屈光度对SMILE术后眼压的预测和校正

目的 探究SMILE术后眼压的预测并校正。方法 回顾性研究+前瞻性验证。纳入2019-12/2021-12于中部战区总医院行SMILE的患者90例(180眼),随机抽取30眼为验证集,余150眼为训练集。前瞻性录入2022-01/2022-09SMILE的51例(102眼)为测试集。首先分析训练集手术前3d、术后1wk、1mo、3mo及6mo的眼压情况,确定眼压稳定期及于术前的眼压变化量(ΔIOP);再选取球柱镜代数和(SC)、球镜度(DS)、柱镜度(DC)及眼压(IOP)、中央角膜厚度(CCT)、角膜平均曲率(Km)、前房深度(ACD),眼压稳定时的变化量(ΔCCT、ΔKm、ΔK1、ΔK2、ΔACD)及个体因素,纳入与ΔIOP相关的因素拟合回归模型;最后建立适合临床SMILE术后眼压预测及校正的公式并作效能验证。结果 除术后1wk、3mo及6moIOP三者间无明显差异外,余时段IOP间差异均有统计学意义(P＜0.05);各因素与ΔIOP拟合最佳回归方程：ΔIOP=0.459IOP_术前-0.183SC-0.041Age+1.292ΔACD-1.270(调节R₂^=0.533,P＜0.001)。仅纳入术前IOP和SC,拟合简化方程：ΔIOP=0.496IOP_术前-0.194SC-2.952(调节R₂^=0.498,P＜0.001);预测和校正公式：IOP_术后预测=0.5IOP_术前+0.2SC+3,IOP_术后校正=IOP_术后+0.5IOP_术前–0.2SC–3。简化公式的效能验证显示良好。结论 SMILE术后3月IOP趋于稳定,且ΔIOP与屈光度相关。SMILE术后IOP可通过简化公式作快速临床预测和校正。     

基于多生理信号的情绪初步识别

以生理信号分析为主,表情行为观察和情绪主观感受评价为辅,对多名被试的情绪进行识别.60名大学女生接受恐惧-快乐-轻松的情绪诱发,有效数据55名,对应每个情绪片段,根据信号标记以及GSR微分,截取1min的生理信号进行处理和分析,应用SPSS对各生理参数进行情绪的单因素方差分析,然后采用逐步多类判别法,提取特征参数以识别情绪.结果表明HR、HRV、R波、T波各生理参数对情绪较敏感;提取出HFP,HR max,PNN50,LF/HF,Ratio,LFP,Mean NN 7个特征参数,构建情绪判别函数Fuction1,Fuction2和Z1、Z2,Z3;轻松的判别正确率为88.0%,快乐的为92.0%,恐惧的为80.0%,总体判别正确率为86.7%.以生理信号分析为主,辅助表情行为观察和情绪主观感受报告,是一种有效的情绪识别方法,所得数据客观、准确,提高了情绪识别率.     

呼气相胸片的识别研究

根据呼气相胸片中肺部区域灰度较高的特点，通过边缘提取和边缘跟踪确定胸部区域，运用最大类间方差方法确定图像分割的阈值．再提取出肺部区域，然后求出肺部的灰度均值。但这样的结果并不理想。于是在此基础上提出修正方案，即以肺部灰度均值和骨骼灰度均值的比值作为特征值．即比例法。此方法可以较客观的识别呼气相胸片。最后用类间距离对这两种方法进行了量化比较。     

基于模糊连接度的多发性硬化症MR图像自动分割算法

多发性硬化症(MS)是一种严重威胁中枢神经功能的疾病,利用磁共振成像技术能够无损伤地检出其病灶。为了自动地对多发性硬化症病灶进行分割,提出了基于模糊连接度的分割算法,实现了种子点的自动选取。作为多发性硬化症分割的预处理,针对脑部MR FLAIR图像的特征,基于区域增长方法,还提出了脑部组织提取算法。通过对临床患者MR图像的分割实验,表明该分割算法能够比较准确地分割多发性硬化症病灶,其分割效果明显好于模糊C-均值聚类算法和基于马尔可夫场模型的分割算法。该算法还具有无监督、运算速度快、稳健性好等优点,能够应用于多发性硬化症的临床辅助诊断。     

Testicular sperm extraction in azoospermic men submitted to bilateral orchidopexy

