Durand L G, Grenier M C, Inderbitzen R, Wieting D W, Stein P D
Biomedical Engineering Laboratory, Clinical Research Institute of Montreal, Quebec, Canada.
J Heart Valve Dis. 1995 Jul;4 Suppl 1:S32-7.
The aim of the study was to evaluate the potential of low frequency analysis of valve opening sounds in order to detect the presence of single leg separation of the outlet strut of Björk-Shiley convexo-concave (BSCC) valves. Single leg separation is believed to precede outlet strut fracture. Phonocardiograms (PCG) of 28 patients with BSCC mitral valves were recorded and filtered to limit the frequency bandwidth between 90 and 1400 Hz (-20 dB). Twenty-four patients had BSCC valves that were presumably unimpaired and four patients had BSCC valves with single leg separation. Spectral analysis consisted of computing the power spectrum for 30 opening sounds for each patient using a fast Fourier transform (FFT) algorithm on an IBM-PC compatible computer. For each patient, the power spectra of the selected opening sounds were averaged and normalized in amplitude on a 0-100% linear scale. A ratio of the high to the low frequency area of the averaged power spectra was computed and evaluated as a diagnostic feature to identify the valves with single leg separation. Six cut-off frequencies serving to distinguish between high and low frequency areas of the spectra were tested: 400, 450, 500, 550, 600, and 650 Hz. The results showed that the high-to-low frequency area ratios tested were all able to identify valves with single leg separation. The correct classification performance was 96.4%. The sensitivity to detect valves with single leg separation was 75% and the specificity was 100%. The positive predictive value was 100% and the negative predictive value 96%. The study confirmed the potential of low frequency analysis of valve opening sounds to detect single leg separation of the outlet strut of BSCC valves implanted in humans.
本研究的目的是评估瓣膜开放音低频分析检测Björk-Shiley凸凹型（BSCC）瓣膜出口支柱单腿分离情况的潜力。人们认为单腿分离先于出口支柱骨折。记录了28例植入BSCC二尖瓣患者的心音图（PCG），并进行滤波以将频率带宽限制在90至1400赫兹（-20分贝）之间。24例患者的BSCC瓣膜推测未受损，4例患者的BSCC瓣膜存在单腿分离。频谱分析包括在一台IBM-PC兼容计算机上使用快速傅里叶变换（FFT）算法为每位患者的30个开放音计算功率谱。对于每位患者，将所选开放音的功率谱进行平均，并在0-100%线性尺度上对幅度进行归一化。计算平均功率谱高频与低频区域的比值，并将其作为诊断特征进行评估，以识别存在单腿分离的瓣膜。测试了六个用于区分频谱高频和低频区域的截止频率：400、450、500、550、600和650赫兹。结果表明，所测试的高频与低频区域比值均能够识别存在单腿分离的瓣膜。正确分类性能为96.4%。检测存在单腿分离瓣膜的敏感性为75%，特异性为100%。阳性预测值为100%，阴性预测值为96%。该研究证实了瓣膜开放音低频分析检测植入人体的BSCC瓣膜出口支柱单腿分离情况的潜力。
