Method and apparatus for displaying body sounds and performing diagnosis based on body sound analysis Number:7,520,861 from the United States Patent and Trademark Office (PTO) owispatent

Title: Method and apparatus for displaying body sounds and performing diagnosis based on body sound analysis

Abstract: A lung sound diagnostic system for use in collecting, organizing and analyzing lung sounds associated with the inspiration(s) and expiration(s) of a patient. The system includes a plurality of transducers that may be placed at various sites around the patient's chest. The microphones are coupled to signal processing circuitry and A/D converters which digitize the data and preferably provides the digital data to a computer station. A data collection and organization program, executing on the computer station, organizes and formats the data into a combination display for display or printing. The combinational display includes at least two display elements. In a first display element, the data is shown for both inspiration and expiration combined in a first time scale. In a second display element, the data for inspiration and expiration are shown individually in a second time scale that is time-expanded relative to the first time scale. The system may also include application programs for detecting and classifying abnormal sounds. The resulting information may be displayed in a variety of formats to facilitate diagnosis. Additionally, the system may include an analysis program for comparing selected criteria corresponding to the detected abnormal sounds with predefined thresholds in order to provide a likely diagnosis.

Patent Number: 7,520,861 Issued on 04/21/2009 to Murphy

---

Inventors:	Murphy; Raymond L. H. (Wellesley, MA)
Appl. No.:	10/884,398
Filed:	July 6, 2004

---

Related U.S. Patent Documents

---

	Application Number	Filing Date	Patent Number	Issue Date
	10080209	Feb., 2002	6790183
	09699546	Oct., 2000	6394967
	09172343	Oct., 1998	6139505

---

Current U.S. Class:	600/529 ; 600/586
Current International Class:	A61B 5/00 (20060101); A61B 7/00 (20060101)

---

References Cited [Referenced By]

---

U.S. Patent Documents

3990435	November 1976	Murphy
4063550	December 1977	Tiep
4267845	May 1981	Robertson, Jr. et al.
4672977	June 1987	Kroll
4928705	May 1990	Sekhar et al.
4951678	August 1990	Joseph et al.
4991581	February 1991	Andries
5010889	April 1991	Bredesen et al.
5035247	July 1991	Heimann
5165417	November 1992	Murphy, Jr.
5213108	May 1993	Bredesen et al.
5218969	June 1993	Bredesen et al.
5301679	April 1994	Taylor
5309922	May 1994	Schechter et al.
5329932	July 1994	Yount
5718227	February 1998	Witlin et al.
5825895	October 1998	Grasfield et al.
5844997	December 1998	Murphy, Jr.
6005951	December 1999	Grasfield et al.
6116241	September 2000	Huygen et al.
6139505	October 2000	Murphy
6152884	November 2000	Bjorgaas
6168568	January 2001	Gavriely
6394967	May 2002	Murphy

Foreign Patent Documents

	201 12 390		Nov., 2001		DE
	WO-00/56218		Sep., 2000		WO

Other References

Earis, J. Woodhouse, N. and Duffy, N., Reproducibility of Crackle Counts in Normals and Patients With Stable Fibrosing Alvelotis . . . , Sep. 2001. cited by other . Kompis, M. and Wodicks, G., Spatial Reconstruction of Acoustic Sources in the Thorax, Oct. 1995. cited by other . Ishikawa, S., Gomez, F., Vyshedskiy, A., Elmaghhiraby, Z., MacDonnell, K.F. and Cell B., "Expiratory Wheeze" in a Patient with "Vocal Cord Dysfunction", Sep. 2001. cited by other . Bergstresser, T., Vyshedskiy, A. and Murphy, R., Sound Speed in the Lung as a Function of Lung Volume, Presented at the annual meeting of the International Lung Sound Association, Berlin, Sep. 2001. cited by other . Murhphy, R., Vyshedskiy, A., Ramirez, A., Brockington, G., Power, V., Gopal, M., Cohen, J. and Paciej, J., Objective Measurement of Wheezes and Crackles in Congestive Heart Failure and Bronchial Asthma, Presented at the annual meeting of the iInternational Lung Sound Association, Berlin, Sep. 2001. cited by other . Murphy, R., Vyshedskiy, A., Power, V-A, Bergstrom, K. and Murphy, M., Objective Measurement of Lung Sounds in Chronic Obstructive Lung Disease, Presented at the annual meeting of the European Respiratory Society, Berlin, Sep. 2001. cited by other . Murphy, R., Vyshediskiy, A., Marinelli, P., Bixby, M. Bergstrom, K., V-A Power, Computerized Lung Sounds in Assessment of ICU Patients, Presented at the Annual Meeting of the American Thoracic Society, 2001. cited by other . Davidson, F., House, C., Power, V-A, Bergstrom K., Wilson, C. and Vyshedskiy, A., Lung Sound Amplitude and Tidal Volume During Positive Pressure Ventilation, Presented at the annual meeting of the International Lung Sound Association, Berlin, 2001. cited by other . Murphy, R., Davidson, F., House, C., Power, V-A, Bergstrom K., Wilson, C. and Vyshedskiy, A., The Relationship of Lung Sound Amplitude and Tidal Volume During Positive Pressure Ventilation, Presented at the annual ACCP meeting, San Franscisco, 2000. cited by other . Murphy, R., Bergstrom, K. and Mylott, Characteristics of Lung Sounds in Patients with Pneumonia and Congestive Heart Failure, Presented at the annual meeting of the International Lung Sound Association, 1997. cited by other . Ishikawa, S., Murphy, M. and Murphy R., "E"-Change in Normal, COPD and Asthmatic Patients, Presented at the 25th International Conference on Lung Sounds, Chicago, IL, Sep. 2000. cited by other . Bergstresser, T., Vyshedskiy, A. and Murphy, R., Sound Speed in the Lung as a Function of Lung Volume, Presented at the 25th International Conference on Lung Sounds, Chicago, IL, Sep. 2000. cited by other . Kiyokawa, H. and Pasterkamp, H., Respiratory phase affects the time delay of lung sounds between adjacent sensors, Presented at the 25th International Conference on Lung Sounds, Chicago, IL, Sep. 2000. cited by other . Tomomasa, Takeshi et al., "Gastrointestinal Sounds and Migrating Motor Complex in Fasted Humans," American Journal of Gastroenterology, New York, NY, US, vol. 94, No. 2, XP002931927, Feb. 1999, pp. 374-381. cited by other . Supplementary European Search Report, European Application No.: 03713564.7-2305, PCT/US0305140, Applicant: Murphy, Raymond, L., H., Date of Mailing: Nov. 29, 2006, pp. 1-5. cited by other.

