Weakly-supervised deep learning for ultrasound diagnosis of breast cancer.

Conventional deep learning (DL) algorithm requires full supervision of annotating the region of interest (ROI) that is laborious and often biased. We aimed to develop a weakly-supervised DL algorithm that diagnosis breast cancer at ultrasound without image annotation. Weakly-supervised DL algorithms were implemented with three networks (VGG16, ResNet34, and GoogLeNet) and trained using 1000 unannotated US images (500 benign and 500 malignant masses). Two sets of 200 images (100 benign and 100 malignant masses) were used for internal and external validation sets. For comparison with fully-supervised algorithms, ROI annotation was performed manually and automatically. Diagnostic performances were calculated as the area under the receiver operating characteristic curve (AUC). Using the class activation map, we determined how accurately the weakly-supervised DL algorithms localized the breast masses. For internal validation sets, the weakly-supervised DL algorithms achieved excellent diagnostic performances, with AUC values of 0.92-0.96, which were not statistically different (all Ps > 0.05) from those of fully-supervised DL algorithms with either manual or automated ROI annotation (AUC, 0.92-0.96). For external validation sets, the weakly-supervised DL algorithms achieved AUC values of 0.86-0.90, which were not statistically different (Ps > 0.05) or higher (P = 0.04, VGG16 with automated ROI annotation) from those of fully-supervised DL algorithms (AUC, 0.84-0.92). In internal and external validation sets, weakly-supervised algorithms could localize 100% of malignant masses, except for ResNet34 (98%). The weakly-supervised DL algorithms developed in the present study were feasible for US diagnosis of breast cancer with well-performing localization and differential diagnosis.

Effect of Different Types of Mammography Equipment on Screening Outcomes: A Report by the Alliance for Breast Cancer Screening in Korea.

OBJECTIVE: To investigate the effects of different types of mammography equipment on screening outcomes by comparing the performance of film-screen mammography (FSM), computed radiography mammography (CRM), and digital mammography (DM). MATERIALS AND METHODS: We retrospectively enrolled 128756 sets of mammograms from 10 hospitals participating in the Alliance for Breast Cancer Screening in Korea between 2005 and 2010. We compared the diagnostic accuracy of the types of mammography equipment by analyzing the area under the receiver operating characteristic curve (AUC) with a 95% confidence interval (CI); performance indicators, including recall rate, cancer detection rate (CDR), positive predictive value1 (PPV1), sensitivity, specificity, and interval cancer rate (ICR); and the types of breast cancer pathology. RESULTS: The AUCs were 0.898 (95% CI, 0.878-0.919) in DM, 0.860 (0.815-0.905) in FSM, and 0.866 (0.828-0.903) in CRM (p = 0.150). DM showed better performance than FSM and CRM in terms of the recall rate (14.8 vs. 24.8 and 19.8%), CDR (3.4 vs. 2.2 and 2.1 per 1000 examinations), PPV1 (2.3 vs. 0.9 and 1.1%), and specificity (85.5 vs. 75.3 and 80.3%) (p < 0.001) but not in terms of sensitivity (86.3 vs. 87.4 and 86.3%) and ICR (0.6 vs. 0.4 and 0.4). The proportions of carcinoma in situ (CIS) were 27.5%, 13.6%, and 11.8% for DM, CRM, and FSM, respectively (p = 0.003). CONCLUSION: In comparison to FSM and CRM, DM showed better performance in terms of the recall rate, CDR, PPV1, and specificity, although the AUCs were similar, and more CISs were detected using DM. The application of DM may help to improve the quality of mammography screenings. However, the overdiagnosis issue of CIS using DM should be evaluated.

Interobserver agreement in breast ultrasound categorization in the Mammography and Ultrasonography Study for Breast Cancer Screening Effectiveness (MUST-BE) trial: results of a preliminary study.

