DSP-based Multi-channel Stethoscope

Abstract

Heart diseases have become, since 1985, the second cause of death worldwide [1]. The development of auxiliary diagnosis tools that help healthcare professionals is essential to obtain within a very short range of time efficient information about the health condition of a patient, particularly regarding heart diseases. Currently, most medical procedures are based on clinical information obtained from electronic systems and computers. In this context, it was developed a multi-channel digital stethoscope that includes Internet connection capabilities. Thus, we intend to demonstrate that is possible multiple acquisition of high quality heart sounds and its subsequent transmission via Internet. The main component of the developed hardware is a DSP (Digital Signal Processor). Using a DSP we aim to achieve greater flexibility and reliability by introducing customizable channel equalization in the signal path, in order to obtain signals of high quality audio. Since there are no standard tests to assess the quality of stethoscopes [2], the performed tests focused on the electrical part of the device, leaving to future work its characterization as a whole. This article describes the hardware developed for the heart sound acquisition system, including digital signal processing and communication modules. Tests were made to the equipment electrical characterization based on procedures commonly used in the evaluation of audio equipment.

Full text

Associação Portuguesa de Engenharia de Áudio
Secção Portuguesa da Audio Engineering Society
Artigo Apresentado no 12º Encontro da APEA
8 e 9 de Outubro de 2010 Universidade de Aveiro
O artigo apresentado nesta Conferência foi seleccionado com base num resumo estendido revisto por pelo menos dois revisores
anónimos qualificados. Este artigo foi reproduzido pelo manuscrito previamente fornecido pelo autor, sem qualquer edição,
correcção ou considerações do quadro de revisores. A APEA não se responsabiliza pelo conteúdo apresentado no artigo. Todos os
direitos reservados. É proibida a reprodução total ou parcial do conteúdo deste artigo sem permissão directa da Associação
Portuguesa de Engenharia de Áudio.
DSP-based Multi-channel Stethoscope
Celso Colaço1, Sérgio I. Lopes1, Carlos Abreu1
1ESTG – Instituto Politécnico de Viana do Castelo, 4901-908 Viana do Castelo, Portugal
A correspondência deverá ser endereçada para: Celso Colaço (celsocolaco@ultrawave.eu)
ABSTRACT
Heart diseases have become, since 1985, the second cause of death worldwide [1]. The development of auxiliary
diagnosis tools that help healthcare professionals is essential to obtain within a very short range of time efficient
information about the health condition of a patient, particularly regarding heart diseases. Currently, most medical
procedures are based on clinical information obtained from electronic systems and computers. In this context, it was
developed a multi-channel digital stethoscope that includes Internet connection capabilities. Thus, we intend to
demonstrate that is possible multiple acquisition of high quality heart sounds and its subsequent transmission via
Internet. The main component of the developed hardware is a DSP (Digital Signal Processor). Using a DSP we aim to
achieve greater flexibility and reliability by introducing customizable channel equalization in the signal path, in order to
obtain signals of high quality audio. Since there are no standard tests to assess the quality of stethoscopes [2], the
performed tests focused on the electrical part of the device, leaving to future work its characterization as a whole. This
article describes the hardware developed for the heart sound acquisition system, including digital signal processing and
communication modules. Tests were made to the equipment electrical characterization based on procedures commonly
used in the evaluation of audio equipment.
RESUMO
As doenças do coração tornaram-se, desde 1985, na segunda causa de morte a nível mundial [1]. O desenvolvimento de
sistemas e meios de diagnóstico que permitam aos profissionais de saúde obter de forma rápida e eficiente o máximo de
informação acerca da condição de saúde de um paciente é indispensável, em particular no que diz respeito a doenças do
coração. Actualmente, a maioria dos actos médicos são efectuados com base em informação clínica obtida a partir de
sistemas electrónicos e/ou informáticos. Neste contexto, pretende-se desenvolver um estetoscópio digital multicanal
com capacidade de ligação à Internet. Desta forma, pretendemos demonstrar a viabilidade da aquisição, com elevada
qualidade, de sons cardíacos e posterior transmissão via Internet. O hardware desenvolvido tem como principal
componente um DSP (microcontrolador especializado em processamento digital de sinais). Com a utilização de um
DSP pretende-se obter maior flexibilidade e fiabilidade no processamento dos sons cardíacos tendo em vista a obtenção
de sinais de áudio de elevada qualidade. Uma vez que não existem testes padrão para avaliar a qualidade dos
estetoscópios [2], os testes realizados focaram-se na parte eléctrica do dispositivo deixando para trabalho futuro a sua
caracterização como um todo. Neste artigo descreve-se o hardware de aquisição e processamento dos sons cardíacos
desenvolvido, bem como a caracterização eléctrica do mesmo através de uma série de testes normalmente utilizados na
avaliação de equipamentos de áudio.
1. INTRODUCTION
A significant number of people from rural and suburban
areas do not have access to optimal medical care due to
lack of medical specialists. This happens because
medical specialists choose to work in large urban centers
where the majority of the population is settled. With the
spread of Internet access, web-based medical tools and
services have been growing popularity within the
medical community and, in many cases, Telemedicine
has been seen as the best solution to minimize this
problem [2]. Telemedicine based auxiliary diagnosis