Primary Examiner: Nasser; Robert L
Attorney, Agent or Firm: Cesari and McKenna LLP Reinemann; Michael R.

---

Parent Case Text

---

This application is a divisional of application Ser. No. 10/080,209 titled METHOD AND APPARATUS FOR DISPLAYING LUNG SOUNDS AND PERFORMING DIAGNOSIS BASED ON LUNG SOUND ANALYSIS, filed on Feb. 21, 2002, now U.S. Pat. No. 6,790,183 which is a continuation-in-part of application Ser. No. 09/699,546, filed Oct. 30, 2000, now U.S. Pat. No. 6,394,967, which is a continuation of application Ser. No. 09/172,343, filed Oct. 14, 1998, now U.S. Pat No. 6,139,505.

---

Claims

---

What is claimed is:

1. A method comprising: receiving digital data corresponding to lung sounds detected during the inspiration and expiration of a patient having a chest by a plurality of transducers placed at various sites around the patient's chest; and generating, in response to the digital data, a display on a graphical user interface, the display having a display element in which the detected lung sounds from a plurality of transducers are plotted in a vertically stacked configuration, wherein the detected lung sounds from at least one transducer is plotted as a function of frequency over time.

2. The method of claim 1 wherein the display element may be one of a plurality of colors and the method further comprises modifying the color of the display element to reflect intensity of a detected sound at a given time and frequency.

3. The method of claim 2 wherein the sound intensity is indicated by one or more colors.

4. The method of claim 1 wherein in the plotted frequency ranges from approximately zero to 500 Hz.

5. The method of claim 1, further comprising: analyzing the data to detect one or more adventitious sounds.

6. The method of claim 5, wherein the one or more adventitious sounds are selected from the group consisting of: crackles, wheezes, and rhonchi.

7. The method of claim 1, further comprising: generating a proposed diagnosis based on the data.

8. A method comprising: receiving digital data corresponding to lung sounds detected during the inspiration and expiration of a patient having a chest by a plurality of transducers placed at various sites around the patient's chest; generating, in response to the digital data, a display on a graphical user interface, the display having a display element in which the detected lung sounds from a plurality of transducers are plotted in a vertically stacked configuration; and locating the approximate physical coordinates of the detected sounds through the analysis of the sounds detected by the plurality of transducers.

9. The method of claim 8, wherein the approximate physical coordinates are three dimensional.

10. The method of claim 8, wherein the display is a visual display of sound locations on a body corresponding to the approximate physical coordinates.

11. A method comprising: receiving digital data corresponding to lung sounds detected during the inspiration and expiration of a patient having a chest by a plurality of transducers placed at various sites around the patient's chest; generating, in response to the digital data, a display on a graphical user interface, the display having a display element in which the detected lung sounds from a plurality of transducers are plotted in a vertically stacked configuration; inputting sound to the patient from an input device; monitoring a speed of the sound to reach a transducer; and determining characteristics of the patient based on the speed of sound from the input to the transducer.

12. A method comprising: receiving digital data corresponding to lung sounds detected during the inspiration and expiration of a patient having a chest by a plurality of transducers placed at various sites around the patient's chest; generating, in response to the digital data, a display on a graphical user interface, the display having a display element in which the detected lung sounds from a plurality of transducers are plotted in a vertically stacked configuration; inputting sound to the patient from an input device; monitoring a resonance of the sound at a transducer; and determining characteristics of the patient based on the resonance of the sound at the transducer.

13. A method comprising: receiving digital data corresponding to lung sounds detected during the inspiration and expiration of a patient having a chest by a plurality of transducers placed at various sites around the patient's chest; generating, in response to the digital data, a display on a graphical user interface, the display having a display element in which the detected lung sounds from a plurality of transducers are plotted in a vertically stacked configuration; and distinguishing the detected sounds between interstitial pulmonary fibrosis and congestive heart failure.

14. A method comprising: receiving digital data corresponding to lung sounds detected during the inspiration and expiration of a patient having a chest by a plurality of transducers placed at various sites around the patient's chest; generating, in response to the digital data, a display on a graphical user interface, the display having a display element in which the detected lung sounds from a plurality of transducers are plotted in a vertically stacked configuration; and dividing the data into phases of inspiration and expiration for analysis.

15. The method of claim 14, wherein the dividing is performed manually by an operator.

---

Description

---

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates generally to non-invasive diagnostic systems and techniques, and more specifically, to a method and apparatus for diagnosis based upon the review and analysis of body sounds.

2. Background Information

Since the time of its invention in the early 1800's, the stethoscope has been used routinely by physicians to amplify sounds in the human body. The physician typically places the chest piece of the stethoscope against the patient's skin and listens through the stethoscope's earpieces. By monitoring a patient's breathing, a physician may detect the existence of adventitious (i.e., abnormal and/or unexpected) lung sounds. The identification and classification of adventitious lung sounds, moreover, often provides substantial information about pulmonary and associated abnormalities.

