BackgroundTo validate the clinical value of simple rules in distinguishing malignant adnexal masses from benign ones and to explore the effect of simple rules for experienced and less-experienced sonographers.MethodsPatients with persistent adnexal masses were enrolled between November 2013 and December 2015. All masses were proven through histological examinations. Five sets of diagnoses were made and compared with one another. Diagnosis 1 was made, according to the simple rules, by a trainee with little clinical diagnostic experience. Diagnoses 2 and 3 were made by experienced and less-experienced sonographers, respectively, according to their clinical experiences. With diagnosis 1 as a reference, the two sonographers were asked to provide a second diagnosis, which were diagnoses 4 and 5. The efficiency of the five sets of diagnoses was compared using ROC curves.ResultsIn total, 75 malignant (37.7%) and 124 benign lesions (62.3%) were enrolled in this study. The mean diameter of the benign masses was obviously smaller than that of the malignant ones (6.8 ± 3.4 cm vs. 9.3 ± 4.9 cm, p < 0.01). The malignant ratio in postmenopausal women was much higher (66.1%) than that in the premenopausal population (25.7%) (p < 0.0001). Totally, 156 of the 199 cases (79.4%) resulted in conclusive diagnoses. Sensitivity and specificity were 98.4% and 73.9%, respectively, among the conclusive cases. The area under the ROC curve (Az) for the simple rule diagnosis was significantly lower than that for the experienced sonographer diagnosis (0.85 vs. 0.96, p < 0.0001); compared with the less-experienced sonographer, this difference was not significant (0.85 vs. 0.86, p = 0.9776). No significant difference was found in the comparison between the diagnoses made by the experienced sonographer before and after referencing the simple rule diagnosis (Az, 0.96 vs. 0.97, p = 0.2055). Using diagnosis 1 as a reference, the diagnostic performance of the less-experienced sonographer increased (from 0.86 to 0.92, p = 0.012); however, it was still lower than that of the experienced sonographer (Az, 96% vs. 92%, p = 0.0241).ConclusionsThe simple rules was an appealing method for discriminating malignant masses from benign ones, particularly for a less-experienced sonographer.

OBJECTIVE:The aim of our study was to evaluate the role of preoperative US, CEUS, and 99m Tc-MIBI scanning with SPECT/CT in localizing diseased parathyroid glands in cases of refractory secondary hyperparathyroidism (SHPT). MATERIAL AND METHODS: Using pathological results as the gold standard, we compared the operative findings with the preoperative localization of each modality in 73 nodules and evaluated the accuracy, and sensitivity of each modality and combinations of the four modalities. RESULTS: The sensitivity of US, CEUS, 99m Tc-MIBI and SPECT/CT was 98.59%, 94.37%, 50.70% and 78.87%, respectively. US had the highest sensitivity of the four imaging methods and the diagnostic sensitivity of US and CEUS was superior to that of 99m Tc-MIBI (p < 0.001 and p < 0.001) and SPECT/CT (p = 0.001 and p = 0.012). In addition, we found that the sensitivity of the combination of US with CEUS, US with 99m Tc-MIBI and/or SPECT/CT, CEUS with 99m Tc-MIBI and/or SPECT/CT, US with CEUS and two other imaging modalities ( 99m Tc-MIBI and/or SPECT/CT) was 98.59%, 100%, 95.77%, and 100%, respectively. CONCLUSIONS: The combination of US with SPECT/CT is the best choice for the comprehensive preoperative localization of glands in refractory SHPT. CEUS can elevate the accuracy of US in differential diagnosis via the interpretation of dynamic microvascular features.

In this paper, we propose a novel hyperspectral image superresolution method based on superpixel spectral unmixing using a coupled encoder-decoder network. The hyperspectral image and multispectral images are fused to generate high-resolution hyperspectral images through the spectral unmixing framework with low-rank constraint. Specifically, the endmember and abundance information is extracted via a coupled encoder-decoder network integrating the priori for unmixing. The coupled network consists of two encoders and one shared decoder, where spectral information is preserved through the encoder. The multispectral image is clustered into superpixels to explore self-similarity, and then, the superpixels are unmixed to obtain an abundance matrix. By imposing a low-rank constraint on the abundance matrix, we further improve the superresolution performance. Experiments on the CAVE and Harvard datasets indicate that our superresolution method outperforms the other compared methods in terms of quantitative evaluation and visual quality.