Acoustic Mapping and Imaging of Thoracic Sounds

Multichannel lung sound recording over the chest in the front and back has received much attention in the recent years due to its diagnostic potentials in a noninvasive manner. The conventional imaging systems for lungs are chest x-rays, CT, MRI that are well-established methods. The ultrasound method is fairly noninvasive; however, it has not been developed successfully due to the very high attenuation of high-frequency sounds at lung parenchyma.

Traditional auscultation by the stethoscope sounds much less reliable and valuable compared to the imaging methods mentioned above and also pulmonary function tests, pulmonary arteriography, radioisotope scanning, and biopsy of the lung that is used in some instances to make a definite diagnosis of the lung condition. However, auscultation remains as one of the most common devices used by physicians due to its simplicity, noninvasive manner, and availability. Indeed, in the situation of an acute respiratory distress in an emergency room, none of the above mentioned tests are likely to be used first. In such a situation, auscultation can be lifesaving.

Physicians, when listening to the lung sounds of a patient with a respiratory disease, often listen to the sounds over a few locations of the patient's chest both in the front and back. In fact by doing so, they gather information about the sounds coming from different parts of the lung and subjectively try to remember the quality of the sound at every location and assess them together for symptomatic sounds. Therefore, it makes their job much easier if there is a simultaneous multisensor recording over the chest wall so that the sounds can be listened to repeatedly. Furthermore, one can analyze the sounds in the time-frequency domain and have an acoustic image of the lung.

One of the first attempts on this topic was the development of a 16 electronic stethoscopes in a backpad (Fig. 8.1) that can be comfortably placed behind the patient and have the simultaneous digital sounds over the back of the chest wall stored in a computer for further analysis [67]. Over the years, the same group developed a software accompanying the multi-stethoscope jacket to display the sounds at every location and automatically count the number

FIGURE 8.1: Backpad microphones. Adopted from [67] with permission

of symptomatic sounds such as crackles, strides, etc., as well as a three-dimensional (3D) image of the concentration of the symptomatic sounds over the chest so that the physician can detect the site affected by the disease objectively and also much faster.

Using simultaneous multisensor recordings of thoracic sounds from the chest wall, an acoustic imaging of the chest has recently been investigated for detecting plausible different patterns between healthy individuals and patients [76]. In that study, a new method for acoustic imaging of the respiratory system was developed and evaluated by a physical model of the lung as well as experimental data on four subjects and one patient. The sound speed, sound wavelengths at the frequencies of diagnostic values, and the geometry of the lung were taken into consideration in developing a model for acoustic imaging to provide a spatial representation of the intrathoracic sounds as opposed to the mapping of sounds on the thoracic surface. The acoustic imaging model was developed based on the calculation of a 3D data array [76], sk(t -\xk - y | /c) = d,xk-y| x r (y, t)/ \xk - y |2 , where xk (k = 1, ... , M) are the positions of M microphones on the thoracic surface, sk (t) is the sound signal recorded at each microphone, y is the hypothetical source site, c is the sound speed, d is the uniform damping factor per unit length, and r (y, t) is the signal emitted by the hypothetical source (intrathoracic sound). Having the signals recorded by Mmicrophones, the resulted M equations can be solved by the least-squares fit to estimate r (y, t), the hypothetical source. Each equation takes into account the delay between the hypothetical source and microphone (the left-hand side of the equation) and the geometric and assumed uniform damping on the source signal (the right-hand side of the equation).

The above model was tested on two subjects with 8 microphones and two other subjects with 16 microphones and one patient (a 5-year-old child with severe pneumonia) as well as on a physical model of the human lung. The results were congruent with the hypotheses that inspiratory sounds are produced predominantly in the periphery of the lung, while the expiratory sounds are produced more centrally [76]. However, it should be noted that the model was tested and evaluated only on four healthy subjects and the results with one ofthese four subjects was not consistent with the other three in terms of the concentration of the sound source on the upper right anterior portion of the chest wall, though this could be due to the fact that the number of microphones was not the same for all the four healthy subjects and the images obtained by 8 microphones showed substantial differences with those obtained by 16 microphones. In conclusion, based on the resulted images provided in [76], it seems that at least 16 microphones are needed for a reliable acoustic imaging of the chest.

Although, as the authors in [76] state, the acoustic imaging method is unlikely to compete with the CT, x-ray, or MRI techniques in terms of the information that they provide, the results are encouraging as it opens a new diagnostic possibility that upon further improvement in sensor technology, hardware design, and advances in computer technology to reduce the computational cost of the algorithm, may become a routine monitoring technique before using the other conventional methods, due to its noninvasive nature.

Was this article helpful?

0 0

Post a comment