英语专业论文 Design a simple apnea detection system

英语专业论文 Design a simple apnea detection system

3.2.1 EXCITATION SOURCE

The wien bridge oscillator which produces 50kHz and 8 V peak to peak signal is used as the excitation signal. The operational amplifier used in the circuit is LF351. The Voltage gain of the amplifier must be at least 3. The input resistance of the amplifier must be high compared toRso that theRCnetwork is not overloaded and alter the required conditions.The output resistance of the amplifier must be low so that the effect of external loading is minimized. Some method of stabilizing the amplitude of the oscillations must be provided because if the voltage gain of the amplifier is too small the desired oscillation will decay and if it is too large the waveform becomes distorted

3.2.2 CONSTANT CURRENT SOURCE

The constant source circuit is used to generate a 4mA constant current to be applied on the resistance circuit. CL100 and CK100 transistors are used in this circuit and these are npn and pnp paired transistors. The base emitter on voltage of these transistors is 0.9V. The collector current can be found by using the formula,

Ic= (Vcc-Vbe)/Rc

Where

Vcc-Supply voltage

Rc-Collector Resistance

Vbe-Base emitter on voltage

3.2.3 PHANTOM MODEL

The model consists of four resistors of 500 ohms which mimics the thoracic resistance.

3.3 DATA COLLECTION

To know about characteristics of normal respiration and apnea their corresponding signals were essential. So 40 respiration data sets with 100 sample values in each data set were collected from PHYSIONET -PHYSIOBANK ATM.

Among these 20 were normal data sets obtained from SLEEP HEART HEALTH STUDY POLYSOMNOGRAPHY DATABASE (SHHPSGDB) while the other 20 were apnea data sets obtained from UCD SLEEP APNEA DATABASE (UCDDB).

In Apnea data sets 10 belonged to Central Sleep Apnea and remaining 10 to Obstructive Sleep Apnea.

Each Data set contained 100 samples whose units are volts(V).They were recorded for 100seconds.So on plotting each data we get time in X-axis and volts in Y axis.

3.4 CLASSIFICATION OF APNEA USING RESPIRATION RATE

Input data which contains 60 samples each. Normalizing of the signal by squaring the signal. Extraction of maximum peak for every 5 samples.Display of respiratory cycles. If the peak value is greater than 6V it will be counted as normal respiratory cycle. If the count is between 10 and 20 the signal will be having normal respiratory rate. If the count is less than 10 the signal will be classified as bradypnea. If the count is greater than 20 the signal will be classified as tachypnea

As the parameter of respiratory rate alone is not enough for classifying the types of apnea the statistical parameters are calculated and then signals are classified using LabVIEW.

FLOWCHART

3.5 CLASSIFICATION OF APNEA USING STATISTICAL PARAMETERS

The signal data was imported from a spread sheet into labview using READ FROM SPREADSHEET block in labview. Then signal was plotted as a graph using WAVEFORM CHART block. The data cannot be manipulated directly so the transpose of the data is taken to find the statistical parameters using TRANSPOSE ARRAY block. Now using the STATISTICS block the signal's various parameters like arithmetic mean, median, mode, maximum peak, minimum peak, range, standard deviation variance, and rms value are found and recorded. Considering the range and mean of the signal it can be classified as its respective type. Give the upper and (or) lower limit for range and mean. Now using AND operator the signal is classified when its condition are satisfied. When the signal s range is greater than 7 and its mean is less than 0.1 it is normal. When the signal s range is lesser than 6 and its mean is greater than 0.21 it is abnormal. When the signal s range lies below 3.0 it is obstructive. When the signal s range lies between 3.1 and 6.99 it is central.

FLOWCHART

CHAPTER 4

4.1 RESULTS AND DISCUSSION

4.1.1 Hardware Results

Output from the excitation source (wein bridge oscillator) was checked in MULTISIM and then implemented using hardware. On applying the constant current to a resistance network that imitates human thoracic impedance , the current varied to a greater extent because of loading effect. The same problem will occur even when the patient is connected to the high frequency, low voltage, constant current module. Also, due to ethical issues the constant current generated cannot be given to the patient directly. So monitoring of real time data could not be done using the hardware design. Hence ,the idea of respiration signal simulation was dropped and offline data were collected from respiration databases for further classification.

4.1.2 Normal and Apnea Data

To know about characteristics of normal respiration and apnea their corresponding signals were essential. So 40 respiration data sets with 100 sample values in each data set were collected from PHYSIONET -PHYSIOBANK ATM. Among these 20 were normal data sets obtained from SLEEP HEART HEALTH STUDY POLYSOMNOGRAPHY DATABASE (SHHPSGDB) while the other 20 were apnea data sets obtained from UCD SLEEP APNEA DATABASE(UCDDB).The Resulting plot for each type of respiration signal is plotted below.

The following figure shows the normal respiration data plotted for 100 samples with time in x-axis and amplitude in y-axis with a maximum peak to peak voltage of 8V and 24 respiration cycles for 100seconds.

The following figure 4.4 shows Central sleep apnea data plotted for 100 samples with time in x-axis and amplitude in y-axis with a maximum peak to peak voltage of 6V and only one respiration cycle for 100 seconds.Obstructive sleep apnea data plotted for 100 samples with time in x-axis and amplitude in y-axis with a maximum peak to peak voltage of 2V and 19 respiration cycles for 100 seconds is shown

4.1.3 Parameter Calculation

The statistical parameters that showed major difference for various respiration data are mean and peak to peak voltage of the signal. Hence those values for 20 normal data sets and apnea data sets are obtained and tabulated in table 4.1 and 4.2 respectively.

Table 4.1: Mean and peak to peak voltage (V-PP) for 20 normal data sets

Table 4.2: Mean and peak to peak voltage (V-PP) for 20 apnea data sets

4.1.4 Results From Parameter Calculation

The tabulated values were analyzed and a specific mean and peak to peak voltage value for normal and apnea data were extracted .Using these extracted values classification of apnea was done using LabVIEW.

The values inside the paranthesis shows the standard deviation for each statistical parameter .eg. For Normal Signal the mean value is 0V with a upper limit of 0.2 and a lower limit of -0.2 . Similarly the peak to peak voltage is 8V with an upper limit of 8.5 and a lower limit of 7.5

Table 4.3: STATISTICAL PARAMETERS FOR APNEA CLASSIFICATION

4.1.5 Results Of Classification Of Apnea In Matlab