRS provided excellent discrimination between FS-SS and FS-CON in our sample, with areas under the receiver operating curve (AUROC) ranging from 0.924 to 1. They also provided excellent discrimination between FS-SS and the combined group FS-CON + NOE in our sample, with AUROC ranging from 0.902 to 1. After one year, hippocampal fractional anisotropy (FA) increased bilaterally, hippocampal radial diffusivity (RD) decreased on the side with the larger initial measurement, and whole brain axial diffusivity (AD) increased in patients with FS-SS; hippocampal volume decreased on the side with the larger initial measurement, hippocampal FA increased bilaterally, hippocampal RD decreased bilaterally and whole brain AD, FA and mean diffusivity increased in the combined group FS-CON + NOE (corrected threshold for significance, q = 0.017). CONCLUSION: We propose a prospective, multicenter study to develop and test models for the prediction of seizure recurrence in patients after a first seizure, based on hippocampal neuroimaging. Further longitudinal neuroimaging studies in patients with a first seizure and new-onset epilepsy may provide clues to the microstructural changes occurring at the earliest stages of epilepsy and yield biomarkers of epileptogenesis.

Ultrasound and CT in the Diagnosis of Appendicitis: Accuracy With Consideration of Indeterminate Examinations According to STARD Guidelines.

OBJECTIVE. The objective of our study was to determine the accuracy of ultrasound (US) and CT in diagnosing appendicitis at our institution while taking into account the number of indeterminate examinations in accordance with the Standards for Reporting Diagnostic Accuracy (STARD) guidelines. MATERIALS AND METHODS. We retrospectively evaluated 790 patients who underwent US, CT, or both for evaluation of suspected appendicitis between May 1, 2013, and April 30, 2015. Patient characteristics and US and CT examination results were recorded. The reference standard was histopathology or 3 months of medical record follow-up if surgery was not performed; 3 × 2 tables were generated, and sensitivity, specificity, overall test yield, and accuracy were calculated according to STARD guidelines. For surgical cases, time to surgery (one-way ANOVA) was compared among patients who underwent US alone, CT alone, or both US and CT. RESULTS. A total of 473 of 562 US examinations had indeterminate findings (overall test yield, 15.8%); sensitivity and specificity in the 89 diagnostic examinations were 98.5% and 54.2%, respectively. Thirteen of 522 CT examinations were indeterminate (overall test yield, 97.5%); sensitivity and specificity in the remaining 509 CT examinations were 98.9% and 97.2%, respectively. Taking indeterminate studies into account, the accuracy was 13.7% for US and 95.6% for CT. The negative appendectomy rates were 17.7% (11/62) for US and 3.3% (9/276) for CT (p = 0.0002). Time to surgery was longer for patients who underwent US and CT (mean ± SD, 17.7 ± 8.9 hours) than US alone (12.9 ± 6.4 hours; p = 0.002) but was not longer for patients who underwent CT alone (16.3 ± 8.4 hours; p = 0.45). CONCLUSION. At our institution, a large proportion of US examinations are indeterminate for appendicitis. CT is the preferred first-line imaging test for evaluating appendicitis in nonobstetric adult patients.

Assessing breast cancer risk within the general screening population: developing a breast cancer risk model to identify higher risk women at mammographic screening.