Adventitious lung sounds may be classified into two major types: crackles (or rales), which are discontinuous (i.e., interrupted) sounds, and wheezes and rhonchi, which are continuous. Crackles may be further classified as coarse, medium or fine, depending on their frequency, characteristics and amplitude. Wheezes may be similarly classified as sibilant or sonorous. An experienced and knowledgeable physician, moreover, may be able to diagnose certain pulmonary diseases, such as pneumonia, asthma, etc., simply by detecting, identifying and noting the location of particular adventitious lung sounds.

Lung sounds may also be recorded and displayed to assist in the detection and identification of adventitious sounds. For example, U.S. Pat. No. 3,990,435, entitled BREATH SOUND DIAGNOSTIC APPARATUS to Raymond L. H. Murphy, Jr., the inventor herein, discloses a system for providing a time-expanded visual display of lung sounds. That is, the time scale of the tracing or waveform detected by a microphone, normally plotted at approximately 25-50 mm/sec. by standard medical strip charts, is expanded to approximately 800 mm/sec. Expanding the time scale of the waveform significantly improves the physician's ability to detect and identify adventitious sounds.

Devices to analyze recorded lung sounds are also known. For example, U.S. Pat. No. 5,010,889, entitled INTELLIGENT STETHOSCOPE to Bredesen et al., discloses a stethoscope capable of digitizing and storing body sounds, including heart and lung sounds, in a memory structure configured to store up to six different sounds. The stethoscope includes a single chest piece with a microphone, which may be moved to one of six locations around the patient's chest. The stethoscope further includes a liquid crystal display (LCD) panel for displaying the waveform of a recorded sound.

Using waveform signature analysis, each of the six recorded waveforms is examined to determine the presence of high-pitch sounds which may correspond to fine crackles or low-pitch sounds which may correspond to coarse crackles. The presence or absence of these sounds is then formed into an array that may be compared with pre-recorded arrays corresponding to known conditions, e.g., normal lung sounds, pneumonia, etc. If a match is found between the recorded waveforms and one of the pre-recorded arrays, a diagnosis may be displayed on the LCD panel of the stethoscope.

Although Bredesen's intelligent stethoscope represents an improvement in diagnostic tools, especially for physicians lacking extensive experience in detecting and identifying adventitious lung sounds, it nonetheless has several disadvantages. First, the intelligent stethoscope has only a single microphone, so that obtaining recordings at multiple locations is time-consuming. A single microphone also makes it impossible to record a given sound (e.g., a particular inspiration or expiration) from more than one point on the chest. Second, the small LCD panel is capable of displaying only a single waveform in one predefined format and is provided simply to determine whether valid data has been obtained. Due to these limitations, the intelligent stethoscope is not that likely to provide accurate diagnoses.

SUMMARY OF THE INVENTION

It is an object of the present invention to provide an improved method and apparatus for facilitating the diagnosis of certain diseases based upon recording, review and analysis of body sounds.

It is a further object of the present invention to provide an improved method and apparatus that provides the diagnostician with a richer, more fully coordinated set of data for rapidly and accurately detecting body sound abnormalities.

Another object of the present invention is to provide a system configured to generate graphical displays of detected abnormal body sounds to facilitate diagnosis.

A still further object of the present invention is to provide a system for automatically providing an accurate diagnosis based upon an analysis of recorded body sounds.

Briefly, the invention relates to a system for recording, displaying and analyzing body sounds to facilitate the diagnosis of various diseases. The system includes a plurality of transducers, such as microphones, that may be placed at preselected sites around a patient's chest. The transducers detect the sound or vibration of the body at these sites. The system also includes signal processing circuitry for conditioning and converting analog signals generated by the transducers into digital data. Additionally, the system includes a computer station coupled to the signal processing and digitizing circuitry. The computer station includes a processor, input/output circuitry, a data storage device, at least one input device, such as a keyboard or a mouse, and a graphical user interface. The system may further include a printer. Executing on the computer station is a first application program that collects and organizes the data for display on the graphical user interface and/or for printing.

More specifically, a plurality of transducers are preferably utilized simultaneously to obtain sound information from the patient. In response to the patient's inspiration and expiration, each transducer generates analog signals that are conditioned and digitized by the signal processing circuitry and stored by the computer station at the data storage device. The first application program organizes the received data from all sites for simultaneous display on the graphical user interface and/or printing in multiple time scales preferably in a vertical stack arrangement, such that all of the information may be reviewed concurrently by an attending physician. The first application program may also display the data in frequency versus time format. In addition, by comparing the displayed or printed combinational data with predefined criteria or guidelines, an accurate diagnosis may be reached.

In a further embodiment of the present invention, a second application program, also executing on the computer station, analyzes the data recorded by the transducers. In particular, the second application program preferably includes means for identifying and counting the number and time of occurrence of adventitious sounds, such as wheezes, rhonchi and crackles, and categorizing the identified crackles as fine, medium or coarse. The second application program may also include means for performing other quantitative analysis, such as the ratio of duration of inspiration to expiration and statistical analysis of the intensity of the recorded sounds. This information may then be provided to the attending physician in a variety of ways. For example, it may be displayed in tabular format or graphically in relation to the point on the patient's chest at which the abnormal sound occurred.

A third application for generating a possible diagnosis may also be included. The third application may be a data analysis program, such as a neural network module or a statistical analysis module using multiple logistic regression models, that interoperates with a database of pre-classified lung sounds. Specifically, the database preferably includes multiple data sets for normal lungs sounds and lung sounds associated with specific diseases, such as COPD, asthma, and IPF. The database may be used to train a neural network classifier or to perform a statistical classification. The neural network module analyzes various quantities computed from the patient's lung sounds in view of the training database and, if a match of sufficient reliability is found, presents a preliminary diagnosis and corresponding probability.