Colaço et al. DSP-based Multi-channel Stethoscope
APEA – 12º Encontro, Aveiro, DETI - Universidade de Aveiro, 8 e 9 de Outubro de 2010
Página 2 de 5
tools can provide quick and efficient information about
the health condition of a patient, in particular regarding
heart diseases. Real-time telemonitoring of vital signs
can be used in emergency scenarios to provide more and
better information to support medical decisions [3].
The first medical tool used by medical students and one
of the most used in the World, especially in the poorest
regions, is the stethoscope [2]. The analysis of heart (and
pulmonary) sounds gives to experienced physicians
valuable information about the health condition of a
patient [3]. However, usage in real-time scenarios is a
difficult and challenging task due to the real-time
connectivity, transmission speed and sound quality [2].
This work describes the development of a multi-channel
digital stethoscope that includes Internet connection
capabilities for real-time patient telemonitoring. Our goal
is to develop a system that can be used by an expert
physician to monitor one patient at a distant location.
This work is a proof-of-concept prototype developed as a
joint project within the courses of “Sensors and
Instrumentation” and “Digital Systems and
Microcontrollers II” (“Sensores e Instrumentação” e
Sistemas Digitais e Microcontroladores II) of the
graduation in Electronic and Computer Networks, School
of Technology and Management (Escola Superior de
Tecnologia e Gestão – ESTG), Polytechnic Institute of
Viana do Castelo (Instituto Politécnico de Viana do
Castelo – IPVC).
2. ACOUSTIC STETHOSCOPE
Taking a look into an acoustic stethoscope will give a
basic idea about the technical specifications and aspects
for the digital stethoscope. Acoustic stethoscopes are
composed of a chest piece connected by hollow tubing to
two earpieces. The chest piece typically has two different
sides, a bell and a diaphragm. The bell side is in the
shape of a hollow cup. When placed on the patient, the
vibration of the skin is transmitted as sound waves
through the tubing to the earpieces. The bell is used to
listen to low-frequency sounds such as those produced by
the heart. The diaphragm side of the chest piece is
usually a plastic disc stretched over a hollow cup. Sound
waves from the body make the diaphragm vibrate,
sending acoustic pressure waves through the tubing to
the earpieces. The diaphragm is used to listen to high-
frequency sounds, such as those produced by the lungs.
The quality of an acoustic stethoscope is greatly affected
by the quality of its chest piece. It was not intended in
this paper to focus the work on studying the mechanical
parts of the stethoscope, but mainly the chest piece [4]
[5] [6].
This work describes only the stethoscope developed from
the audio signal perspective, i.e., after the sound waves
acquisition by the microphone. The position of the
microphone placed inside the chest piece has not been
studied in this work, since it requires a dedicated study in
order to support which position should be the best to
place the microphone inside. Having this in mind, and
considering that we do not have the technical knowledge
necessary to design an improved chest piece, we decided
to make only the electrical characterization of this
stethoscope prototype.
3. MULTI-CHANNEL STETHOSCOPE
ARCHITECTURE
The multi-channel digital stethoscope architecture is
presented in Fig. 1. The acquisition front-end is
responsible for the heart sounds acquisition, through
active microphones followed by filtering, analog-to-
digital conversion and channel equalization. A DSP is
used to do channel equalization, which permits to
customize the stethoscope frequency response. Different
frequency responses can be programmed into the DSP,
depending on the desired clinic speciality. The Micro
Controller Unit (MCU) handles Ethernet
communications and other system control functions, such
as, bank-filter selection.
Fig. 1: Digital multi-channel stethoscope architecture.
3.1. Signal Acquisition Stage
A Panasonic WM-61A Omni-directional electret
condenser microphone was used due its high linearity of
±1dB for frequencies from 20Hz to 5kHz and +1/-3dB
for 5kHz to 20kHz, low noise floor and high sensibility
of –35±4dB (0db = 1V/pa, 1kHz). It will be built-in the
chest piece to capture the acoustic pressure waves. The
performance of the WM-61A is excellent [7], but it
requires further signal processing to conform to
phantom-powered-microphone standards. Sound pressure
moves a diaphragm, producing variations in its
capacitance. This action produces an ac-output voltage
with extremely high impedance [7]. The electric signal
produced by the WM-61A, is a very weak signal, in the
order of tens of mV.
Fig. 2: Signal acquisition state with active microphone.