OBJECTIVES: To develop a breast cancer risk model to identify women at mammographic screening who are at higher risk of breast cancer within the general screening population. METHODS: This retrospective nested case-control study used data from a population-based breast screening program (2009-2015). All women aged 40-75 diagnosed with screen-detected or interval breast cancer (n = 1882) were frequency-matched 3:1 on age and screen-year with women without screen-detected breast cancer (n = 5888). Image-derived risk factors from the screening mammogram (percent mammographic density [PMD], breast volume, age) were combined with core biopsy history, first-degree family history, and other clinical risk factors in risk models. Model performance was assessed using the area under the receiver operating characteristic curve (AUC). Classifiers assigning women to low- versus high-risk deciles were derived from risk models. Agreement between classifiers was assessed using a weighted kappa. RESULTS: The AUC was 0.597 for a risk model including only image-derived risk factors. The successive addition of core biopsy and family history significantly improved performance (AUC = 0.660, p < 0.001 and AUC = 0.664, p = 0.04, respectively). Adding the three remaining risk factors did not further improve performance (AUC = 0.665, p = 0.45). There was almost perfect agreement (kappa = 0.97) between risk assessments based on a classifier derived from image-derived risk factors, core biopsy, and family history compared with those derived from a model including all available risk factors. CONCLUSIONS: Women in the general screening population can be risk-stratified at time of screen using a simple model based on age, PMD, breast volume, and biopsy and family history. KEY POINTS: • A breast cancer risk model based on three image-derived risk factors as well as core biopsy and first-degree family history can provide current risk estimates at time of screen. • Risk estimates generated from a combination of image-derived risk factors, core biopsy history, and first-degree family history may be more valid than risk estimates that rely on extensive self-reported risk factors. • A simple breast cancer risk model can avoid extensive clinical risk factor data collection.

Dosing Iodinated Contrast Media According to Lean Versus Total Body Weight at Abdominal CT: A Stratified Randomized Controlled Trial.

RATIONALE AND OBJECTIVES: To compare the magnitude and interpatient variability in normalized mean hepatic enhancement (MHE) indices when dosing contrast media (CM) according to total body weight (TBW) and lean body weight (LBW). MATERIALS AND METHODS: This ethics-approved stratified randomized controlled study allocated 280 outpatients for abdominal Computed Tomography (CT) between February-November 2018 to TBW- or LBW-dosing using computer-generated tables. CTs were acquired in portal venous phase after fixed 35-second injection of Iohexol 350. Patients with missing precontrast image, incorrect dose, or chronic kidney, liver or heart disease were excluded. The number of included patients and CM doses were: TBW arm, 51 women and 60 men, 1.22 mL/kg; LBW arm, 59 women, 1.66 mL/kg LBW, and 59 men, 1.52 mL/kg LBW. Liver attenuations were obtained from regions of interest. Values and standard deviations in MHE indices normalized to iodine dose (MHE/I) and iodine dose per kg TBW (aMHE = MHE/[I/TBW]) were compared (unpaired t tests and F-tests). RESULTS: Cohorts were similar in age, sex, TBW, and LBW. TBW groups received more CM than LBW groups: men, 106.5 ± 20 versus 98.4 ± 11 mL, p = 0.007; women, 93.7 ± 20 versus 77.5 ± 11 mL, p < 0.0001. TBW and LBW groups showed no significant difference in MHE/I (women, 1.75 ± 0.5 versus 1.86 ± 0.6 HU/g, p = 0.31; men, 1.53 ± 0.4 versus 1.52 ± 0.4 HU/g, p = 0.90) or aMHE (women, 0.03 ± 0.01 versus 0.03 ± 0.01 HU/g/kg, p = 0.25; men, 0.02 ± 0.01 versus 0.02 ± 0.01 HU/g/kg, p = 0.52). Variances in MHE/I and aMHE were not significantly different for all groups (p > 0.05). CONCLUSION: TBW- and LBW-based CM dosing yield a similar magnitude and interpatient variability in normalized MHE indices at routine abdominal CT.

The yield and patient factors associated with CT colonography C-RADS results in a non-screening patient population.

