Coronary artery segmentation under class imbalance using a U-Net based architecture on computed tomography angiography images.

Coronary artery disease is caused primarily by vessel narrowing. Extraction of the coronary artery area from images is the preferred procedure for diagnosing coronary diseases. In this study, a U-Net-based network architecture, 3D Dense-U-Net, was adopted to perform fully automatic segmentation of the coronary artery. The network was applied to 474 coronary computed tomography (CT) angiography scans performed at Wanfang Hospital, Taiwan. Of these, 10% were used for testing. The CT scans were divided into patches of 16 original high-resolution slices. The slices were overlapped between patches to take advantage of surrounding imaging information. However, an imbalance between the foreground and background presents a challenge in smaller-object segmentation such as with coronary arteries. The network was optimized and achieved a promising result when the focal loss concept was adopted. To evaluate the accuracy of the automatic segmentation approach, the dice similarity coefficient (DSC) was calculated, and an existing clinical tool was used. The subjective ratings of three experienced radiologists were used to compare the two ratings. The results show that the proposed approach can achieve a DSC of 0.9691, which is significantly higher than other studies using a deep learning approach. In the main trunk, the results of automatic segmentation agree with those of the clinical tool; they were significantly better in some small branches. In our study, automatic segmentation tool shows high-performance detection in coronary lumen vessels, thereby providing potential power in assisting clinical diagnosis.

Characterization and management of various renal cystic lesions by sonographic features.

Renal cysts are common incidental findings in clinical practice. Most renal cysts detected in medical imaging are benign simple cysts. However, some are complicated by hemorrhage or infection or are associated with calcification. In these instances, difficulties can be encountered distinguishing the complicated cysts from cystic renal tumors such as cystic renal cell carcinoma, multilocular cystic nephroma, and mixed epithelial and stromal tumors. The Bosniak classification is widely used to categorize cystic renal lesions but is used to classify those discovered via computed tomography. Ultrasonography (US) and color Doppler US are the most frequently used imaging techniques for abdominal surveys and long-term follow-up because of their noninvasiveness, relatively low cost, wide availability, and frequently, lack of contrast medium. Herein, we review the features of various cystic lesions of the kidney that can be found using US, discuss differential diagnoses using US, and propose a feature-oriented algorithmic approach to classifying renal cystic lesions using US.

Large-volume iodinated contrast medium extravasation: low frequency and good outcome after conservative management in a single-centre cohort of more than 67,000 patients.

OBJECTIVES: Our aim was to retrospectively investigate the frequency and outcome of large-volume iodinated contrast medium (CM) extravasation in our institution and to compare our management protocol to current practice. METHODS: Institutional review board approval was obtained, and informed consent was waived because the study was retrospective. From January 2008 to September 2016, radiological examinations with intravenous non-ionic iodinated CM administration were performed in 67,129 patients. Contrast medium extravasation events on CT scans and intravenous pyelograms but not on angiograms were included. All data were collected prospectively and stratified according to age, injection method (manual vs auto-injection), prevention of extravasation by various means (including intercom alarm), management of extravasation (routine application of silver sulfadiazine ointment, clobetasol propionate cream, and damp gauze at room temperature), etc. RESULTS: The incidence of large-volume CM extravasation was very low (0.04% [27/67,129] overall; 0.03% related to manual injection [age range, 59-92 years; mean, 75.4 years], and 0.045% related to auto-injection [age range, 36-86 years; mean, 65.8 years]). The CM extravasation volume in majority of patients was 20-40 ml in 5 of 9 patients (55.6%) in the manual injection group and 14 of 18 (77.8%) in the auto-injection group. Swelling and pain were the most common symptoms. No patient developed severe signs or needed surgical intervention. CONCLUSIONS: Results show a very low incidence of large-volume CM extravasation without severe complications or sequelae. The casual effect between our protocols and good outcome cannot be scrutinised thoroughly because the study lacks a control group and is retrospective. KEY POINTS: • The incidence of large-volume contrast medium extravasation (≥20 ml) was 0.04%. • No patient needed surgical intervention, and most recovered within 7 days. • Each element of our management protocol contributed to good outcome.

The proper use of coronary calcium score and coronary computed tomography angiography for screening asymptomatic patients with cardiovascular risk factors.

