Feature representation is the crucial factor for the magnetic resonance imaging (MRI) based computer-aided diagnosis (CAD) of Parkinson’s disease (PD). Deep polynomial network (DPN) is a novel supervised deep learning algorithm, which has excellent feature representation for small dataset. In this work, a stacked DPN (SDPN) based ensemble learning framework is proposed for diagnosis of PD, which can improve diagnostic accuracy for small dataset. In the proposed framework, SDPN was performed on each subset of extracted features from MRI images to generate new feature representation. The support vector machine (SVM) was then adopted to perform classification task on each subset. The ensemble learning algorithm was then performed on all the SVM classifiers to generate the final diagnosis for PD. The experimental results on the Parkinson’s Progression Markers Initiative dataset (PPMI) showed that the proposed algorithm achieved the classification accuracy, sensitivity and specificity of 90.15%, 85.48% and 93.27%, respectively, with the brain network features, and it also got the classification accuracy of 87.18%, sensitivity of 86.90% and specificity of 87.27% on the multi-view features extracted from different brain regions. Moreover, the proposed algorithm outperformed other algorithms on the MRI dataset from PPMI. It suggests that the proposed SDPN-based ensemble learning framework has the feasibility and effectiveness for the CAD of PD.
Accurate reconstruction of tissue elasticity modulus distribution has always been an important challenge in ultrasound elastography. Considering that existing deep learning-based supervised reconstruction methods only use simulated displacement data with random noise in training, which cannot fully provide the complexity and diversity brought by in-vivo ultrasound data, this study introduces the use of displacement data obtained by tracking in-vivo ultrasound radio frequency signals (i.e., real displacement data) during training, employing a semi-supervised approach to enhance the prediction accuracy of the model. Experimental results indicate that in phantom experiments, the semi-supervised model augmented with real displacement data provides more accurate predictions, with mean absolute errors and mean relative errors both around 3%, while the corresponding data for the fully supervised model are around 5%. When processing real displacement data, the area of prediction error of semi-supervised model was less than that of fully supervised model. The findings of this study confirm the effectiveness and practicality of the proposed approach, providing new insights for the application of deep learning methods in the reconstruction of elastic distribution from in-vivo ultrasound data.
Objective To introduce a modified Sakakibara classification system for a ruptured sinus of Valsalva aneurysm （RSVA），and suggest different surgical approaches for corresponding types of RSVA. Methods Clinical data of 159 patients undergoing surgical repair for RSVA in Fu Wai Hospital between February 2006 and January 2012 were retrospectively analyzed. There were 105 male and 54 female patients with their age of 2-71 （33.4±10.7） years. All these patients were divided into 5 types as a modified Sakakibara classification system. Type I: rupture into the right ventricle just beneath the pulmonary valve （n=66），including 84.8% patients with ventricular septal defect （VSD） and 53.8% patients with aortic valve insufficiency （AI）. TypeⅡ：rupture into or just beneath the crista supraventricularis of the right ventricle （n=17），including 88.2% patients with VSD and 23.5% patients with AI. Type Ⅲ：rupture into the right atrium （typeⅢ a，n=21） or the right ventricle （typeⅢv，n=6） near or at the tricuspid annulus，including 18.5% patients with VSD and 25.9% patients with AI. TypeⅣ：rupture into the right atrium （n=46），including 23.9% patients with AI but no patient with VSD. TypeⅤ：other rare conditions，such as rupture into the left atrium，left ventricle or pulmonary artery （n=3），including 100% patients with AI and 33.3% patients with VSD. Most RSVA originated in the right coronary sinus （n=122），and others originated in the noncoronary sinus （n=35） or left coronary sinus （n=2）. Results All the type V patients （100%） and 50% patients with typeⅢv received RSVA repair through aortotomy. In most patients of typeⅠ，II andⅣ，repair was achieved through the cardiac chamber of the fistula exit （71.2%，64.7% and 69.6% respectively）. Both routes of repair were used in 76.2% patients with typeⅢ a. The cardiopulmonary bypass time （92.4±37.8 minutes） and aortic cross-clamp time （61.2±30.7 minutes） was the shortest to repair typeⅣRSVA. There was no in-hospital death in this group. Two patients （type I andⅡrespectively） underwent reoperation during the early postoperative period because of restenosis of the right ventricular outflow tract. Most patients received reinforcement patch for RSVA repair （n=149），and only 10 patients received simple suture repair （including 5 patients with typeⅣ，4 patients with typeⅢ a and 1 patient with typeⅡ）. Aortic valve replacement was performed for 33 patients （66.7% of those with typeⅠ）. A total of 147 patients （92.5%） were followed up after discharge. Two patients （type I andⅢ a respectively） developed atrial fibrillation and received radiofrequency ablation treatment，1 patient （typeⅣ） underwent reoperation for residual shunt，and there was no late death during follow-up. Conclusion Modified Sakakibara classification system for RVSA provides a guidance to choose an appropriate surgical approach，and satisfactory clinical outcomes can be achieved for all types of RSVA.