PURPOSE: The purpose of this study was to record and evaluate interobserver agreement as quality control for the modified categorization of screening breast ultrasound developed by the Alliance for Breast Cancer Screening in Korea (ABCS-K) for the Mammography and Ultrasonography Study for Breast Cancer Screening Effectiveness (MUST-BE) trial. METHODS: Eight breast radiologists with 4-16 years of experience participated in 2 rounds of quality control testing for the MUST-BE trial. Two investigators randomly selected 125 and 100 cases of breast lesions with different ratios of malignant and benign lesions. Two versions of the modified categorization were tested. The initially modified classification was developed after the first quality control workshop, and the re-modified classification was developed after the second workshop. The re-modified categorization established by ABCS-K added size criteria and the anterior-posterior ratio compared with the initially modified classification. After a brief lecture on the modified categorization system prior to each quality control test, the eight radiologists independently categorized the lesions using the modified categorization. Interobserver agreement was measured using kappa statistics. RESULTS: The overall kappa values for the modified categorizations indicated moderate to substantial degrees of agreement (initially modified categorization and re-modified categorization: κ=0.52 and κ=0.63, respectively). The kappa values for the subcategories of category 4 were 0.37 (95% confidence interval [CI], 0.24 to 0.52) and 0.39 (95% CI, 0.31 to 0.49), respectively. The overall kappa values for both the initially modified categorization and the re-modified categorization indicated a substantial degree of agreement when dichotomizing the interpretation as benign or suspicious. CONCLUSION: The preliminary results demonstrated acceptable interobserver agreement for the modified categorization.

Diagnostic Usefulness of Combination of Diffusion-weighted Imaging and T2WI, Including Apparent Diffusion Coefficient in Breast Lesions: Assessment of Histologic Grade.

PURPOSE: This study aimed to compare the diagnostic values of a combination of diffusion-weighted imaging and T2-weighted imaging (DWI-T2WI) with dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), and to evaluate the correlation of DWI with the histologic grade in breast cancer. MATERIALS AND METHODS: This study evaluated a total of 169 breast lesions from 136 patients who underwent both DCE-MRI and DWI (b value, 1000s/mm2). Morphologic and kinetic analyses for DCE-MRI were classified according to the Breast Imaging-Reporting and Data System. For the DWI-T2WI set, a DWI-T2WI score for lesion characterization that compared signal intensity of DWI and T2WI (benign: DWI-T2WI score of 1, 2; malignant: DWI-T2WI score of 3, 4, 5) was used. The diagnostic values of DCE-MRI, DWI-T2WI set, and combined assessment of DCE and DWI-T2WI were calculated. RESULTS: Of 169 breast lesions, 48 were benign and 121 were malignant (89 invasive ductal carcinoma, 24 ductal carcinoma in situ, 4 invasive lobular carcinoma, 4 mucinous carcinoma). The mean apparent diffusion coefficient (ADC) of invasive ductal carcinoma (0.92 ± 0.19 × 10-3 mm2/s) and ductal carcinoma in situ (1.11 ± 0.13 × 10-3 mm2/s) was significantly lower than the value seen in benign lesions (1.36 ± 0.22 × 10-3 mm2/s). The specificity, positive predictive value (PPV), and accuracy of DWI-T2WI set and combined assessment of DCE and DWI-T2WI (specificity, 87.5% and 91.7%; PPV, 94.3% and 96.2%; accuracy, Az = 0.876 and 0.922) were significantly higher than those of the DCE-MRI (specificity, 45.8%; PPV, 81.7%; accuracy, Az = 0.854; P < .05). A low ADC value and the presence of rim enhancement were associated with a higher histologic grade cancer (P < .05). CONCLUSION: Combining DWI, T2WI, and ADC values provides increased accuracy for differentiation between benign and malignant lesions, compared with DCE-MRI. A lower ADC value was associated with a higher histologic grade cancer.

Analysis of Participant Factors That Affect the Diagnostic Performance of Screening Mammography: A Report of the Alliance for Breast Cancer Screening in Korea.

