Developing a weakly supervised deep learning framework for breast cancer diagnosis with HR status based on mammography images.

The status of hormone receptors (HR) at the molecular level is crucial for accurate diagnosis and effective treatment of breast cancer. Meanwhile, mammography is an effective screening method for detecting breast cancer, which significantly improve survival. However, diagnosing the molecular status of breast cancer involves a pathological biopsy, which can affect the accuracy of the diagnosis. To non-invasively diagnose the hormone receptor (HR) status of breast cancer and reduced manual annotation, we proposed a weakly supervised deep learning framework BSNet which detected breast cancer with HR status and benign tumors. BSNet was trained on 2321 multi-view mammography cases from female undergoing digital mammography for the general population at Harbin Medical University Cancer Hospital in Heilongjiang Province during the period 2017-2018 and was validated on the external cohort. The average AUCs of BSNet on the test set and the external validation set were 0.89 and 0.92, respectively. BSNet demonstrated excellent performance in non-invasive breast cancer diagnosis with HR status, using multiple mammography views without pixel annotation. Furthermore, we developed a web server (http://bsnet.edbc.org) for easy use. BSNet described high-dimensional mammography of breast cancer subtypes, which helped inform early management options.

Evaluation of the peritumoral features using radiomics and deep learning technology in non-spiculated and noncalcified masses of the breast on mammography.

Objective: To assess the significance of peritumoral features based on deep learning in classifying non-spiculated and noncalcified masses (NSNCM) on mammography. Methods: We retrospectively screened the digital mammography data of 2254 patients who underwent surgery for breast lesions in Harbin Medical University Cancer Hospital from January to December 2018. Deep learning and radiomics models were constructed. The classification efficacy in ROI and patient levels of AUC, accuracy, sensitivity, and specificity were compared. Stratified analysis was conducted to analyze the influence of primary factors on the AUC of the deep learning model. The image filter and CAM were used to visualize the radiomics and depth features. Results: For 1298 included patients, 771 (59.4%) were benign, and 527 (40.6%) were malignant. The best model was the deep learning combined model (2 mm), in which the AUC was 0.884 (P < 0.05); especially the AUC of breast composition B reached 0.941. All the deep learning models were superior to the radiomics models (P < 0.05), and the class activation map (CAM) showed a high expression of signals around the tumor of the deep learning model. The deep learning model achieved higher AUC for large size, age >60 years, and breast composition type B (P < 0.05). Conclusion: Combining the tumoral and peritumoral features resulted in better identification of malignant NSNCM on mammography, and the performance of the deep learning model exceeded the radiomics model. Age, tumor size, and the breast composition type are essential for diagnosis.

Focus on the Predictive Value of Subclassification of Extratumoral Structural Abnormalities for Malignant Nonspiculate and Noncalcified Masses on Digital Mammography.

Purpose: To determine the independent risk factors associated with malignant nonspiculate and noncalcified masses (NSNCMs) and evaluate the predictive values of extratumoral structural abnormalities on digital mammography. Methods: A total of 435 patients were included between January and May 2018. Tumor signs included shape, density, and margin, which were evaluated. Extratumoral signs were classified into extratumoral structural abnormalities (parenchymal and trabecular) and halo; subclassification included contraction, distortion, pushing and atrophy sign of parenchyma, parallel, vertical, and reticular trabecula sign, and narrow and wide halo. Univariate and multivariate analysis was performed. The positive predictive value (PPV) of the independent predictor was calculated, and diagnostic performance was evaluated using the receiver operating characteristic curve. Results: Of all cases, 243 (55.8%) were benign and 192 (44.2%) were malignant. Extratumoral contraction sign of parenchyma was the strongest independent predictor of malignancy (odds ratio [OR] 36.2, p < 0.001; PPV = 96.6%), followed by parenchymal distortion sign (OR 10.2, p < 0.001; PPV = 92%), parallel trabecula sign (OR 7.2, p < 0.001; PPV = 85.6%), and indistinct margin of tumor (OR 4.3, p < 0.001; PPV =70.9%), and also parenchymal atrophy sign, wide halo, vertical trabecula, age ≥ 47.5 years, irregular shape, and size ≥ 22.5 mm of tumor (OR range, 1.3-4.0; PPV range, 56.6-83.6%). The diagnostic performance of most of the extratumoral signs was between that of indistinct margin and irregular shape of tumor. Conclusion: The subclassification of extratumoral structural abnormalities has important predictive value for mammographic malignant NSNCM, which should be given more attention.

Value of Low-Dose Dobutamine Stress Real-Time Myocardial Contrast Echocardiography in the Diagnosis of Coronary Heart Disease.

