No abstract

Dynamic contrast-enhanced magnetic resonance imaging (DCE MRI) of the breast is a routinely used imaging method which is highly sensitive for detecting breast malignancy. Specificity, though, remains suboptimal. Dynamic susceptibility contrast magnetic resonance imaging (DSC MRI), an alternative dynamic contrast imaging technique, evaluates perfusion-related parameters unique from DCE MRI. Previous work has shown that the combination of DSC MRI with DCE MRI can improve diagnostic specificity, though an additional administration of intravenous contrast is required. Dual-echo MRI can measure both T1W DCE MRI and T2*W DSC MRI parameters with a single contrast bolus, but has not been previously implemented in breast imaging. We have developed a dual-echo gradient-echo sequence to perform such simultaneous measurements in the breast, and use it to calculate the semi-quantitative T1W and T2*W related parameters such as peak enhancement ratio, time of maximal enhancement, regional blood flow, and regional blood volume in 20 malignant lesions and 10 benign fibroadenomas in 38 patients. Imaging parameters were compared to surgical or biopsy obtained tissue samples. Receiver operating characteristic (ROC) curves and area under the ROC curves were calculated for each parameter and combination of parameters. The time of maximal enhancement derived from DCE MRI had a 90% sensitivity and 69% specificity for predicting malignancy. When combined with DSC MRI derived regional blood flow and volume parameters, sensitivity remained unchanged at 90% but specificity increased to 80%. In conclusion, we show that dual-echo MRI with a single administration of contrast agent can simultaneously measure both T1W and T2*W related perfusion and kinetic parameters in the breast and the combination of DCE MRI and DSC MRI parameters improves the diagnostic performance of breast MRI to differentiate breast cancer from benign fibroadenomas.

Abstract. Weather radar echo is one of the fundamental data for meteorological workers to weather systems identification and classification. Through the technique of weather radar echo extrapolation, the future short-term weather conditions can be predicted and severe convection storms can be warned. However, traditional extrapolation methods cannot offer accurate enough extrapolation results since their modeling capacity is limited, the recent deep learning based methods make some progress but still remains a problem of blurry prediction when making deeper extrapolation, which may due to they choose the mean square error as their loss function and that will lead to losing echo details. To address this problem and make a more realistic and accurate extrapolation, we propose a deep learning model called Adversarial Extrapolation Neural Network (AENN), which is a Generative Adversarial Network (GAN) structure and consist of a conditional generator and two discriminators, echo-frame discriminator and echo-sequence discriminator. The generator and discriminators are trained alternately in an adversarial way to make the final extrapolation results be realistic and accurate. To evaluate the model, we conduct experiments on extrapolating 0.5h, 1h, and 1.5h imminent future echoes, the results show that our proposed AENN can achieve the expected effect and outperforms other models significantly, which has a powerful potential application value for short-term weather forecasting.

Background Breast cancer is the most common cancer in women worldwide, high-resolution dynamic contrast-enhanced MRI (DCE-MRI) can better evaluate the tissue microenvironment and texture characteristics. The purpose of this study was to investigate the value of the texture-based analysis for breast DCE-MRI in the diagnosis of breast lesions and background enhancement. Methods This study prospectively enrolled 128 patients with clinically suspected breast lesions in our hospital from April 2015 to June 2017. Among them, 62 patients underwent preoperative high temporal resolution DCE-MRI (1 + 26 phases) scan with 39 malignant and 23 benign lesions. The control group retrospectively and randomly contained 78 patients who underwent preoperative low temporal resolution DCE-MRI (1 + 5 phases) scans with 46 malignant and 32 benign lesions. Quantitative parameters were obtained using a two-compartment Extended Tofts and volume of interest model for the lesion center, surrounding peripheral area and background enhancement, including pharmacokinetic parameters (Ktrans, Kep, Ve and Vp) and texture features based on the Ktrans map. The Student’s t-test was used to compare the differences of means. LASSO was used for dimension reduction and logistic regression analysis was used for model construction. A receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance. Results Pharmacokinetic parameters were significantly different between high temporal resolution and low temporal resolution DCE-MRI (P < 0.05). In the malignant group, the average Ktrans of the lesion area on high temporal resolution DCE-MRI was significantly correlated to the pathological grading (r = 0.400, P = 0.012). In the differentiation between benign and malignant lesions, the ROC analysis demonstrated that the diagnostic value of high temporal resolution DCE-MRI offered slightly significant advantages in the realms of the lesion, peripheral areas and background enhancement. Conclusions The use of texture analysis based on high temporal resolution DCE-MRI may potentially improve breast cancer diagnostic performance. Specifically, combining the lesion, peripheral, BE area, and Ktrans-mean parameters can contribute to the diagnosis of breast lesions, background enhancement and the pathological grading of malignant tumors.