Colaço et al. DSP-based Multi-channel Stethoscope
APEA – 12º Encontro, Aveiro, DETI - Universidade de Aveiro, 8 e 9 de Outubro de 2010
Página 3 de 5
An integrated circuit with dual ultra-low noise and low
distortion op-amp will make use of one op-amp as simple
voltage buffer that provides low-impedance drive for one
output and the other op-amp will act as a unity-gain
inverter that derives its drive from the output of the first
op-amp, generating a phase inverted signal. The output
of this active microphone is a low-impedance differential
signal. The amplifier outputs, see Fig. 2, use ac coupling
in the active microphone output terminals, to block the
dc phantom-power voltage, on the audio lines. Phantom-
powered microphones derive power for their active
circuitry from the receiving-end circuit through the same
leads that transmit the audio signal. A balanced
microphone cable will be used between the active
microphone chest piece and the receiving-end amplifier.
3.1.1. Pre-Amplifier and Filtering
The receiving-end for the active microphones provides
phantom power and takes advantage of THAT 1510, an
ultralow-noise instrumentation amplifier, from THAT
Corporation with three internal op-amps. The
THAT 1510 improves on existing integrated microphone
pre-amps by offering lower noise at low gains, wider
bandwidth, higher slew rate, lower distortion, lower
supply current and provides excellent CMRR (common-
mode-rejection) properties [8]. The high CMRR rejects
noise and power-line hum that appear equally in both
signal lines. This instrumentation amplifier is used as an
audio pre-amplifier and its gain is adjusted with a
trimmer from 20 to 60 dB, see Fig. 3.
Fig. 3: Audio pre-amplifier and filtering stage.
The output signal from the audio pre-amplifier is then
filtered by an active anti-aliasing filter formed by a
variable Sallen and Key low pass filter design, with cut-
off frequencies between 3,1 and 42 kHz. The anti-
aliasing filter output is then sampled by the Analog
Devices AD1974 ADC. This audio ADC has 4-channels
and operates with 24 bits sampling depth and must be
driven from a differential signal source for best
performance, see Fig. 4.
Fig. 4: AD1974 differential input conversion.
The input pins of the AD1974 connect to internal
switched capacitors. To isolate the external driving op-
amp from the glitches caused by the internal switched
capacitors, each input pin is isolated by using a series
connected resistor together with a 1nF capacitor
connected from each input to ground. This capacitor is a
high quality ceramic NPO [9]. Some care was taken to
choose high quality capacitors and high precision metal
film resistors in all audio paths.
3.2. DSP Unit
The DSP Unit used was the ADAU1701 SigmaDSP®
from Analog Devices. This unit is a complete single-chip
audio system, with a built-in audio DSP, built in DACs,
and MCU-like control interfaces.
Signal processing capabilities includes, equalization,
audio enhancement, among others, and can be used to
improve limitations of real-world audio signals,
providing improvements in the perceived audio [10].
The ADAU1701 can be programmed with the Analog
Devices SigmaStudio™ software, which allows us to
graphically configure a custom signal processing
workflow using blocks such as filters, level controls or
GPIO interface controls. Compiled code is generated by
the SigmaStudio™, and can be loaded on power-up
either from a serial EEPROM through a self-boot
mechanism or from an external MCU. On power-down,
the current state of the memory of the ADAU1701 is
written back to the EEPROM, to be recalled the next
time the chip is started [11].
Channel equalization was made using three-second order
IIR Filters that can be found under the General Purpose
Filter Library in SigmaStudio™ [10], see Fig. 5.
The stethoscope channel equalization used in the testes,
follows a simple topology, based on these specifications:
• Should attenuate 8 dBs, the 50 Hz line noise.
• Should boost 3 dBs @ 25 Hz.
• Should boost 2.5 dBs @ 1000 Hz.
Fig. 5: SigmaStudio™ GUI - Stethoscope Channel
Equalization Filter Design.
3.3. Headphone Output
To implement the Headphone output, see Fig. 6, the
processed audio stream is converted back into an