OBJECTIVE: To analyze participant factors that affect the diagnostic performance of screening mammography. MATERIALS AND METHODS: We enrolled 128756 cases from 10 hospitals between 2005 and 2010. We analyzed recall rate, cancer detection rate (CDR) per 1000 examinations, positive predictive value (PPV), sensitivity, specificity, false positive rate (FPR), and interval cancer rate (ICR) per 1000 negative examinations according to participant factors including age, breast density, and number of visit to the same institution, and adjusted for confounding variables. RESULTS: Increasing age improved recall rates (27.4% in 40's, 17.5% in 50's, 11.1% in 60's, and 8.6% in 70's), CDR (2.7, 3.2, 2.0, and 2.4), PPV (1.0, 1.8, 1.8, and 2.8%), sensitivity (81.3, 88.8, 90.3, and 94.7%), specificity (72.7, 82.7, 89.0, and 91.7%), and FPR (27.3, 17.3, 11.0, and 8.4%) (p < 0.05). Higher breast density impaired recall rates (4.0% in P1, 9.0% in P2, 28.9% in P3, and 27.8% in P4), PPV (3.3, 2.3, 1.2, and 1.3%), specificity (96.1, 91.2, 71.4, and 72.5%), and FPR (3.9, 8.9, 28.6, and 27.6%) (p < 0.001). It also increased CDR (1.3, 2.1, 3.3, and 3.6) and ICR (0.2, 0.3, 0.6, and 1.6) (p < 0.05). Successive visits to the same institution improved recall rates (20.9% for one visit, 10.7% for two visits, 7.7% for more than three visits), PPV (1.6, 2.8, and 2.7%), specificity (79.4, 89.6, and 92.5%), and FPR (20.6, 10.4, and 7.5%) (p < 0.001). CONCLUSION: Young age and dense breasts negatively affected diagnostic performance in mammography screening, whereas successive visits to the same institution had a positive effect. Examinee education for successive visits to the same institution would improve the diagnostic performance.

Potential Use of American College of Radiology BI-RADS Mammography Atlas for Reporting and Assessing Lesions Detected on Dedicated Breast CT Imaging: Preliminary Study.

RATIONALE AND OBJECTIVES: Dedicated breast computed tomography (DBCT) is an emerging and promising modality for breast lesions. The objective of this study was to evaluate the potential use of applying the BI-RADS Mammography Atlas 5th Edition for reporting and assessing breast lesions on DBCT. Currently, no atlas exists for DBCT. MATERIALS AND METHODS: Four radiologists trained in breast imaging were recruited in this institutional review board-approved, Health Insurance Portability and Accountability Act-compliant study. The enrolled radiologists, who were blinded to mammographic and histopathologic findings, individually reviewed 30 randomized DBCT cases that contained marked lesions. Thirty-four lesions were included in this study: 24 (70.6%) masses, 7 (20.6%) calcifications, and 3 (8.8%) architectural distortions. Eight (23.5%) lesions were malignant and 26 (76.5%) were benign. The reader was asked to specify according to the BI-RADS Mammography Atlas for each marked DBCT lesion: primary findings, features, breast density, and final assessment. We calculated readers' diagnostic performances for differentiating between benign and malignant lesions and interobserver variability for reporting and assessing lesions using a generalized estimating equation and the Fleiss kappa (κ) statistic. RESULTS: The estimated overall sensitivity of the readers was 0.969, and the specificity was 0.529. There were no significant differences in the sensitivity and the specificity between lesion types. For reporting the presence of a primary finding, the overall substantial agreement (κ = 0.70) was seen. In assigning the breast density and the final assessment, the overall agreement was moderate (κ = 0.53) and fair (κ = 0.30). CONCLUSION: The use of the BI-RADS Mammography Atlas 5th Edition for DBCT showed high performance and good agreement among readers.