A fourth application for automatic localization of the origin of adventitious and normal sounds may also be included. Preferably, the localized sounds are displayed in three-dimensions (3D) using preselected images or glyphs for the corresponding adventitious sounds that have been detected.

In addition, sounds can be input to the patient. These input sounds can then be detected by the plurality microphones disposed around the patient and the information can be displayed and analyzed.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and further advantages of the invention may be better understood by referring to the following description in conjunction with the accompanying drawings, in which:

FIG. 1 is a block diagram of a system for implementing a preferred embodiment of the present invention;

FIG. 2 is a block diagram of the computer station of FIG. 1 illustrating the relationship of an operating system and several application programs;

FIG. 3 is a flow diagram of the steps performed by a data collection and organization application program;

FIG. 4 is a representative display of data for a single location;

FIGS. 5A and 5B are a highly schematic representation of a combinational display of information;

FIGS. 6A-C are exemplary plots of lung sound data versus time illustrating the appearance of several adventitious sounds;

FIG. 7 is a highly schematic representation of another combinational display of information;

FIG. 8 is a flow diagram of the steps performed by an adventitious sound detection application program;

FIGS. 9A and 9B are data plots that may be generated by the adventitious sound detection application program;

FIG. 10 is a highly schematic illustration of a graphical display showing the points at which abnormal sounds were detected;

FIG. 11 is a highly schematic illustration of a graphical display showing sound intensity levels as determined by the present invention;

FIGS. 12 and 13 are exemplary displays of lung sound data in frequency versus time format;

FIGS. 14A-14B is a flow diagram of a preferred localization method in accordance with an aspect of the present invention;

FIG. 15 is a highly schematic illustration of an abnormal sound segment as detected at a plurality of microphone channels in amplitude versus time format;

FIGS. 16 and 17 are representative three-dimensional displays of adventitious sounds in accordance with an aspect of the present invention;

FIG. 18 is an exemplary representation of an input sound signal in accordance with an aspect of the present invention;

FIG. 19 is an exemplary representation of an input sound signal and a detected sound signal in amplitude versus time format; and

FIGS. 20-22 are a highly schematic illustrations of a microphone cassette in accordance with another aspect of the present invention.

DETAILED DESCRIPTION OF AN ILLUSTRATIVE EMBODIMENT

FIG. 1 is a block diagram of the lung sound recording and analysis system 100 of the present invention. The system 100 includes a sensor system 101 which includes a plurality of sound transducers, such as analog microphones 102, that may be placed at various sites around the chest or other area of a patient 104. In the preferred embodiment of the invention, the system 100 uses sixteen different sites, of which, fifteen are located around the chest and one is located at the patient's trachea. More specifically, there is one site on the left side, one site on the right side, two sites on the upper front chest separated by the spinal column (proximate to the top portion of the lungs), one site on the lower right front chest, two sites on the upper back (proximate to the top portion of the lungs), four sites in the middle back (proximate to the mid portion of the lungs), four sites at the lower back (proximate to the bottom of the lungs) and one site at the trachea. It should be understood that other chest sites may be utilized by the system 100. Furthermore, a simpler system may use nine sites, eight around the chest and one at the trachea. The eight chest sites may include two on the upper front chest (separated by the spinal column), one on each side and four on the back (two upper and two lower) each pair separated by the spinal column.

Additionally, sixteen microphones 102, one located at each of the sites, are preferably utilized concurrently by the system 100 during the data collection process, although fewer are shown in FIG. 1 for clarity. This allows the data from all sites to be collected concurrently. Nonetheless, a simpler system may utilize one microphone 102 positioned sequentially at the nine or more sites for data collection and the data collection process repeated at each site. To isolate the microphones 102 from external sounds, they may be embedded in the chest pieces of conventional stethoscopes (not shown). The microphones 102 may also be taped or applied with suitable strapping to the patient 104 to prevent dislocation or movement during the data acquisition process.

Leads 106 extending from each microphone 102 are used to connect the microphones 102 to a signal conditioning circuit 108. In general, the signal conditioning circuit 108 modifies the analog audio signals generated by the microphones 102 in order to remove unwanted noise and boost the signal strength for subsequent digitizing. A suitable signal conditioning circuit for use in the present invention is disclosed in U.S. patent application Ser. No. 08/729,272, filed Oct. 10, 1996, entitled "Method And Apparatus For Locating The Origin Of Intrathoracic Sounds," now U.S. Pat. No. 5,884,997, the specification of which is hereby incorporated by reference in its entirety.

It should be understood that the sensor system 101 may utilize other sound transducers besides analog microphones. System 101 may use, for example, digital microphones, one or more lasers configured to scan the selected sites, accelerometers, etc.

The outputs from the signal conditioner 108 (i.e., processed audio signals from each microphone 102) are provided to a computer station 110. The computer station 110, which may be implemented, at least in part, using a personal computer or workstation, includes a central processing unit (CPU) 112 coupled to a memory 114 and input/output circuitry 116 by a bidirectional bus 118. The memory 114 typically comprises random access memory (RAM) for the temporary storage of information, including application programs and an operating system, and read only memory (ROM) for permanent storage of the computer's configuration and basic operating commands. The operating system controls the operations of the CPU 112. The computer station 110 also includes a sound card 119 connected to a speaker 121 for generating output sounds as discussed in more detail below.

The I/O circuitry 116 preferably connects the computer station 110 to a digital storage device 120, such as a disk drive or removable digital storage media, for storage and retrieval of data as described below. The I/O circuitry 116 also connects the computer station 110 to cursor/pointer control and input devices, such as a mouse 122 and a keyboard 124. A window-based graphical user interface 126 and a printer 130 are also preferably connected to the I/O circuitry 116 of the computer station 110. The input/output circuitry 116 preferably contains the necessary hardware, e.g., buffers and adapters, needed to interface with the control devices 122, 124, the graphical user interface 126, printer 130, memory 114 and digital storage device 120.