OBJECTIVES: To determine the proportion of diagnostic computed tomography colonography (CTC) Reporting and Data System (C-RADS) categories in a non-screening population, and which patient factors are associated with a positive CTC (C2-4), a non-diagnostic CTC (C0), and potentially relevant extracolonic findings (ECF, E3-4). METHODS: Diagnostic CTCs performed at a single academic center from 2017 to 2018 were retrospectively reviewed. For each examination, the indications, age, sex, admission status, and C-RADS categories were recorded. Multivariate logistic regression was performed of patient demographic factors and clinical indications, with adjusted odds ratios (OR) and 95% confidence intervals. RESULTS: 1373 CTCs were included. The mean age was 66.4 ± 13 years (range 24-97). There were 782 women and 75 inpatients. The number of CTCs reported as C0-C4 were 194/1373 (14.1%), 970/1373 (70.6%), 77/1373 (5.6%), 86/1373 (6.3%), and 46/1373 (3.4%), respectively, and 134/1373 (9.8%), 960/1373 (69.9%), 173/1373 (12.6%), and 106/1373 (7.7%) CTCs were reported as E1-4, respectively. Factors that demonstrated the strongest associations were as follows: with C2-4, age groups 50-79 (OR 2.8, 95% confidence interval 1.4-6.1), 80-89 (6.2, 2.9-14.5) and ≥ 90 (7.6, 2.0-29.1), and inpatients (3.4, 1.8-6.4); with C0, age groups 50-79 (5.9, 2.2-24.4), 80-89 (9.8, 3.4-41.8), and ≥ 90 (22.5, 5.8-113.0), incomplete colonoscopy (3.2, 2.0-5.1) and melena or gastrointestinal bleeding (4.1, 1.8-9.4); and with E3-4, age groups 50-79 (1.6, 1.0-2.9), 80-89 (2.0, 1.1-3.9), and ≥ 90 (3.2, 1.2-8.8), and inpatients (2.3, 1.3-3.9). CONCLUSION: Older age is increasingly associated with a positive test, a non-diagnostic test and potentially relevant ECF. Inpatients are also associated with positive tests and E3-4 findings. Symptoms are not strongly associated with a positive CTC.

Maternal pre-pregnancy obesity and health care utilization and costs in the offspring.

BACKGROUND/OBJECTIVE: The association between maternal pre-pregnancy obesity and adverse child health outcomes is well described, but there are few data on the relationship with offspring health service use. We examined the influence of maternal pre-pregnancy obesity on offspring health care utilization and costs over the first 18 years of life. METHODS: This was a population-based retrospective cohort study of children (n = 35,090) born between 1989 and 1993 and their mothers, who were identified using the Nova Scotia Atlee Perinatal Database and linked to provincial administrative health data from birth through 2014. The primary outcome was health care utilization as determined by the number and cost of physician visits, hospital admissions and days, and high utilizer status (>95th percentile of physician visits). The secondary outcome was health care utilization by ICD chapter. Maternal pre-pregnancy weight was categorized as normal weight, overweight, or obese. Multivariable-adjusted regression models were used to examine the association between maternal weight status and offspring health care use. RESULTS: Children of mothers with pre-pregnancy obesity had more physician visits (10%), hospital admissions (16%), and hospital days (10%) than children from mothers of normal weight over the first 18 years of life. Offspring of mothers with obesity had C$356 higher physician costs and C$1415 hospital costs over 18 years than offspring of normal weight mothers. Children of mothers with obesity were 1.74 times more likely to be a high utilizer of health care and had higher rates of physician visits and hospital stays for nervous system and sense organ disorders, respiratory disorders, and gastrointestinal disorders compared to children of normal weight mothers. CONCLUSION: Our findings suggest that maternal pre-pregnancy overweight and obesity are associated with slightly higher offspring health care utilization and costs in the first 18 years of life.

Classification tree for the prediction of malignant disease and the prediction of non-diagnostic biopsies in patients with small renal masses.

INTRODUCTION: Preoperative prediction of benign vs. malignant small renal masses (SRMs) remains a challenge. This study: 1) validates our previously published classification tree (CT) with an external cohort; 2) creates a new CT with the combined cohort; and 3) evaluates the RENAL and PADUA scoring systems for prediction of malignancy. METHODS: This study includes a total of 818 patients with renal masses; 395 underwent surgical resection and 423 underwent biopsy. A CT to predict benign disease was developed using patient and tumour characteristics from the 709 eligible participants. Our CT is based on four parameters: tumour volume, symptoms, gender, and symptomatology. CART modelling was also used to determine if RENAL and PADUA scoring could predict malignancy. RESULTS: When externally validated with the surgical cohort, the predictive accuracy of the old CT dropped. However, by combining the cohorts and creating a new CT, the predictive accuracy increased from 74% to 87% (95% confidence interval 0.84-0.89). RENAL and PADUA score alone were not predictive of malignancy. One limitation was the lack of available histological data from the biopsy series. CONCLUSIONS: The validated old CT and new combined-cohort CT have a predictive value greater than currently published nomograms and single-biopsy cohorts. Overall, RENAL and PADUA scores were not able to predict malignancy.

