Search

The receiver operating characteristic (ROC) curve is a useful way to display the performance of a medical diagnostic test for detecting whether or not a patient is diseased or healthy. The diagnostic data consist of independent random samples on continuous measurement scales from diseased and healthy populations. We propose assessing the goodness-of-fit of a model by comparing a model-based estimate with a nonparametric estimate of the area under the curve (AUC). We focus on two parametric models, so-called Bi-Normal and Bi-Weibull models, and briefly on associated semiparametric transformation models. We also consider the null hypothesis that a parametric model is valid after an unspecified monotone transformation of the measurement scales. High power of the test implies sensitivity of the AUC to model assumptions; low power implies robustness of the estimate. The test is exemplified with a data set on the diagnosis of pancreatic cancer. A simulation study of the statistical power of the test is included.

Background: We compared nonenhanced and dynamic gadolinium (Gd)–enhanced magnetic resonance imaging (MRI) appearances of hepatic focal nodular hyperplasia (FNH) as depicted with breath-hold MR sequences and assessed the detectability of the individual MR sequences used. Methods: We retrospectively reviewed 48 consecutive patients with FNH. All patients underwent nonenhanced (T1 fast low-angle shot [FLASH] and T2 half-Fourier acquisition [HASTE]) and dynamic Gd-enhanced (T1 FLASH) MRI between December 1997 and March 2000. Individual MR sequences were analyzed separately for number of lesions, signal intensity features, dynamic enhancement pattern, and the presence and enhancement profile of a central scar. Ninety-five percent confidence intervals of absolute discrepancy were calculated to define differences in lesion detection. Results: Seventy-seven lesions were found in 48 patients. Nonenhanced FLASH imaging depicted 59 (76.6%) lesions in 45 patients. HASTE images showed 55 (71.4%) lesions in 44 patients. On T1- and T2-weighted images, lesions appeared predominantly hypointense (69.5%) and hyperintense (72.7%), respectively. Arterial and portal venous dominant phase Gd-enhanced MRI demonstrated all 77 lesions (100%), most of which showed hypervascular (94.8%), homogeneous (97.4%), and incomplete (except the central scar: 58.4%) enhancement in the arterial phase. Portal venous phase images showed lesion isointensity (50.6%) or moderate hyperintensity (46.8%) with complete enhancement (central scar: 94.8%). A central scar was detected on nonenhanced T1-weighted images (hypointense: 100%), T2-weighted images (hyperintense: 100%), arterial phase (hypointense: 59.7%) and portal venous phase (hyperintense: 71.4%) Gd-enhanced images in 78%, 69.1%, 77.9%, and 75.3% of tumors, respectively. Conclusion: Arterial and portal venous phase Gd-enhanced T1-weighted sequences are superior to nonenhanced images in the detection of FNH. Typical MRI appearances include hypointensity on T1-weighted and hyperintensity on nonenhanced T2-weighted images. Most commonly, FNH shows a homogeneous (without scar) and strong enhancement during the arterial phase, with lesion isointensity or slight hyperintensity during the portal venous phase.

Purpose: To assess the reproducibility and accuracy compared to radiologists of three automated segmentation pipelines for quantitative magnetic resonance imaging (MRI) measurement of brain white matter signal abnormalities (WMSA). Materials and Methods: WMSA segmentation was performed on pairs of whole brain scans from 20 patients with multiple sclerosis (MS) and 10 older subjects who were positioned and imaged twice within 30 minutes. Radiologist outlines of WMSA on 20 sections from 16 patients were compared with the corresponding results of each segmentation method. Results: The segmentation method combining expectationmaximization (EM) tissue segmentation, template-driven segmentation (TDS), and partial volume effect correction (PVEC) demonstrated the highest accuracy (the absolute value of the Z-score was 0.99 for both groups of subjects), as well as high interscan reproducibility (repeatability coefficient was 0.68 mL in MS patients and 1.49 mL in aging subjects). Conclusion: The addition of TDS to the EM segmentation and PVEC algorithms significantly improved the accuracy of WMSA volume measurements, while also improving measurement reproducibility.