Early detection and treatment of coronary artery disease (CAD) can reduce incidences of acute myocardial infarction. In this study, we determined the proper use of contributing risk factors and coronary artery calcium score (CACS) when screening asymptomatic patients with coronary arterial stenoses using coronary computed tomography angiography (CCTA). We reviewed 934 consecutive patients who received CACS and CCTA between December 2013 and November 2016. At least one cardiovascular disease risk factor was present in each of the 509 asymptomatic participants. Patients were grouped based on CACS into "zero," "minimal" (0 < CACS ≤ 10), "mild" (10 < CACS ≤ 100), "moderate" (100 < CACS ≤ 400), and "excessive" (CACS > 400). Males over 45 years old with diabetes mellitus and hypertension had a higher risk of significant coronary stenosis. In multivariate analysis, age, sex, hypertension, and diabetes mellitus remained significant predictors of stenosis. A CACS of zero occurred in 227 patients (44.6%). There were no significant differences between the "zero" and "minimal" groups (p = 0.421), but the "mild," "moderate," and "excessive" groups showed correlations with significant coronary stenosis. Age, sex, diabetes mellitus, and hypertension were associated with higher risk of significant coronary stenosis. Asymptomatic patients with CACSs of zero do not require CCTA, and thereby avoid unnecessary radiation exposure.

Assessment of radiation safety in cardiac CT angiography.

This study investigates the radiation dose and image quality of patients not receiving ß-blockers for cardiac CT angiography (CCTA) with or without the optimization of electrocardiographic (ECG) pulsing window. The differences in patient characteristics are also characterized.Normal-weight and obese patients (n = 154) with heart rates between 65 and 80 beats per minutes (bpm) during the prospective axial scanning were enrolled retrospectively. The ECG pulsing windows were set at 50% to 75% (Group A) or 60% to 75% (Group B) of the R-R interval for patients with heart rate variability higher than or not exceeding ±5 bpm, respectively. The effective doses of individual patient were estimated from the dose length product of the CCTA scan. Two radiologists independently reviewed the images and applied a 4-point Likert scale for image quality assessment. The patients' characteristics were compared along with the patients' effective doses between groups.The optimized pulsing window significantly reduced the average radiation dose for normal-weight and obese patients by 33% and 27%, respectively. The CCTA image quality of patients in Group A was not different overall from those obtained from Group B. Nondiabetic obese patients were more likely to be accepted for the use of the optimized pulsing window. Unlike obese patients, normal-weight patients revealed no characteristic difference between Groups A and B.This study indicates an equivalent efficacy of using optimized pulsing windows for reducing the radiation dose for patients without ß-blocker administration between different body weight groups. Nevertheless, gender and diabetic status became prominent characteristics in the obese group when matching up with the optimized pulsing window.

Sonographic criteria predictive of benign thyroid nodules useful in avoiding unnecessary ultrasound-guided fine needle aspiration.

BACKGROUND/PURPOSE: There has been no individual ultrasound feature of having high accuracy for diagnosis of thyroid malignancy. In this study, we aimed to establish feature-oriented criteria to characterize benign thyroid nodules that do not require ultrasound (US)-guided fine needle aspiration (FNA). METHODS: We reviewed 374 patients with thyroid nodules who had undergone US-guided FNA at our institution (2005-2008). Thyroid nodules were classified into two groups: Category 1 (benign nodules that required follow-up 6-12 months later but not US-guided FNA); and Category 2 (indeterminate nodules or suspected carcinoma that required US-guided FNA). To test the validity, we reviewed 315 consecutive patients who had histologically proven thyroid carcinoma (n = 39) and randomly selected 40 of the 276 patients with benign nodules (2009-2010). RESULTS: Of 374 nodules, 354 (95%) were benign and 20 (5%) malignant. On US, 260 nodules had no calcification, no increase in vascularity, well-defined margin, and no lymphadenopathy (Category 1). Using a combination of these four features, we were able to discriminate benign from indeterminate nodules or suspected malignant nodules with a sensitivity of 73%, and specificity and positive predictive value of 100%. Validity testing revealed that none of the 39 malignant thyroid nodules had all four US features. All Category 1 nodules (2005-2008) remained benign at the 3-years follow up. CONCLUSION: The combination of four US features of Category 1 nodules is highly predictive of benign disease, and we could avoid unnecessary US-guided FNA in 69.5% of our patients using this combined features.