Impact of Detection Method and Age on Survival Outcomes in Triple-Negative Breast Cancer: A Population-Based Cohort Analysis.

BACKGROUND: Most investigations have compared triple-negative breast cancer (TNBC) to non-TNBC to elucidate clinical or epidemiologic differences between subtypes. We examined a contemporary cohort of patients with primary TNBC by detection and age at diagnosis within a population-based breast screening program to examine survival outcomes. PATIENTS AND METHODS: All women with a diagnosis of primary TNBC between January 1, 2005, and December 31, 2012, in Nova Scotia, Canada, were included. Clinicopathologic and detection variables were abstracted from the Nova Scotia Breast Screening Program. Patient charts were abstracted for adjuvant therapies and survival outcomes, supplemented by provincial vital statistical data. RESULTS: A total of 412 patients comprised the study population, with almost half aged over 60 years (46.3%) and 30.2% having screen-detected disease. There were no significant differences in prognostic variables between age groups. Younger patients were more likely to receive adjuvant chemotherapy (96.3% ≤ 49 years vs. 31.2% ≥ 70 years), but there were no differences in disease-free or breast cancer-specific survival between the age groups. For those with disease recurrence, median time to recurrence and death was shorter for younger patients (17 vs. 26 months, 16 vs. 33 months respectively; age 40-49 vs. 70+). Those with screen-detected disease had better disease-free, breast cancer-specific, and overall survival outcomes. CONCLUSION: Detection method may play a role in TNBC survival outcomes, thus supporting novel screening strategies for TNBC. Shorter time to survival events in the younger patient groups suggests that TNBC is a clinically heterogeneous disease despite similarities in prognostic factors across age.

How Do Different Indices of Hepatic Enhancement With Gadoxetic Acid Compare in Predicting Liver Failure and Other Major Complications After Hepatectomy?

OBJECTIVE: The objective of this study was to assess the accuracy of gadoxetic acid hepatic enhancement indices in predicting posthepatectomy liver failure (PHLF) and other major complications (OMCs). METHODS: Sixty-five patients underwent prehepatectomy gadoxetic acid-enhanced magnetic resonance imaging. Enhancement indices were calculated by obtaining regions of interest on magnetic resonance images and segmented volumes of the liver and spleen. Multivariate regression analysis was performed to predict PHLF and OMC as a function of the indices, and areas under the receiver operator characteristic (AUROC) curves were calculated. RESULTS: Areas under the receiver operator characteristic values varied from 0.412 to 0.681 and 0.462 to 0.738 in predicting PHLF and OMC, respectively. The most accurate indices in predicting PHLF were the region of interest-based, fat-normalized relative liver enhancement and liver enhancement index (AUROC, 0.681). The most accurate index in predicting OMC was the volumetric least-squares regression slope of a pharmacokinetic model (Khep_V, AUROC, 0.738). CONCLUSIONS: Indices of gadoxetic acid liver enhancement demonstrate variable performance in predicting PHLF and OMC.

Breast density scales: the metric matters.

OBJECTIVE: Measures of percent mammographic density (PMD) are often categorized using various density scales. The purpose of this study was to examine information loss associated with the use of categorical density scales. METHODS: Baseline PMD was assessed at 1% precision for 2,374 females. The data were used to create 21-category, 4-category and 2-category density scales. R-squared and root mean square error were used to evaluate the effect of categorizing PMD. The area under the receiver operator characteristic curves were compared between cancer risk models employing solely categorical PMD scales and solely baseline PMD for a subset of females (424 cases, 848 controls). RESULTS: R-squared value decreased from 1.00 (1% PMD) to 0.56 (2-category scale), while root mean square error increased from 0.00 (1% PMD) to 10.83 (2-category scale). The area under the receiver operator characteristic curve decreased from 0.64 for a cancer risk model using 1% PMD to 0.58 for a risk model using a 21-category density scale (p < 0.0001), 0.55 for a 4-category Breast Imaging, Reporting and Data System-like scale (p < 0.0001) and 0.50 for a 2-category Breast Imaging, Reporting and Data System-like scale (high vs low) (p < 0.0001). CONCLUSION: Categorizing PMD measures into categorical density scales leads to a significant loss of information. Indeed, a simple high versus low split of PMD using a 50% cut point yields a cancer risk model with no discriminatory power. Advances in knowledge: Use of categorical mammographic density scales rather than continuous percent mammographic density measures leads to significant loss of information. Breast cancer risk models using categorical mammographic density scales perform more poorly than models using continuous PMD measures.