The validity of segmentation is an important issue in image processing because it has a direct impact on surgical planning. Binary manual segmentation is not only time-consuming but also lacks the ability of differentiating subtle intensity variations among voxels, particularly for those on the border of a tumor and for different tumor types. Previously we have developed an automated segmentation method that yields voxel-wise continuous probabilistic measures, indicating a level of tumor presence. The goal of this work is to examine three accuracy metrics based on two-sample statistical methods, against the estimated composite latent ground truth derived from several experts' manual segmentation by a maximum likelihood algorithm. We estimated the distribution functions of the tumor and control voxel data parametrically by assuming a mixture of two beta distributions with different shape parameters. We derived the resulting receiver operating characteristic curves, Dice similarity coefficients, and mutual information, over all possible decision thresholds. Based on each validation metric, an optimal threshold was then computed via maximization. We illustrated these methods on MR imaging data from nine brain tumor cases, three with meningiomas, three astrocytomas, and three other low-grade gliomas. The automated segmentation yielded satisfactory accuracy, with varied optimal thresholds.

Characterizing the performance of image segmentation approaches has been a persistent challenge. Performance analysis is important since segmentation algorithms often have limited accuracy and precision. Interactive drawing of the desired segmentation by domain experts has often been the only acceptable approach, and yet suffers from intra-expert and inter-expert variability. Automated algorithms have been sought in order to remove the variability introduced by experts, but no single methodology for the assessment and validation of such algorithms has yet been widely adopted. The accuracy of segmentations of medical images has been difficult to quantify in the absence of a "ground truth" segmentation for clinical data. Although physical or digital phantoms can help, they have so far been unable to reproduce the full range of imaging and anatomical characteristics observed in clinical data. An attractive alternative is comparison to a collection of segmentations by experts, but the most appropriate way to compare segmentations has been unclear.We present here an Expectation-Maximization algorithm for computing a probabilistic estimate of the "ground truth" segmentation from a group of expert segmentations, and a simultaneous measure of the quality of each expert. This approach readily enables the assessment of an automated image segmentation algorithm, and direct comparison of expert and algorithm performance.

Christine L Baker,1 Kelly H Zou,1 Jun Su21Pfizer Inc., New York, NY, USA; 2Boehringer-Ingelheim Pharmaceuticals Inc., Ridgefield, CT, USABackground: Treatment of stable chronic obstructive pulmonary disease (COPD) with long-acting bronchodilator (LABD) medications is recommended by the 2014 Global initiative for chronic Obstructive Lung Disease (GOLD) guidelines. The primary objective of this study was to examine LABD prescription fills after a COPD-related hospitalization.Methods: This retrospective observational study used claims from Truven Health MarketScan® Commercial and Medicare Supplemental databases. Patients (age ≥40, commercial; age ≥65, Medicare supplemental) had a first hospitalization with a primary COPD diagnosis between April 1, 2009 and June 30, 2011 (index hospitalization) and were continuously enrolled for 1 year before and 9 months after hospitalization. Patients were categorized according to pre-index and/or post-index pharmacy claims.Results: A total of 27,738 patients had an index hospitalization and met inclusion/exclusion criteria. Of those, 19,783 patients had COPD as a primary or secondary diagnosis during the year before index hospitalization and were included in the analysis. Approximately one quarter of the patients (26.32%) did not fill a prescription for an LABD or short-acting bronchodilator both 90 days before and 90 days after hospitalization. During the 90-day pre-index period, 40.57% of patients filled an LABD (with or without a short-acting bronchodilator) prescription. Over half of the patients (56.88%) filled an LABD prescription at some point during the 180-day post-index period, but, of those, a significantly greater proportion of patients filled an LABD prescription in the 1- to 90-day post-index period than in the 91- to 180-day post-index period (51.27% versus 43.66%; P