The computer station 110 may be a personal computer of the IBM.RTM. series of computers sold by International Business Machiness or the Macintosh.RTM. series of computers sold by Apple Computer Inc. These computers have resident thereon, and are controlled and coordinated by, operating system software, such as IBM OS2.RTM., Microsoft Windows.RTM. or Mac OS 9 operating systems. It should be understood that the system 100 may also be implemented on other computer platforms, such as UNIX-based workstations manufactured and sold by Hewlett Packard Co. of Palo Alto, Calif., or hand-held computers, such as those running the Palm or Win CE operating systems, among others.

As mentioned above, the signal conditioning circuit 108 is preferably connected to the computer station 110 such that the processed audio signals from each microphone 102 are received by computer station 110. Specifically, each output of signal conditioning circuit 108 (which preferably corresponds to a particular microphone 102) is connected to an analog-to-digital ("A/D") converter 132 that may be part of the computer station 110. The A/D converter 132 converts the processed analog audio information into a digital data stream. The sampling rate of the A/D converter 132 is preferably greater than 8000 samples per second and the bit rate is preferably greater than eight bits per sample.

Additionally, the A/D converter 132 synchronously pairs a master time signal from a system clock 133 which, for example, may be included in the A/D converter 132 or be internal to the CPU 112 with the digital audio information corresponding to each microphone 102. The master time signal provides a uniform time index to the signals received from the microphones 102. The paired digital-audio/time information associated with each microphone 102 is then forwarded to the digital storage device 120. The A/D converter 132 is preferably a multi-channel, high bandwidth data acquisition printed circuit board, such as those manufactured and sold by Keithley Metrabyte, Inc. It should be understood that the CPU 112, rather than the A/D converter 132, may pair the master clock signal to the digital audio information. It should be further understood that digital transducers, rather than analog microphones, may be utilized. The use of digital transducers would obviate the need for an A/D converter. Although the analog signal is digitized in the preferred embodiment, those skilled in the art will recognize that the analog signal can be used as well.

FIG. 2 is a highly schematized illustration of the computer station 110 illustrating the interaction of several software elements, including a data collection and organization application program 202, an adventitious-sound detection program 203, a probable-diagnosis prediction program 204, an automatic localization program 205, and an operating system 206. The application programs 202-204 execute on the computer station 110. Interacting with the probable-diagnosis prediction program 204, moreover, is a training database 207. The application programs 202-204 and the operating system 206 interact, as shown by arrows 208, 209, 211, and 212, via system calls to control the operations of the computer station 110.

Included within the operating system 206 are system facilities, including a window manager 214 and a printer manager 216, which, inter alia, implement at least some of the system calls. Lower layers of the operating system 206 (or the computer station 110) may also include device drivers, such as a display driver 218 and printer driver 220. Drivers 218, 220 interface directly with hardware components, such as the graphical user interface 126 and the printer 130, respectively.

It should be understood that computer station 110 may include additional application programs. It should be further understood that the computer station 110 may omit the adventitious-sound detection and/or the probable-diagnosis prediction programs 203, 204.

Data Collection and Organization

In operation, the microphones 102 are preferably taped or strapped to the patient's skin at the sixteen sites. Next, the system 100 is initialized and the data collection and organization application program 202 is preferably opened. FIG. 3 represents a flow chart of operations 300 performed by the data collection and organization application program 202 (FIG. 2). As shown by block 302, the data collection and organization program 202 first requests patient identifying information (such as name, identification number, physician, etc.), which may be entered by a system operator through the keyboard 124 or mouse 122. This information may be displayed on the graphical user interface 126 in a data collection window (not shown). Next, the patient is instructed to breath in (inspiration) and out (expiration) several times. While the patient breathes, lung sounds detected by the microphones 102 are converted to audio signals and provided to the signal conditioning circuit 108. Preferably, data is continuously received for a sufficient period of time (e.g., ten seconds) to ensure that useful data is obtained for at least one inspiration/expiration pair. As indicated at block 304, the audio signals are measured, conditioned and provided to the analog-to-digital converter 132 for digitization. The digital data for each site is then stored at either the memory 114 of computer station 110 or the digital storage device 120. If fewer microphones are used, they may be situated at the next location(s) and the process repeated.

Next, the data collection and organization application program 202 retrieves the data for display or printing, as shown by block 306. Specifically, the data collection and organization application program 202 interacts with the operating system 206 so as to retrieve the data corresponding to each microphone site from memory 114 or the storage device 120. The data for each site, which represents both inspiration and expiration combined, is preferably displayed on screen 228. A preferred form of display is described below.

As indicated by block 308 (FIG. 3), the system operator then selects a particular inspiration and expiration for further analysis by identifying the corresponding starting and stopping points of the selected inspiration and expiration. Preferably, this is accomplished with the aid of a display formed in accordance with the present invention and illustrated in FIG. 4. FIG. 4 is a highly schematic representation of a preferred display 400 of data obtained from sixteen sites. The display 400 preferably includes a set of body maps 402-405, which illustrate the various sites at which data was recorded, and a data plot area 408, which contains an illustration of the recorded data. More specifically, the data plot area 408 includes the actual data tracings (i.e., signal tracings) obtained at each microphone 102 (FIG. 1) and preferably includes a corresponding time axis 410. The data in data area 408 corresponds to both inspiration and expiration combined and is preferably a plot of signal amplitude (e.g., millivolts or decibels) from the microphones 102, each associated with a particular channel number, versus time in seconds as shown by time axis 410. It should be understood that the display 400 may include other areas, such as a patient data area 412, a command bar area 414 and a root-mean-square (RMS) field 416.