Diagnostic Reference Levels and Monitoring Practice Can Help Reduce Patient Dose From CT Examinations.

OBJECTIVE: The purpose of this study is to establish provincial diagnostic reference levels (DRLs) and to determine whether this process may help reduce the patient radiation dose from the most frequently performed CT examinations. MATERIALS AND METHODS: We investigated the following CT examinations: head, chest, low-dose chest, abdomen and pelvis, and chest, abdomen, and pelvis examinations. The sample for each protocol included 15 patients of average body weight (mean [± SD], 70 ± 20 kg). The differences in dose between scanners were evaluated using one-way ANOVA. Correlations between dose, scanner age, and the number of detector rows were assessed using the Pearson correlation coefficient. A sample of abdominal and chest examinations were randomized and blinded for review by experienced radiologists who graded diagnostic image quality. Provincial DRLs were calculated as the 75th percentile of patient dose distributions. For hospitals with doses exceeding the DRLs, dose reduction was recommended, followed by another survey. RESULTS: The initial survey included data of 1185 patients, and an additional 180 patients were surveyed after protocol optimization. The differences between the mean values of the dose distributions from each scanner were statistically significant (p < 0.05) for all examinations. The variation was greatest for low-dose chest CT, with a greater than fivefold difference in the mean dose values noted between scanners. A very weak correlation was found between dose and scanner age or the number of detector rows. Analysis of image quality revealed no statistically significant differences in any scoring categories, with the exception of the noise category in abdominal imaging. Implementation of the DRLs allowed a reduction in patient dose of up to 41% as a result of a protocol change. CONCLUSION: Establishing provincial DRLs allows an effective reduction in patient dose without resulting in degradation of image quality.

Secular pattern of aneurismal rupture with the lunar cycle and season.

Background The lunar cycle and seasons may be associated with rates of rupture of intracranial aneurysms, but the literature is mixed. Studies of the association between the lunar cycle and rates of aneurysm rupture used the eight qualitative moon phases. The purpose of this study was to assess any association of aneurysm rupture with the lunar cycle and with the season. Materials and methods We retrospectively reviewed all cases of subarachnoid haemorrhage secondary to ruptured intracranial aneurysm treated with endovascular coiling in our institution over a 10-year period. We included only cases with a known rupture date. We used the degree of illumination of the moon to quantitatively code the lunar cycle. Results A total of 212 cases were included in our analyses. The odds of aneurysm rupture were significantly greater ( p < 0.001) when the moon was least (new moon) and most (full moon) illuminated, as compared to the middle of the lunar cycle. The odds of rupture tended to be higher ( p = 0.059) in the summer, compared to autumn. Conclusions The odds of aneurysm rupture were greater when the moon was least illuminated (new moon) and most illuminated (full moon), compared to the middle of the lunar cycle.

The Trio Trial - A Randomized Controlled Clinical Trial Evaluating the Effect of a Biocompatible Peritoneal Dialysis Solution on Residual Renal Function.

