Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Table 1.

Percentage of correctly classified signals for different heart sounds.

More »

Table 1 Expand

Figure 1.

A flowchart of the hybrid LPC-SVM-MCS system training process.

The system first collects PCG signals and performs their segmentation to extract useful information for LPC estimation. Then the training process commences where the Modified Cuckoo Search algorithm optimizes parameters of a Support Vector Machine classifier.

More »

Figure 1 Expand

Figure 2.

Result of the heart tone segmentation algorithm.

The waveform presented contains normal S1, S2 and S3 heart sounds, which are segmented by a variable size time window for further analysis.

More »

Figure 2 Expand

Figure 3.

A S3 heart tone and LPC filter spectra.

A comparison between a real S3 heart tone spectrum and spectra of filters estimated by the LPC algorithm. The 24th order filter provides the closest representation of the original heart tone spectrum.

More »

Figure 3 Expand

Figure 4.

Spectrum matching error for different filter orders.

A comparison of matching errors in replicating a S3 heart tone spectrum. The presented curves indicate errors for three filters estimated by a Linear Predictive Coding algorithm with a transfer function of the 5th (red dotted line), 18th (green dotted line) and 24th (blue line) order. The 24th order filter obtained significantly lower error.

More »

Figure 4 Expand

Figure 5.

Spectrum comparison of selected heart sounds and LPC filters.

Presented curves demonstrate the effectiveness of the modified LPC algorithm in estimating different heart sounds.

More »

Figure 5 Expand

Figure 6.

Data flow in the proposed classification system.

The picture presents data transfer in the proposed classification system. The modified LPC algorithm estimates filter coefficients and passes them to the training part. After the training and optimisation process, the selection of appropriate coefficients, selected kernel function and its parameters, and the penalty parameter C can be used in validation of the testing data set.

More »

Figure 6 Expand

Figure 7.

Example heart sounds used in the tests.

A – early systolic murmur, B – S4, C – pansystolic murmur, D – S3, E – late systolic murmur, F – normal split S2, G – normal split S1, H – ejection click, I – diastolic rumble, J – opening snap.

More »

Figure 7 Expand

Table 2.

Percentage of correctly classified signals for different heart sounds from a leave-one-out test.

More »

Table 2 Expand

Table 3.

Classification accuracy of compared methods for various number of considered classes.

More »

Table 3 Expand

Table 4.

Classification accuracy of proposed methods for various number of considered classes.

More »

Table 4 Expand

Figure 8.

Comparison of classification results for all test groups.

A comparison of accuracy of all tested methods (being: ANN – Artificial Neural Network, SVM-poly – Support Vector Machine with polynomial kernel function, SVM-rbf – Support Vector Machine with radial basis kernel function, SVM-quad – Support Vector Machine with quadratic kernel function, SVM-MCS-ca – Support Vector Machine with the Modified Cuckoo Search optimizer and classification accuracy fitness function) for a different number of recognizable classes. The SVM-MCS-ce method shows overall the best quality of classification.

More »

Figure 8 Expand