To investigate the value of low-dose dobutamine stress real-time myocardial contrast echocardiography (RT-MCE) in the diagnosis of coronary heart disease (CHD). A total of 65 hospitalized patients with suspected or confirmed CHD were detected by RT-MCE combined with low-dose dobutamine stress (0.84 mg/kg). Perfusion curves were quantitatively analyzed using QLAB software. Peak intensity (A), slope of curves (ß), and perfusion (A × ß) were also calculated. Based on the results of coronary angiography, patients were divided into no obvious stenosis group (<50%), mild stenosis group (50%-74%), moderate stenosis group (75%-89%), and severe stenosis group (≥90%). The A, ß, and A × ß values before and after low-dose dobutamine stress of each group were compared. In the basal state and after low-dose dobutamine stress, the A, ß, and A × ß values significantly decreased as the stenosis degree of the myocardial segments increased. The same variation tendency was also found in the A, ß, and A × ß reserve values, and there was significant difference in these reserve values between moderate and severe stenosis groups and no obvious stenosis and mild stenosis groups. Collateral circulation had marked effects on the values of myocardial perfusion parameters and their reserve values, especially in the segments with severe stenosis. Low-dose dobutamine stress RT-MCE can be a sensitive method for clinical diagnosis and risk assessment of CHD and may provide a basis for further treatment of CHD.

Study on the Antinociceptive Effects of Herba Epimedium in Mice.

The present study was conducted to investigate the antinociceptive action of relationship between Herba Epimedium (HE) and 5-HT1A receptor, between Herba Epimedium (HE) and 5-HT2A receptor. We used the hot-plate method and the writhing assay in mice by the intracerebroventricular (i.c.v.) injection and observed the analgesic effect of HE. Furthermore, through the i.c.v. injection, 5-HT1A receptor partial agonist Buspirone, antagonist Propranolol, the adrenaline ß 1-receptor selective blocking agent Metoprolol, and 5-HT2A receptor agonist hydrochloride DOI and antagonist Ketanserin were used, and, 5 min later, HE was used to investigate the impacts of drugs on the analgesic effect in the same way. Results showed that HE had fast and significant antinociception in nervous system, and the effects can persist for a long time. Buspirone and Hydrochloride DOI can remarkably increase the antinociception of HE in nervous system. Ketanserin leads to a significant decrease in its antinociception in nervous system; Metoprolol also has antinociceptive action in nervous system, but it can inhibit the antinociceptive effect of Herba Epimediumin peripheral region. These results suggest that HE has significant antinociception effect and its mechanism is related with 5-HT1A receptor, 5-HT2A receptor, and adrenaline ß 1-receptor.

Diagnostic significance of apparent diffusion coefficient values with diffusion weighted MRI in breast cancer: a meta- analysis.

AIMS: Apparent diffusion coefficient (ADC) values of nodes in diffusion-weighted imaging (DWI) are widely used in differentiating metastatic from non-metastatic lymph nodes. The purpose of this meta-analysis was to demonstrate whether DWI could contribute to the precise diagnosis of breast cancer (BC) with and without lymph node metastasis (LNM). MATERIALS AND METHODS: English and Chinese electronic databases were searched for relevant studies followed by a comprehensive literature search. Two reviewers independently assessed the methodological quality of the included trials based on the quality assessment of diagnostic accuracy studies (QUADAS). Summary odds ratios (ORs) and corresponding 95% confidence intervals (95% CIs) were calculated. RESULTS: Final analysis of 624 BC subjects (patients with LNM = 254, patients without LNM = 370) were incorporated into the current meta-analysis from 9 eligible cohort studies. Combined ORs of ADCs suggested that ADC values in BC patients without LNM were higher than in patients with LNM (OR=0.56, 95%CI: 0.11-1.01, p=0.015). Subgroup analysis stratified by country indicated a low ADC value in BC patients with LNM rather than those without LNM among Chinese (OR=1.27, 95%CI: 0.89-1.66, p<0.001), Italians (OR=0.75, 95%CI: 0.13-1.38, p=0.018), and Egyptians (OR=1.27, 95%CI: 0.71-1.84, p<0.001). The findings of subgroup analysis by MRI machine type revealed that ADC values from diffusion MRI may be potential diagnostic indicators for BC using Non-Philips 1.5T (OR=1.10, 95%CI: 0.84-1.36, p<0.001). CONCLUSIONS: The main findings of our meta-analysis demonstrated that increased signal intensity on DWI and decreased signals on ADC are helpful in diagnosis of BC patients with or without LNM. DWI could therefore be an important imaging investigation in patients suspected of BC.

[Effect of kinetin on ovary and uterus in D-galactose-induced female mouse model of aging].

The present study was to investigate the effect of kinetin on ovary and uterus of D-galactose-induced female mouse model of aging. Aging female mice model caused by D-galactose were used as model group, the aging model mice intragastrically administered with kinetin solution (daily 25 mg/kg or 50 mg/kg) were used as kinetin groups, and the mice with solvent as normal group (n = 20). To detect the effects of kinetin, estrous cycle, estradiol content, ovarian and uterine wet weight and organ index, SOD and GSH-Px activities, MDA and total protein contents, as well as the reserve function of ovaries were examined. The results showed that, kinetin-induced changes in two kinetin groups were observed, compared with the model group: (1) the estrous cycle was shortened; (2) serum estradiol content was significantly increased; (3) the wet weights of the ovary and uterus were increased significantly; (4) SOD and GSH-Px activities of ovary and uterus were significantly higher; (5) the MDA contents of the ovary and uterus were reduced significantly; (6) total protein contents of the ovary and uterus were increased significantly; (7) the numbers of mature oocytes in fallopian tubes were increased significantly. The results show that kinetin can protect ovary and uterus against oxidative damage, prevent low estrogen secretion caused by ovarian oxidative damage, shorten the estrous cycle in mice, and eventually maintain ovarian and uterine vitalities.