To mark the starting and stopping points of the selected inspiration and expiration, the system operator moves a pointer (not shown) associated with the mouse 122 across the data area 408 to the start of inspiration and executes a mouse "click" at that location, thereby associating a particular time (based on the corresponding point on the time axis 410) with the start of inspiration. The data collection and organization application program 202 preferably includes conventional means to associate the position of the pointer with the time value vertically aligned therewith upon execution of the mouse click. The system operator similarly associates respective times with the end of inspiration and with the starting and stopping times of expiration. The starting and stopping points for inspiration and expiration are best identified by examining the signal or tracing recorded at the trachea microphone (i.e., microphone channel 16). In particular, inspiration is typically associated with a first continuous, high amplitude segment 416 of the trachea signal. When the amplitude of the segment 416 diminishes to near zero, inspiration is typically at an end. Expiration is similarly associated with a continuous, high amplitude segment 418 that directly follows inspiration segment 416. When the amplitude of this second continuous segment 418 diminishes to zero, expiration is typically complete.

It should be understood that the data collection and organization program 202 may as an alternative, or a supplement, to operator selection, include one or more modules or routines that automatically identify the starting and stopping points of inspiration and expiration in a similar fashion.

Following the identification of the starting and stopping points of a selected inspiration and expiration, the data collection and organization program 202 proceeds to organize the corresponding data for display as a function of time. In particular, as indicated by block 310 (FIG. 3), the program 202 preferably plots the data corresponding to each site for both inspiration and expiration in a time-expanded format. Execution of the time expansion function is preferably in accordance with the description set forth in U.S. Pat. No. 3,990,435, which is hereby incorporated by reference in its entirety. Specifically, the data collection and organization program 202 preferably generates two increments of time-expanded data: (i) slightly time-expanded and (ii) fully time-expanded. In particular, the data collection and organization program 202 modifies a copy of the data for each site obtained at step 304 so that it may be displayed or printed in a slightly time-expanded scale (e.g., on the order of 200-400 mm/sec.) and in a fully time-expanded scale (e.g., on the order of 800 mm/sec.) in addition to the more conventional, non-expanded time scale of around 20-50 mm/sec. As block 312 indicates, the data collection and organization program 202 then displays and/or prints-out the data corresponding to inspiration and expiration for each site (which is now maintained in three formats: (i) unexpanded, (ii) slightly time-expanded and (iii) fully time-expanded), in a common display, where it can be viewed in all formats simultaneously. Those skilled in the art will recognize that multiple degrees of expansion can be performed.

FIGS. 5A and 5B illustrate a highly schematic representation of a preferred combinational display or print-out 500 of lung sound data generated by the data collection and organization program 202. The combinational display 500 includes first, second and third plot elements 502, 504 and 506, respectively, and a data field 508. Each plot element 502, 504 and 506 includes a data or signal trace (e.g., trace 510 in first plot element 502) of the amplitude of the detected lung sounds (vertical axis) for each microphone as a function of time (horizontal axis). As mentioned above, the microphones 102 (FIG. 1) may be identified by channel (e.g., channels one through sixteen) at least one of which (e.g., channel sixteen) corresponds to the patient's trachea. Channels one to fifteen preferably represent the fifteen microphones each located at a different site around the patient's chest, as described above. The first plot element 502 represents the lung sound data for all inspirations and expirations over the predefined time period (e.g., ten seconds) combined in an unexpanded time scale. That is, the data is formatted for display at approximately 20-50 mm/sec. The time period represented by the first plot element 502 is preferably selected so that at least one set of inspiration and expiration data of sufficient quality is obtained. The previously selected inspiration and expiration are preferably enclosed within blocks located in the first plot element 502. In particular, the selected inspiration is enclosed in a first block 512 and the selected expiration is enclosed in a second block 514.

Second plot element 504 represents the lung sounds corresponding to the selected inspiration as detected by each microphone in a slightly time-expanded format. That is, the data is displayed on an approximately 200-400 mm/sec. scale. Depending on the length of the selected inspiration, second plot element 504 may comprise more than one (e.g., two) panels 504a and 504b. Similarly, the third plot element 506 represents the lung sounds corresponding to the selected expiration for each microphone also in a slightly time-expanded format. Since the selected expiration was adjacent to the selected inspiration, the time scale (horizontal axis) for the third plot element 506 continues on from the time scale for the second plot element 504. The third plot element 506 may also be represented by multiple panels, such as panels 506a, 506b and 506c, depending on its length. The signal tracings within the first, second and third plot elements 502, 504 and 506, moreover, are preferably arranged in a vertical stack configuration relative to each other.

The data field 508 preferably includes several computed quantities as determined from the selected inspiration and expiration information. More specifically, the data field 508 preferably contains a root-mean-square (RMS) value calculated in a conventional manner for each channel during the selected inspiration and expiration. The RMS values may be provided in column format adjacent to the first plot element 502. In addition, data field 508 may include the length of time of the selected inspiration 512 and the selected expiration 514, preferably in seconds. Data field 508 may further include other computed statistical quantities identified as R1, R2, R3, R4, R5, R6, R7 and R8, that are of interest to the attending physician.

For example, R1 may represent the ratio of time of selected inspiration to time of selected expiration. R2 may be the ratio of R1 to the average RMS value at the trachea during inspiration. R3 may be the ratio of the average RMS value during inspiration for the microphones located on the chest to the RMS value for the trachea during inspiration. R4 may be the standard deviation of the RMS values during inspiration for all of the channels. R5 may represent the ratio of the mean interchannel non-homogeneity of the start of inspiration to the duration of the inspiration at the trachea. R6 may represent the ration of the mean interchannel non-homogeneity of the end of inspiration to the duration of the inspiration at the trachea. R7 may represent the product of inspiratory sounds root mean square (RMS) averaged between chest sites and the duration of inspiration at the trachea (time integrated amplitude). R8 may represent the ratio of sound energy below 80 Hz to that from 80 Hz to 800 Hz.