Performance of Screening Mammography: A Report of the Alliance for Breast Cancer Screening in Korea.

OBJECTIVE: To analyze the diagnostic accuracy and trend in screening mammography in Korea. MATERIALS AND METHODS: We retrospectively linked the information from hospitals participating in the Alliance of Breast Cancer Screening in Korea (ABCS-K) and the database of the National Cancer Screening Program. We calculated performance indicators, including the recall rate, cancer detection rate (CDR), positive predictive value (PPV), sensitivity, specificity, false-positive rate (FPR), and interval cancer rate (ICR). Changes in the performance indicators were calculated as the annual percent change with 95% confidence interval (CI). RESULTS: We enrolled 128756 cases from 10 hospitals from 2005 to 2010. The recall rate was 19.1% with a downward trend over time (-12.1% per year; 95% CI, -15.9 to -8.2). The CDR was 2.69 per 1000 examinations, without a significant trend. The PPV was 1.4% with an upward trend (20.8% per year; 95% CI, 15.2 to 26.7). The sensitivity was 86.5% without a significant trend, whereas the specificity was 81.1% with an upward trend (3.3% per year; 95% CI, 2.1 to 4.5). The FPR was 18.9% with a downward trend (-12.4% per year; 95% CI, -16.2 to -8.4). The ICR was 0.5 per 1000 negative examinations without a significant trend. There were institutional variations in the diagnostic accuracy and trend except for the CDR, sensitivity, and ICR. CONCLUSION: The sensitivity and CDR of screening mammography in the ABCS-K from 2005 to 2010 were compatible with those for Western women. The recall rate, PPV and specificity, however, were suboptimal, although they showed significant improvements over this period. A further analysis is required to explain institutional variations.

Image presentation. The nipple-areolar complex: a pictorial review of common and uncommon conditions.

OBJECTIVE: The purpose of this presentation is to show the radiologic findings of normal variants and benign and malignant diseases that affect the nipple-areolar complex. METHODS: We evaluated the imaging findings of nipple-areolar complex lesions, using multiple breast imaging modalities including mammography, sonography, galactography, contrast-enhanced magnetic resonance imaging (MRI), and positron emission tomography/computed tomography. RESULTS: Radiologic features of nipple-areolar complex lesions, including Montgomery tubercles, nipple inversion, benign calcifications, inflammation, duct dilatations, intraductal papillomas, fibroadenomas, neurofibromatosis, dermatosis of the nipple, and breast malignancy, have been illustrated. CONCLUSIONS: A clinical examination is essential and an appropriate imaging evaluation with multiple modalities is often necessary to accurately diagnose an underlying abnormality of the nipple-areolar complex. Given the limitations of conventional mammography, supplemental mammographic views often are needed, and sonography may be performed to further characterize a mammographic or clinical finding. Also, contrast-enhanced MRI may be useful for additional evaluation.

Acute hypertensive pulmonary edema after Cesarean section in a patient with an antepartum myocardial infarction -A case report-.

We report a case of 29-year-old, morbidly obese, diabetic primigravida who had undergone previously primary percutaneous coronary intervention with stent placement for an inferior wall myocardial infarction at 10 weeks of gestation. She remained asymptomatic with medication during the remainder of her pregnancy, but preoperative echocardiography revealed left ventricular dilation and a restrictive diastolic dysfunction with a preserved ejection fraction (46%). She developed acute pulmonary edema associated with hypertension after an elective Cesarean delivery under continuous epidural anesthesia despite the meticulous restriction of fluid.

Prevalence and significance of carotid plaques in patients with coronary atherosclerosis.