Declining fee schedules, decreasing operating margins, and increasingly stringent compliance regulations create a need for intense scrutiny and optimization of a radiology organization's billing and collection procedures. The authors' goal was to analyze the effectiveness of departmental professional billing procedures, identify controllable factors, and intervene when they could be improved.A detailed audit of professional claims and payments was performed for all patients who underwent abdominal-pelvic computed tomography (CT) during July 1999 (n = 717). The adequacy of indication for the CT examination as given by the referring physician and modified by the radiology staff, the time required for claim generation, and the status of reimbursement within 120 days were assessed by an interdisciplinary team. After an intervention was performed to improve adequacy of the available clinical indication, the audit was repeated in December 1999 (n = 710).Despite a significant (P.05) improvement in wording of clinical indications for billing purposes between July (68%) and December (85%), there was no significant change in reimbursement against gross charges. The vast majority of claims (97% in July, 99% in December) were generated in less than 30 days. At 120 days after the date of service, payments had been received that amounted to only 66% and 54% of discounted professional fees for July and December, respectively. For examinations performed in December, payment was delayed beyond contracted time periods in 138 cases (19%).Optimum billing and collection for imaging studies is an increasingly complex task. Even when substantial efforts are devoted to eliciting the proper indication for the study, reimbursement remains low primarily because of payer delays.

Rational and Objectives It is common to administer the same diagnostic test more than once to the same set of patients. The purpose of this study was to develop two statistical methods for estimating and comparing correlated receiver operating characteristic (ROC) curves for data derived from repeated diagnostic tests. Material and Methods Parametric and semiparametric transformation models were developed. These estimation methods were illustrated with data from 72 pigmented lesions suspected of being malignant melanoma. A diagnostic scoring system based on asymmetry, border irregularity, color variation, and diameter was used repeatedly, with or without a dermoscope. Statistical hypothesis tests were conducted to evaluate whether a dermoscope improved the clarity of the lesion features in the scoring system. The resulting ROC curves were constructed, along with characteristics and summary measures. Results The areas under the ROC curves were 0.885 (parametric method) and 0.893 (semiparametric method) without the dermoscope, and 0.916 (parametric) and 0.912 (semiparametric) with the dermoscope. The statistical hypothesis tests did not yield statistically significant differences between the underlying ROC curves for either estimation method. Conclusion The two transformation models yielded similar results for estimation and comparison of the ROC curves. Although a dermoscope did not add extra information, the scoring system was accurate for diagnosing malignant melanoma.

To describe the cryoablation of liver tumors by using a percutaneous approach and intraprocedural magnetic resonance (MR) imaging monitoring and to assess the feasibility and safety of the procedure.Fifteen hepatic tumors (mean diameter, 2.9 cm) in 12 patients were treated (18 total cryoablations). Fourteen were metastases and one was a hemangioma; all were proved at biopsy. By using a 0.5-T open MR imaging system, cryoneedles were placed and lesions ablated by using real-time monitoring. Clinical signs and symptoms were assessed and laboratory tests performed. Intraprocedural depictions of iceballs were compared with contrast material-enhanced MR imaging-based estimates of cryonecrosis that were obtained 24 hours after cryoablation.MR imaging-guided percutaneous cryotherapy resulted in no serious complications and no clinically important changes in serum liver enzymes or creatinine or myoglobin levels. Intraprocedural MR imaging demonstrated iceballs as sharply marginated regions of signal loss that expanded and engulfed tumors. The maximal iceball size was 4.9 x 2.2 x 2.2 cm with the use of one cryoneedle and 6.0 x 5.6 x 4.9 cm with three cryoneedles. Intraprocedural iceball depictions correlated well with postprocedural cryonecrosis estimates.MR imaging-guided percutaneous cryotherapy of liver tumors is feasible and safe. MR imaging can be used to estimate cryotherapy effects and guide therapy intraprocedurally.