In a preferred embodiment, adventitious sounds, such as crackles, are subtracted out of the signal before computation of RMS values. This subtraction operation is preferably performed as crackles can be disproportionately loud, e.g., have much higher amplitudes, in comparison with the rest of the signal, thereby resulting in a calculated RMS value that is not necessarily reflective of regional ventilation.

Once the recorded data has been displayed and/or printed in the manner illustrated in FIGS. 5A and 5B, it is preferably reviewed by the attending physician. As shown, the combinational display 500 concisely and effectively presents the data obtained at multiple sites to the attending physician. In particular, examination of plot elements 504 and 506, which represent the slightly time-expanded data, quickly reveals the occurrence of any adventitious sounds. Further review of these elements 504 and 506 provides detail information regarding the existence, identity and location of the adventitious sounds. For example, by simply referring to the corresponding channel number, a physician may quickly ascertain at which location an adventitious sounds was recorded. By arranging the signal tracings in a vertical stack as shown in the combinational display 500, he or she may also judge whether the same event produced adventitious sounds detected at more than one location. This may all be performed, moreover, without having to switch back-and-forth between a plurality of screens or sheets. That is, the data collection and organization application program 202 (FIG. 2), in cooperation with the operating system 206, may adjust the size of the plot elements 502, 504 and 506 so that they fit in their entirety in one or two windows on the display screen 228 or on one or two sheets of paper, if printed. Moreover, the system 100 may mark the location of detected adventitious sounds within the second and third plot elements 504 and 506, utilizing a set of abbreviations as identifiers, as described below. For example, "C" stands for a coarse crackle, "M" stands for a medium crackle, "F" for a fine crackle, "W" for a wheeze and "R" for rhonchi.

By comparing the data contained within combinational display 500 with information indicative of various pulmonary conditions or diseases, moreover, the physician may be able to render a diagnosis with a relatively high degree of accuracy. In particular, Table 1 lists the criteria or characteristics of lung sounds associated with four possible conditions: normal, COPD, asthma and IPF, based on empirical studies and analysis of numerous subjects with the indicated conditions. As shown in Table 1, for example, a normal patient's expiration should last about 20% longer than his inspiration. This information, moreover, may be quickly obtained by simply reviewing the R1 value in the data field 508. For a patient suffering COPD, expiration is typically on the order of 60% longer than inspiration.

TABLE-US-00001 TABLE 1 CHARACTERISTICS OF PULMONARY SOUND TRACINGS Normal COPD Asthma IPF Ratio of Time Expiration 20% Expiration 60% Expiration variable of Inspiration longer than longer than typically to Time of Inspiration Inspiration much Expiration (on average) (on average) longer than In- spiration Distribution Relatively high Amplitude of Relatively variable of Sounds over amplitude of sounds during uniform the Patient's sounds during Inspiration distri- Chest Inspiration, variable, but bution of little or no higher than sounds variation amplitude of across the in sound sounds during chest, amplitude Expiration wheezes across the typically chest present Sounds appear random broken, wheezes crackles Occurring irregular typically typically During present present Inspiration Occurrence few wheezes and prominent many of Abnormal rhonchi wheezing, crackles Sounds During typically rhonchi typically Inspiration and present, early may also present Expiration inspiratory be present crackles also common

A diagram of illustrative adventitious or abnormal sounds may also be utilized by the attending physician, in combination with the information contained in Table 1, when reviewing combinational pulmonary display 500 so as to assist in arriving at a diagnosis. FIG. 6A is an exemplary plot of lung sound amplitude (vertical axis) versus time (horizontal axis) for a plurality of microphones illustrating the appearance of crackles. FIG. 6B is a similar exemplary plot illustrating the appearance of a wheeze. FIG. 6C is another exemplary plot illustrating the appearance of Type I and Type II rhonchus.

The preferred combinational display thus portrays the detected sounds so that the presence or absence of the characteristics specified in Table I may be ascertained. In particular, the combinational display portrays the following information: (1) the ratio of inspiration to expiration, preferably as a percentage; (2) the distribution of adventitious sounds over the chest and their relative amplitudes; (3) the occurrence of adventitious during inspiration and expiration; and (4) whether the adventitious sounds are crackles, wheezes or rhonchi. As shown, the preferred display 500 provides all of this information to the attending physician in a coherent, efficient manner.

Nonetheless, it should be understood that other combinational displays may be generated by the system. For example, FIG. 7 is a highly schematic representation of another combinational display or print-out 700 of data by the data collection and organization program 202 (FIG. 2) from block 312. As shown, data from four chest regions (right back, left back, right side and left side) is simultaneously portrayed either on the graphical user interface 128 and/or printed preferably on a single sheet of paper from printer 130. The combinational display 700 is preferably divided into four sections 710-713, each corresponding to a particular chest region at which data was obtained. Within each section 710-713, moreover, may be a graphical body illustration 714a-714d, corresponding to the particular chest region at which the respective data was obtained. For example, graphical illustration 714b, associated with the data in section 711, corresponds to the patient's right back region.

Each section 710-713 preferably includes a representation of the data in multiple time scales and formats. In particular, a first display element 716a-716d, disposed with each section 710-713, respectively, illustrates the data obtained at each microphone for several repetitions of inspiration and expiration combined in an unexpanded time scale. That is, the data is formatted for display at approximately 20-50 mm/sec. A second display element 718a-718d, similarly disposed within each section 710-713, respectively, illustrates the data obtained at each microphone for inspiration only in a slightly expanded scale (e.g., on the order of approximately 200-400 mm/sec.). A third display element 720a-720d illustrates the data obtained by each microphone during expiration only, also in a slightly expanded scale (e.g., on the order of approximately 200-400 mm/sec.).