BACKGROUND AND OBJECTIVES: Carotid artery intima-media thickness (CIMT) has recently been recommended as a non-invasive tool for primary prevention of cardiovascular events; the association between CIMT and adverse cardiovascular events is well-known. We sought to evaluate the prevalence and significance of carotid artery plaque, especially in patients with coronary atherosclerosis. SUBJECTS AND METHODS: The study population consisted of 1,705 consecutive patients {933 males (54.7%); mean age, 59.7+/-10.9 years} who underwent coronary angiography and carotid artery scanning using high-resolution ultrasonography. Carotid plaque was defined as a focal structure encroaching into the arterial lumen by at least 50% of the surrounding IMT value or a thickness >1.2 mm. RESULTS: Carotid plaque was identified in 30.3% (516/1,705) of the patients. Of patients in whom the plaque location could be evaluated (n=1,027), carotid plaque was located at the common carotid artery {n=64/267 (24.0%)}, carotid bulb {n=194/267 (72.7%)}, and at both sites {n=9/267 (3.4%)}. The prevalence of hypertension (58.5% vs. 45.2%, p<0.001) and diabetes mellitus (30.6% vs. 23.5%, p=0.007) was higher in patients with carotid plaques. The patients with carotid plaques were older (65.4+/-8.9 years vs. 57.2+/-10.7 years, p<0.0001), had a thicker CIMT (0.89+/-0.20 mm vs. 0.77+/-0.16 mm, p<0.001), and higher fasting blood sugar (FBS) levels (132.1+/-60.7 mg/dL vs. 121.6+/-47.1 mg/dL, p<0.001) than those without carotid plaque. Patients with carotid plaque more frequently presented with acute coronary syndrome (32.4% vs. 23.9%, p<0.001) than those without carotid plaque. Significant coronary artery stenosis by coronary angiography (75.4% vs. 58.3%, p<0.001), especially multi-vessel disease (46.3% vs. 27.2%, p<0.001), was more frequent in patients with carotid plaques. On multivariate analysis, old age (>/=65 years), hypertension, and increased CIMT (>/=1.0 mm) were independent predictors of carotid plaque. Carotid plaque (odds ratio, 1.85; 95% confidence interval, 1.39-2.45; p<0.001) was an independent predictor of multivessel disease based on multivariate regression analysis. CONCLUSION: Carotid plaque was common (30.3%) in Korean patients with coronary atherosclerosis, but it is still relatively uncommon compared to Western populations. Carotid plaque was associated with old age, hypertension, and increased IMT, and was an independent predictor of multi-vessel disease.

Percutaneous treatment of extrahepatic bile duct stones assisted by balloon sphincteroplasty and occlusion balloon.

OBJECTIVE: To describe the technical feasibility and usefulness of extrahepatic biliary stone removal by balloon sphincteroplasty and occlusion balloon pushing. MATERIALS AND METHODS: Fifteen patients with extrahepatic bile duct stones were included in this study. Endoscopic stone removal was not successful in 13 patients, and two patients refused the procedure due to endoscopy phobia. At first, all patients underwent percutaneous transhepatic biliary drainage (PTBD). A few days later, through the PTBD route, balloon assisted dilatation for common bile duct (CBD) sphincter was performed, and then the stones were pushed into the duodenum using an 11.5 mm occlusion balloon. Success rate, reason for failure, and complications associated with the procedure were evaluated. RESULTS: Eight patients had one stone, five patients had two stones, and two patients had more than five stones. The procedure was successful in 13 patients (13/15). In 12 of the patients, all stones were removed in the first trial. In one patient, residual stones were discovered on follow-up cholangiography, and were subsequently removed in the second trial. Technical failure occurred in two patients. Both of these patients had severely dilated CBD and multiple stones with various sizes. Ten patients complained of pain in the right upper quadrant and epigastrium of the abdomen immediately following the procedure, but there were no significant procedure-related complications such as bleeding or pancreatitis. CONCLUSION: Percutaneous extrahepatic biliary stone removal by balloon sphincteroplasty and subsequent stone pushing with occlusion balloon is an effective, safe, and technically feasible procedure which can be used as an alternative method in patients when endoscopic extrahepatic biliary stone removal was not successful.