Objective: To compare MR imaging techniques with differing contrast and spatial resolution for evaluation of complete disruption of the ulnar collateral ligament (UCL) anterior bundle in a cadaveric elbow model. Design: Complete UCL tears were surgically created at the typical location for clinical tears in eight of 28 fresh frozen cadaveric elbow specimens. All specimens underwent 1.5 T MR imaging in the oblique coronal plane, using an extremity coil. The sequences employed were: T1-weighted spin echo (T1 SE), proton density-weighted (PD) fast spin echo (FSE), fat-suppressed T2-weighted FSE (T2 FSE), gradient recalled echo (GRE) with a high matrix, PD FSE with a high matrix (HRPD), and fat suppressed T1-weighted SE with intra-articular gadolinium (MRAr). Two radiologists independently graded the UCL with separate and side-by-side assessments. Results: Sensitivity/specificity pairs were as follows for reader A and reader B, respectively: T1 SE: 0.25/0.95, 0.50/0.95; PD FSE: 0.38/1.00, 0.25/1.00; T2 FSE: 0.50/0.95, 0.63/0.95; GRE: 0.63/0.85, 0.63/0.60; MRAr: 0.88/1.00, 1.00/0.80; HRPD: 0.50/1.00, 0.88/0.80. Kappa statistics for measuring interobserver reliability for each sequence were poor under T1 SE (–0.13) and GRE (0.19), moderate under HRPD (0.41) and T2 FSE (0.44) and good under MRAr (0.62) and PD FSE (0.78). For both readers, the rating for overall image quality was highest for HRPD, and the rating for UCL lesion conspicuity was the highest for MRAr. Conclusions: Of the MR imaging pulse sequences tested, MRAr showed the greatest ability to identify complete ligamentous injuries with good agreement between readers and had the highest subjective preference for lesion conspicuity. However, HRPD had the least interobserver variability and the highest subjective preference for overall image quality

One contribution of 13 to a Theme Issue ‘Mathematical and statistical methods for diagnoses and therapies’.

The purpose of this study was to determine whether unenhanced helical CT alone can be used for diagnosis and treatment planning of patients with obstructing ureteral stones.Medical records of 100 patients with ureteral stones and a clearly discernible clinical outcome who had undergone unenhanced helical CT were reviewed to determine the number of urography procedures and results of excretory urograms performed within 72 hr of helical CT. CT scans were then reviewed by two radiologists for six findings: in-plane stone diameter, z-axis stone diameter, location of stone, periureteral stranding, hydronephrosis, and perinephric fluid. Seventy-one patients passed stones spontaneously, and 29 patients required intervention including basket retrieval, extracorporeal shock-wave lithotripsy, laser lithotripsy, or a combination of the three treatments. Data were analyzed to determine those findings that correlated with the need for intervention.Five excretory urograms were obtained, all of which agreed with findings revealed by CT. Excretory urography added no information. CT findings of in-plane diameter (p.001), z -axis diameter (p.001), and location of stone (p = .003) all significantly correlated with the need for intervention.Helical CT can be used in place of excretory urography to plan treatment of patients with flank pain caused by obstructing ureteral stones. Stones that are larger than 5 mm, located within the proximal two thirds of the ureter, and seen on two or more consecutive CT images are more likely to require endoscopic removal, lithotripsy, or both.

