Irregular Breathing Classification from Multiple Patient Datasets - Prediction and Classification of Respiratory Motion - page 124

Fig. 6.14 ROC graph of
irregular detection with
different observation period

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

T i = 100
T i = 300
T i = 900

0.2

0.1

0

0 0.2 0.4 0.6 0.8 1

FP Rate

Figure 6.14 shows ROC graphs to evaluate how different observation periods
affect the classification performance. Here, we fixed the regular threshold Wth of
0.92 that is the mean value of the ratio c ðÞ , shown in Fig. 6.10 . In Fig. 6.14 we
can see that the proposed classifier shows a better performance with a long
observation period (T i ). That means the classifier can be improved by extending
the observation period for feature extraction.
Figure 6.15 shows ROC graphs of irregular detection with different regular
thresholds Wth of 0.8, 0.85, and 0.9. In this figure, the ratio c ðÞ of patients i are
extracted with observation periods T i of 100, 300, and 900 s.
The smaller the regular threshold Wth, the better the classifier performance.
Here, we notice that the true positive rate (TPR) for the proposed classifier is
97.83 % when the False Positive Rate (FPR) is 50 % in Fig. 6.15 c.
Based on the result of ROC graph in Fig. 6.15 c, we notice that the breathing
cycles of any given patient with a length of at least 900 s can be classified reliably
enough to adjust the safety margin prior to therapy in the proposed classification.
For the overall analysis of the curve, we have shown the area under the ROC curve
(AUC) in Fig. 6.16 . The AUC value can be increased by lowering the regular
threshold Wth. The maximum AUCs for observation period T i of 100, 300, and
900 s are 0.77, 0.92, and 0.93, respectively. Based on Figs. 6.15 a-c, 6.16 picked
0.8, 0.85, and 0.9 for Wth.
Some studies investigated the classification of regular/irregular breathing pat-
terns for the detection of lung diseases with spirometry [ 14 - 17 ]. The irregular
breathing patterns can also impact on the dosimetric treatment for lung tumors in
stereotactic body radiotherapy [ 5 , 11 , 12 ]. However, there are few studies with the
results on the classification of breathing irregularity in this area. The following
table shows the classification performance of irregular breathing detection using a
variety of respiratory measurement datasets.