UNLABELLED: ♦ BACKGROUND AND OBJECTIVE: Residual renal function (RRF) correlates with mortality and morbidity rates in patients receiving peritoneal dialysis (PD). We examined the effect of a biocompatible PD solution (Gambrosol Trio; Gambro Lundia AB, Lund, Sweden) with lower concentrations of glucose degradation products on rates of decline in RRF. ♦ DESIGN, SETTING, PARTICIPANTS, AND MEASUREMENTS: Incident patients at 2 centers in Canada and 1 in Hong Kong were randomized (by minimization) in an open-label parallel group trial to receive Gambrosol Trio or standard PD solution (Dianeal; Baxter Healthcare, Mississauga, Canada) for 2 years. Primary outcome was slope of RRF. Secondary outcomes were urine volumes, fluid and nutrition indices, PD and membrane characteristics, peritonitis rates, adverse events, and PD technique survival. ♦ RESULTS: Residual renal function declined by 0.132 mL/minute/1.73 m(2)/month in 51 patients allocated to biocompatible, and 0.174 mL/minute/1.73 m(2)/month in 50 patients allocated to standard PD solution (difference 0.042 mL/minute/1.73 m(2)/month, p = 0.001). Urine volume, body mass index, normalized protein catabolic rates, and fat mass were higher; total body water, peritoneal ultrafiltration, and D/P creatinine did not differ; and serum phosphate, rates of icodextrin, and automated cycler use were lower with Gambrosol Trio use. There were more peritonitis events with Gambrosol Trio use, while PD technique survival did not differ between groups. ♦ CONCLUSIONS: The use of the biocompatible PD solution Gambrosol Trio was associated with slower rates of decline in RRF, fluid and nutrition benefits, and increased peritonitis rates. TRIAL NUMBER: ISRCTN26252543.

Predicting EGFR mutation status in lung cancer:Proposal for a scoring model using imaging and demographic characteristics.

OBJECTIVE: To determine if a combination of CT and demographic features can predict EGFR mutation status in bronchogenic carcinoma. METHODS: We reviewed demographic and CT features for patients with molecular profiling for resected non-small cell lung carcinoma. Using multivariate logistic regression, we identified features predictive of EGFR mutation. Prognostic factors identified from the logistic regression model were then used to build a more practical scoring system. RESULTS: A scoring system awarding 5 points for no or minimal smoking history, 3 points for tumours with ground glass component, 3 points for airbronchograms, 2 points for absence of preoperative evidence of nodal enlargement or metastases and 1 point for doubling time of more than a year, resulted in an AUROC of 0.861. A total score of at least 8 yielded a specificity of 95 %. On multivariate analysis sex was not found to be predictor of EGFR status. CONCLUSIONS: A weighted scoring system combining imaging and demographic data holds promise as a predictor of EGFR status. Further studies are necessary to determine reproducibility in other patient groups. A predictive score may help determine which patients would benefit from molecular profiling and may help inform treatment decisions when molecular profiling is not possible. KEY POINTS: • EGFR mutation-targeted chemotherapy for bronchogenic carcinoma has a high success rate. • Mutation testing is not possible in all patients. • EGFR associations include subsolid density, slow tumour growth and minimal/no smoking history. • Demographic or imaging features alone are weak predictors of EGFR status. • A scoring system, using imaging and demographic features, is more predictive.

Consistency of visual assessments of mammographic breast density from vendor-specific "for presentation" images.

Discussions of percent breast density (PD) and breast cancer risk implicitly assume that visual assessments of PD are comparable between vendors despite differences in technology and display algorithms. This study examines the extent to which visual assessments of PD differ between mammograms acquired from two vendors. Pairs of "for presentation" digital mammography images were obtained from two mammography units for 146 women who had a screening mammogram on one vendor unit followed by a diagnostic mammogram on a different vendor unit. Four radiologists independently visually assessed PD from single left mediolateral oblique view images from the two vendors. Analysis of variance, intra-class correlation coefficients (ICC), scatter plots, and Bland-Altman plots were used to evaluate PD assessments between vendors. The mean radiologist PD for each image was used as a consensus PD measure. Overall agreement of the PD assessments was excellent between the two vendors with an ICC of 0.95 (95% confidence interval: 0.93 to 0.97). Bland-Altman plots demonstrated narrow upper and lower limits of agreement between the vendors with only a small bias (2.3 percentage points). The results of this study support the assumption that visual assessment of PD is consistent across mammography vendors despite vendor-specific appearances of "for presentation" images.