Marco daCosta DiBonaventura1, Ryne Paulose-Ram2, Jun Su3, Margaret McDonald2, Kelly H Zou2, Jan-Samuel Wagner1, Hemal Shah31Health Sciences Practice, Kantar Health, New York, NY, USA; 2Pfizer, Inc, New York, NY, USA; 3Boehringer Ingelheim Pharmaceuticals, Inc, Ridgefield, CT, USAObjective: To examine quality of life, work productivity, and health care resource use among employed adults ages 40–64 years with chronic obstructive pulmonary disease (COPD) in the United States.Methods: Data from the 2009 National Health and Wellness Survey were used. All employed adults ages 40–64 years with or without a self-reported diagnosis of COPD were included in the study. Impact on quality of life (using the mental and physical component summary scores and health utilities from the Short Form-12v2), work productivity and activity impairment (using the Work Productivity and Activity Impairment questionnaire), and resource use were analyzed using regression modeling.Results: There were 1112 employed adults with COPD versus 18,912 employed adults without COPD. After adjusting for demographics and patient characteristics, adults with COPD reported significantly lower mean levels of mental component summary (46.8 vs 48.5), physical component summary (45.6 vs 49.2), and health utilities (0.71 vs 0.75) than adults without COPD. Workers with COPD reported significantly greater presenteeism (18.9% vs 14.3%), overall work impairment (20.5% vs 16.3%), and impairment in daily activities (23.5% vs 17.9%) than adults without COPD. Employed adults with COPD also reported more mean emergency room visits (0.21 vs 0.12) and more mean hospitalizations (0.10 vs 0.06) in the previous 6 months than employed adults without COPD. All of the above differences were significant at two-sided P < 0.05.Conclusion: After adjusting for various confounders, employed adults with COPD reported significantly lower quality of life and work productivity, and increased health care resource utilization than employed adults without COPD. These results highlight the substantial impact and burden of COPD in the United States workforce.Keywords: chronic obstructive pulmonary disease, quality of life, work productivity, activities of daily living, health care resource use

ContextThe clinical validity of using computed tomography (CT) to diagnose peripheral pulmonary embolism is uncertain. Insufficient sensitivity for peripheral pulmonary embolism is considered the principal limitation of CT.ObjectiveTo review studies that used a CT-based approach to rule out a diagnosis of pulmonary embolism.Data SourcesThe medical literature databases of PubMed, MEDLINE, EMBASE, CRISP, metaRegister of Controlled Trials, and Cochrane were searched for articles published in the English language from January 1990 to May 2004.Study SelectionWe included studies that used contrast-enhanced chest CT to rule out the diagnosis of acute pulmonary embolism, had a minimum follow-up of 3 months, and had study populations of more than 30 patients.Data ExtractionTwo reviewers independently abstracted patient demographics, frequency of venous thromboembolic events (VTEs), CT modality (single-slice CT, multidetector-row CT, or electron-beam CT), false-negative results, and deaths attributable to pulmonary embolism. To calculate the overall negative likelihood ratio (NLR) of a VTE after a negative or inconclusive chest CT scan for pulmonary embolism, we included VTEs that were objectively confirmed by an additional imaging test despite a negative or inconclusive CT scan and objectively confirmed VTEs that occurred during clinical follow-up of at least 3 months.Data SynthesisFifteen studies met the inclusion criteria and contained a total of 3500 patients who were evaluated from October 1994 through April 2002. The overall NLR of a VTE after a negative chest CT scan for pulmonary embolism was 0.07 (95% confidence interval [CI], 0.05-0.11); and the negative predictive value (NPV) was 99.1% (95% CI, 98.7%-99.5%). The NLR of a VTE after a negative single-slice spiral CT scan for pulmonary embolism was 0.08 (95% CI, 0.05-0.13); and after a negative multidetector-row CT scan, 0.15 (95% CI, 0.05-0.43). There was no difference in risk of VTEs based on CT modality used (relative risk, 1.66; 95% CI, 0.47-5.94; P = .50). The overall NLR of mortality attributable to pulmonary embolism was 0.01 (95% CI, 0.01-0.02) and the overall NPV was 99.4% (95% CI, 98.7%-99.9%).ConclusionThe clinical validity of using a CT scan to rule out pulmonary embolism is similar to that reported for conventional pulmonary angiography.