CARDIOVASCULAR SOUND 



685 



at the diaphragm. For example, if the output le\cl of 

 a microphone is —45 db, the voltage generated is 45 

 db below the reference level of i volt per dyne per 

 cm-. The smaller the absolute value (in db) of the 

 output expression the more sensitiv'e is the micro- 

 phone. 



Input 



He 



c 



R Output 



FILTRATION, INCLUDING AMPLIFICATION AND 

 LOW-FREqUENCV ATTENUATION (EqUALIZATION) 



Overloading from the intense precordial vibrations 

 of low frequency must be avoided in phonocardiog- 

 raphy. For example, in early phonocardiography so- 

 called "low-frequency attenuation" was achieved by 

 acoustic filtration; an air leak eliminated gross low- 

 frequency components of the precordial vibration. In 

 more recent work electronic filtration has been used 

 for the same purpose. 



In optical registration the amplitude which can be 

 encompassed is of the order of 50: i (34 db). (In this 

 estimate it is assumed that i inch is allowed for maxi- 

 mal deflection and that .02 inch is just distinguish- 

 able.) Because of the logarithmic compression charac- 

 teristic of the function of the ear, in accordance with 

 the Weber-Fechner law, sounds with an intensity 

 range of 1,000,000:1 (120 db) can be detected. The 

 intensity of cardiovascular sound decreases roughly 

 with the second power of the frequency, or about 12 

 db per octave. Obviously a weak high-frequency sig- 

 nal requires much greater amplification for \isual 

 registration than does an intense low-frequency signal. 



A single oscillogram cannot adequately represent 

 the entire frequency intensity range of cardiovascular 

 sound, and various techniques of differential filtration 

 or amplification are necessary. Filters used in phono- 

 cardiography can be classed as a) low-cut, high-pass; 

 b) high-cut, low-pass; and c) band-pass. The general 

 definition of each is evident from its designation. The 

 first and third are the types used in phonocardiog- 

 raphy. The band-pass filter essentially combines a 

 high-pass and a low-pass filter in series. Figure i 

 presents (upper) a schematic view of the circuit for 

 low-frequency attenuation (equalization; high-pass 

 filtration). In the low-pass filter the positions of the 

 resistor and capacitor are interchanged. In the lower 

 section the output characteristics of the circuit above 

 are schematized. Figure 2 presents the characteristics 

 of two systems which have used multiple high-pass 

 filters. More detailed descriptions of the characteristics 

 of these and other svstems are available elsewhere 

 (20). 



OUTPUT 

 AS 7„ OF 

 INPUT I 



I Constant Amplitude 

 Input 

 I 



f =- 



ZTTRC 



100 10 



f lOf lOOf 



FREQUENCY 



FIG. I. Upper: circuit for low-frequency attenuation. Lower: 

 output of circuit of type shown above. [Courtesy of VVilliams & 

 Wilkins (12).] 



Experience with spectral phonocardiography sug- 

 gests, and the work of Bekkering and his colleagues 

 (personal communication) appears to establish, that 

 five oscillographic channels, each with different 

 frequency filtration and amplification, and each 

 separated from the next by one octave, can adequately 

 portray cardiovascular sound. 



A schematic representation of a multichannel, 

 multifilter system for oscillographic phonocardiog- 

 raphy is presented in figure 3. The decrement in 

 intensity of cardiovascular sound with frequency is 

 represented by line AB. Band-pass filters may be 

 unnecessary; high-pass filters as shown in figure 3 

 here may suffice, inasmuch as the natural decrement 

 of cardiovascular sound takes care of the upper end of 

 the frequency scale. The system indicated here can 

 theoretically be calibrated and standardized if the 

 amplification in each channel is so adjusted as to 

 bring the registered amplitude to a uniform level, and 

 if the amount of amplification necessary to accom- 

 plish this is recorded. An alternative to the use of 

 different amplification in each channel as the findings 

 in the individual patient may require is to apply a 

 different and somewhat arbitrarily selected amplifica- 

 tion in each channel, let us say o, 12, 24, 48, and 60 

 db in the five schematized in figure 3, keeping said 

 amplification the same in all patients studied. 



Much of the information represented in a single 

 spectral (SPCG) display can be derived from such a 

 system as this. For example, improved representation 

 of the shape of murmurs useful in diagnosis and in 

 hemodynamic correlation, and of the splitting of heart 

 sounds, tends to appear in the channels with higher 



