Vast quantities of Magnetic Resonance Images (MRI) are routinely acquired in clinical practice but, to speed up acquisition, these scans are typically of a quality that is sufficient for clinical diagnosis but sub-optimal for large-scale precision medicine, computational diagnostics, and large-scale neuroimaging collaborative research. Here, we present a critic-guided framework to upsample low-resolution (often 2D) MRI full scans to help overcome these limitations. We incorporate feature-importance and self-attention methods into our model to improve the interpretability of this study. We evaluate our framework on paired low- and high-resolution brain MRI structural full scans (i.e., T1-, T2-weighted, and FLAIR sequences are simultaneously input) obtained in clinical and research settings from scanners manufactured by Siemens, Phillips, and GE. We show that the upsampled MRIs are qualitatively faithful to the ground-truth high-quality scans (PSNR = 35.39; MAE = 3.78E−3; NMSE = 4.32E−10; SSIM = 0.9852; mean normal-appearing gray/white matter ratio intensity differences ranging from 0.0363 to 0.0784 for FLAIR, from 0.0010 to 0.0138 for T1-weighted and from 0.0156 to 0.074 for T2-weighted sequences). The automatic raw segmentation of tissues and lesions using the super-resolved images has fewer false positives and higher accuracy than those obtained from interpolated images in protocols represented with more than three sets in the training sample, making our approach a strong candidate for practical application in clinical and collaborative research.
Among electroencephalogram (EEG) signal emotion recognition methods based on deep learning, most methods have difficulty in using a high-quality model due to the low resolution and the small sample size of EEG images. To solve this problem, this study proposes a deep network model based on dynamic energy features. In this method, first, to reduce the noise superposition caused by feature analysis and extraction, the concept of an energy sequence is proposed. Second, to obtain the feature set reflecting the time persistence and multicomponent complexity of EEG signals, the construction method of the dynamic energy feature set is given. Finally, to make the network model suitable for small datasets, we used fully connected layers and bidirectional long short-term memory (Bi-LSTM) networks. To verify the effectiveness of the proposed method, we used leave one subject out (LOSO) and 10-fold cross validation (CV) strategies to carry out experiments on the SEED and DEAP datasets. The experimental results show that the accuracy of the proposed method can reach 89.42% (SEED) and 77.34% (DEAP).
This study aims to investigate the correlation between the enhancement degree of contrast-enhanced ultrasound (CEUS) and the expression of CD147 and MMP-9 in carotid atherosclerotic plaques in patients with carotid endarterectomy and evaluate the diagnostic efficacy of CEUS using pathological results as the gold standard. Thirty-eight patients who underwent carotid endarterectomy (CEA) for carotid stenosis in the Department of Neurovascular Surgery of the Second People’s Hospital of Shenzhen from July 2019 to June 2020 were selected. Preoperatively, two-dimensional (2D) ultrasound scan was performed on all patients to assess the characteristics of the plaque and degree of stenosis, and CEUS was used to evaluate the surface morphology of the plaque and the distribution of neovascularization. Postoperatively, pathological sections and immunohistochemical analysis of CD147 and MMP-9 levels in the plaque were performed on the stripped plaque tissue, and the results were analyzed against the CEUS grading and pathological results. Among the 38 patients, pathological results showed that 10 and 28 were in the stable and vulnerable plaque groups, respectively. There were more smokers in the vulnerable plaque group than in the stable plaque group, with higher intraplaques CD147 and MMP-9. The difference in ultrasound plaque surface morphology grading and CEUS grading between the two groups was statistically significant. There was no significant difference in age, sex, incidence of complications such as hypertension, diabetes, and coronary heart disease between the two groups. CD147 was higher in the CEUS grade IV group than in the grades I (P = 0.040) and II (P = 0.010) groups. MMP-9 was higher in the CEUS grade IV group than in the grade II group (P = 0.017); MMP-9 was higher in the grade III group than in the grade II group (P = 0.015). Intraplaque contrast enhancement intensity was positively correlated with CD147 (r = 0.462, P = 0.003) and MMP-9 (r = 0.382, P = 0.018) levels. There was moderate consistency between the assessment of plaque vulnerability by 2D-ultrasound and by histopathological hematoxylin-eosin (HE) (kappa = 0.457, P > 0.05). 2D diagnosis of vulnerable plaque had a sensitivity of 85.7%, a specificity of 60.0%, a positive predictive value of 85.7%, a negative predictive value of 60.0%, and an accuracy of 78.0%. There was a strong consistency between the assessment of plaque vulnerability by CEUS and histopathological HE (kappa = 0.671, P < 0.01). CEUS had a sensitivity of 89.2%, a specificity of 80.0%, a positive predictive value of 92.6%, a negative predictive value of 72.7%, and an accuracy of 86.8% for the diagnosis of vulnerable plaques; CEUS is a reliable, non-invasive test that can show the distribution of neovascularization within vulnerable plaques, evaluate the vulnerability and risk of intraplaque hemorrhage, with a high consistency with pathological findings. The degree of intraplaque enhancement and the levels of CD147 and MMP-9 in the tissue were positively correlated.