The effective classification of multi-task motor imagery electroencephalogram (EEG) is helpful to achieve accurate multi-dimensional human-computer interaction, and the high frequency domain specificity between subjects can improve the classification accuracy and robustness. Therefore, this paper proposed a multi-task EEG signal classification method based on adaptive time-frequency common spatial pattern (CSP) combined with convolutional neural network (CNN). The characteristics of subjects' personalized rhythm were extracted by adaptive spectrum awareness, and the spatial characteristics were calculated by using the one-versus-rest CSP, and then the composite time-domain characteristics were characterized to construct the spatial-temporal frequency multi-level fusion features. Finally, the CNN was used to perform high-precision and high-robust four-task classification. The algorithm in this paper was verified by the self-test dataset containing 10 subjects (33 ± 3 years old, inexperienced) and the dataset of the 4th 2018 Brain-Computer Interface Competition (BCI competition Ⅳ-2a). The average accuracy of the proposed algorithm for the four-task classification reached 93.96% and 84.04%, respectively. Compared with other advanced algorithms, the average classification accuracy of the proposed algorithm was significantly improved, and the accuracy range error between subjects was significantly reduced in the public dataset. The results show that the proposed algorithm has good performance in multi-task classification, and can effectively improve the classification accuracy and robustness.
Objective To identify clinical risk factors for early major adverse cardiovascular events (MACEs) following surgical correction of supravalvar aortic stenosis (SVAS). Methods Patients who underwent SVAS surgical correction between 2002 and 2019 in Beijing and Yunnan Fuwai Cardiovascular Hospitals were included. The patients were divided into a MACEs group and a non-MACEs group based on whether MACEs concurring during postoperative hospitalization or within 30 days following surgical correction for SVAS. Their preoperative, intraoperative, and postoperative clinical data were collected for multivariate logistic regression. Results This study included 302 patients. There were 199 males and 103 females, with a median age of 63.0 (29.2, 131.2) months. The incidence of early postoperative MACEs was 7.0% (21/302). The multivariate logistic regression model identified independent risk factors for early postoperative MACEs, including ICU duration (OR=1.01, 95%CI 1.00-1.01, P=0.032), intraoperative cardiopulmonary bypass (CPB) time (OR=1.02, 95%CI 1.01-1.04, P=0.014), aortic annulus diameter (OR=0.65, 95%CI 0.43-0.97, P=0.035), aortic sinus inner diameter (OR=0.75, 95%CI 0.57-0.98, P=0.037), and diameter of the stenosis (OR=0.56, 95%CI 0.35-0.90, P=0.016). Conclusion The independent risk factors for early postoperative MACEs include ICU duration, intraoperative CPB time, aortic annulus diameter, aortic sinus inner diameter, and diameter of the stenosis. Early identification of high-risk populations for MACEs is beneficial for the development of clinical treatment strategies.