BACKGROUND: This study was carried out to evaluate whether bilateral orchidopexy represents a poor or good prognostic factor in azoospermic men undergoing testicular sperm extraction (TESE). METHODS: One hundred and seven presumed non-obstructive azoospermia (NOA) patients, according to conventional clinical parameters (volume of testis, FSH, clinical history) were submitted to testicular biopsy with TESE. Thirty men (28%) had a history of bilateral orchidopexy for cryptorchidism. RESULTS: Normal spermatogenesis or mild hypospermatogenesis was diagnosed in 12/30 ex-cryptorchid patients and in 7/77 presumed NOA patients (P = 0.0004). Conversely, pure Sertoli cell-only syndrome or complete maturation arrest was found in 10/30 ex-cryptorchid patients and in 48/77 presumed NOA patients (P = 0.0094). In 53/107 patients (49.5%), TESE allowed a positive sperm retrieval. At least one spermatozoon was observed in 22/30 ( approximately 73%) ex-cryptorchid patients and in 31/77 ( approximately 40%) presumed NOA patients (P = 0.0026). A large number of spermatozoa (equivalent to an obstructive pathology) were retrieved in 13/30 ex-cryptorchid and in 10/77 presumed NOA patients (P = 0.001). A history of bilateral orchidopexy in presumed NOA patients correlates positively for the chance of retrieving testicular spermatozoa (odds ratio 3.8; 95% confidence interval 1.41-10.21; P = 0.008). CONCLUSIONS: Although bilateral cryptorchidism is usually considered a testicular secretive dysfunction, TESE permits retrieval of a large number of spermatozoa in almost 40% of cases. Our data suggest the existence of congenital or acquired obstructive anomalies of the seminal ducts in azoospermic orchidopexed men.     

An intelligent system for diagnosis of the heart valve diseases with wavelet packet neural networks

In this paper, an intelligent system is presented for interpretation of the Doppler signals of the heart valve diseases based on the pattern recognition. This paper especially deals with combination of the feature extraction and classification from measured Doppler signal waveforms at the heart valve using the Doppler Ultrasound. Because of this, a wavelet packet neural network model developed by us is used. The model consists of two layers: wavelet and multi-layer perceptron. The wavelet layer is used for adaptive feature extraction in the time-frequency domain and is composed of wavelet packet decomposition and wavelet packet entropy. The multi-layer perceptron used for classification is a feed-forward neural network. The performance of the developed system has been evaluated in 215 samples. The test results showed that this system was effective in detecting Doppler heart sounds. The correct classification rate was about 94% for abnormal and normal subjects.     

基于离散小波变换提取脑机接口中脑电特征

在脑机接口中，针对脑电特征提取利用单一种类信息、使用数据量大、分类性能较差等缺点，提出一种新颖的基于离散小波变换的方法。分析了小波变换特征提取的特点和特征表示方式，用Daubechies类db4小波函数对脑电信号进行6层分解，抽取小波变换各子带关键的部分逼近系数、小波系数、小波子带系数均值组成特征向量。以分类正确率为指标检验了提取特征的性能。实验结果表明，这种方法能够利用少量数据提取脑电信号本质特征，具有较高的分类性能，为利用脑电识别人的不同意图提供了快速而有效的手段。     

基于脑电复杂度的意识任务的特征提取与分类

本研究提出了利用事件相关脑电复杂度提取大脑运动意识特征，应用Mahalanobis距离判别式分析法，对人脑想象左右手运动任务进行分类，获得了满意的结果。对受试者想象左右手运动期间在大脑初级感觉运动皮层区记录的脑电信号采用复杂度分析方法量化了事件相关去同步（ERD）和事件相关同步（ERS）时程，结果表明EEG复杂度特征较好反映了ERD／ERS变化时程。最后对测试数据进行分类，最大分类正确率达到86．43％，通过最大分类正确率，最大信噪比，最大互信息等评价指标比较，验证了该方法的有效性，从而为大脑运动意识任务的特征提取及分类提供了新思路。     

基于自适应小波神经网络的心电图检测法

自适应小波神经网络检测法就是利用小波函数取代通常神经网络中隐层的作用函数来实现的，通过网络的学习自适应地调整尺度参数和时移因子，提高了特征提取能力，因此，该网络可提高心电信号的检测率和可靠性。