A fourth display element 722a-722d illustrates inspiration only for each microphone in a fully expanded time scale. That is, the data is displayed on an approximately 800 mm/sec. scale. A fifth display element 724a-724d, corresponding to each section 710-713, respectively, illustrates the data for expiration only, also in a fully expanded scale. In the preferred embodiment, the first through third display elements (i.e., elements 716, 718 and 720) are all preferably arranged side-by-side above display element 722. Additionally, display element 724 corresponding to fully expanded expiration, which is often the longest, is preferably arranged below display element 722 and may wrap around as necessary.

Other arrangements of the display elements 716-724 forming combinational display 700 may also be employed. Nonetheless, all of the display elements 716-724 are preferably arranged so as to be shown simultaneously. That is, all of the display elements 716-724 are preferably arranged to appear on the graphical user interface 128 at the same time and/or printed on a single sheet of paper.

To ensure that the display elements 716-724 of combination pulmonary display 700 are placed on the graphical user interface 128 at the same time and/or preferably printed on a single sheet of paper, the data collection and organization application 202 (FIG. 2), in cooperation with the operating system 206, may adjust the size of the display elements 716-724 so that they will fit in their entirety either on the graphical user interface 128 or on a sheet of paper. Nonetheless, the relative relationships between unexpanded, slightly time-expanded, and fully time-expanded are preferably maintained.

The arrangement of information within the combinational display 500 or 700 facilitates various disease diagnosis by highlighting their distinctive and identifying characteristics to the attending physician. In addition, the above-described procedure, unlike x-rays or exploratory surgery, presents little risk or discomfort to the patient.

The data collection and organization program 202 may also display body, e.g., lung, sounds recorded from all or a portion of the sixteen microphones 102 in power versus frequency and time, as shown in FIGS. 12 and 13. FIG. 12 illustrates a display or spectrogram 1200 for eight microphone channels 1202a-h where time is on the horizontal axis, frequency is on the vertical axis and signal power is shown by color and/or color intensity. The frequency range for each microphone channel is 0 to 500 Hz. Nonetheless, those skilled in the art will recognize that other ranges may be displayed. The display 1200, which extends for 20 continuous seconds, shows a plurality of inspirations and expirations by the patient. An exemplary expiration is designated generally 1204 and an exemplary inspiration is designated generally 1206. The intensity of sound at the particular frequency and time may be indicated by color. The colors preferably range from light yellow for low sound intensity to dark red for high sound intensity. An attending physician can identify abnormal sounds by simply reviewing the spectrogram 1200. For example, a wheeze can be identified as a continuous high intensity band (dark red) at is about 125 Hz as seen in channel 1202f (i.e., channel 12).

FIG. 13 is a display or spectrogram 1300 of a single channel, i.e., channel 1202f, corresponding to the same time period as FIG. 12. Here, the frequency range shown on the vertical axis is 0 to 1000 Hz. The display 1300 thus illustrates the frequencies in greater detail. It should be understood that the data collection and organization program 202 may provide one or more drop down menus or buttons on the display screen 126 for selecting the desired display format, e.g., number of microphone channels, range of time, frequency range, etc.

Adventitious-Sound Detection

In a preferred embodiment, the computer station 110 (FIG. 2) further includes an adventitious-sound detection program 203, as mentioned above. The adventitious-sound detection program 203 preferably parses the data recorded by each microphone 102 (FIG. 1) to identify the occurrence of any adventitious sounds, such as crackles, wheeze or rhonchi. The adventitious-sound detection program 203 preferably operates in accordance with the methods and procedures described in U.S. patent application Ser. No. 406,152, titled LUNG SOUND DETECTION SYSTEM AND METHOD, now U.S. Pat. No. 5,165,417, to Raymond L. H. Murphy, Jr., the inventor herein, which is also incorporated by reference herein in its entirety.

More specifically, the consecutive waves of each sound signal are preferably analyzed to determine when a particular wave meets established predefined amplitude and cycle period criteria. Once such a wave is identified, the next adjacent waves are similarly analyzed to determine whether they meet other predefined cycle period and/or amplitude criteria. Depending on the number of consecutive waves that are found to meet particular period and/or amplitude requirements, the adventitious sound detection program 203 may categorize these portions of the signal as crackles, wheeze, rhonchi, or other adventitious sound, depending on the criteria that were utilized in setting the thresholds.

For example, FIG. 8 is a flow chart of operations 800 executed by the adventitious-sound detection program 203. At block 802, the program 203 first generates a corresponding amplitude threshold trace for the data signal (i.e., signal trace) corresponding to each microphone site. To generate the amplitude threshold trace, the program 203 first determines a running average amplitude trace corresponding to the absolute value of the signal over 600 data points (e.g., 300 data points on either side of the data point for which the running average is currently being calculated). As described above, information from the microphones 102 is preferably sampled at 8000 data points per second. The program 203 proceeds to determine the mean of the running average amplitude trace which is then multiplied by an amplitude threshold constant (e.g., 1.5). The amplitude threshold constant is used to distinguish adventitious sounds, such as crackles, from background lung noise. Empirical studies have shown that an amplitude threshold constant of 1.5 is adequate to distinguish crackle events in most cases, although other values may also be employed. The resulting value is added to the running average amplitude trace to form an amplitude threshold trace.

FIG. 9A is a data plot 902 for a particular microphone 102 (FIG. 1) which represents either inspiration or expiration plotted as a function of time. The data plot 902 includes a data signal 910 and a corresponding amplitude threshold trace 912, generated as described above. As shown, the data signal 910 exceeds the amplitude threshold trace 912 at various points (e.g., points A, B, C, etc.). The adventitious-s