Colaço et al. DSP-based Multi-channel Stethoscope
APEA – 12º Encontro, Aveiro, DETI - Universidade de Aveiro, 8 e 9 de Outubro de 2010
Página 4 de 5
analogue audio signal for real-time listening. The
ADAU1701 provides four high quality on-board Delta-
Sigma DACs. Each DAC channel is followed by a low-
pass filter that reconstructs the signal existent in the DAC
output. Finally, the audio signal is amplified by a high-
output-current op-amp, the NJM4556AD, to drive the
headphone speakers.
Fig. 6: Headphone output stage.
3.4. Micro Controller Unit
The MCU chosen was the Microchip™ PIC32MX785L
which has built-in Ethernet capabilities. The MCU
controls all the on-board peripherals, and makes the
“bridge” between the stethoscope and the exterior world.
Control data communication between the MCU and the
ADAU1701 is made through a serial SPI port. Audio
data communication between MCU and the ADAU1701
is performed by a serial I2S port, running in TDM-8
Channel mode.
Due to the high rate audio data transmitted between the
stethoscope and a possible client application running in a
PC in the local area network (LAN), and taking into
account the common use of Ethernet today, the
communication protocol chosen in the Ethernet-based
communications was the User Datagram Protocol (UDP).
The advantages of using UDP are its simple message-
based, connectionless protocol and it does not set up a
dedicated end-to-end connection. Communication is
achieved by transmitting information in one direction
from source to destination without verifying the
readiness or state of the receiver, making it lightweight
[12].
4. AUDIO TESTS
Audio tests were performed only for the electrical part of
the stethoscope. The characterization of the mechanical
parts has been left for future work, given the difficulties
normally encountered in the characterization of the
acoustic parts of stethoscopes. The equipment used for
the tests was the Portable One, made by the well-known
Audio Precision Inc. This equipment allows performing
standard tests, commonly used in industry, for quality
certification of audio equipment. Figure 7 shows the
prototype under testing.
Fig. 7: Audio tests performed with the Portable One.
The experimental results were grouped in Tab. 1. The
input signal used was a sinusoidal function with
25 mVpp and 1 kHz frequency.
5. RESULTS
The results were grouped into table 1. The input signal
used under testing, was a sinusoidal function, with 25
mV peak-to-peak and 1 kHz frequency.
Parameter
Value
SNR
78.70 dB
THD+N
0.0251 %
Load
1990.0 !
Tab. 1: Experimental results.
We also carried out a sweep of frequencies with the
Portable One to obtain the stethoscope frequency
response. The figure below shows the result obtained.
Fig. 8: Frequency response obtained with the
Portable One.
6. CONCLUSIONS
As already stated, our final goal was to develop a
proof-of-concept prototype of a multi-channel digital
stethoscope.
The stethoscope channel equalization proved to be a
major feature. With this feature, it is possible to include
different frequency responses in the stethoscope memory
and customize the device for different clinical
specialities.
The developed stethoscope was tested using one of the
most reliable equipment in the Audio industry, the
Portable One from Audio Precision Inc. A high value for
the Signal to Noise Ratio (SNR=78.7dB) was achieved,
and a low Total Harmonic Distortion plus Noise
(THD+N=0.0251%) was measured. These results show
that it is possible to build low-cost digital stethoscopes
with reasonable quality and some advanced features.

Colaço et al. DSP-based Multi-channel Stethoscope
APEA – 12º Encontro, Aveiro, DETI - Universidade de Aveiro, 8 e 9 de Outubro de 2010
Página 5 de 5
7. FUTURE WORK
As future work we intend to develop and implement
algorithms to extract valuable information from heart
sounds in order to detect abnormal functioning of the
heart. Finally, the device should be able to collect ECG
signals synchronized with the heart sounds. Regarding
the communications, the device must evolve to wireless.
8. ACKNOWLEDGMENTS
We would like to thank the Department of Electronics,
Telecommunications and Informatics of the University of
Aveiro, in the person of Professor José Neto Vieira, the
access to the test equipment used to evaluate this project.
9. REFERENCES
[1]
H. Wang, J. Chen, Y. Hu, Z. Jiang, and C. Samjin,
"Heart Sound Measurement and Analysis System
with Digital Stethoscope," Biomedical Engineering
and Informatics, 2009. BMEI '09. 2nd International
Conference on, pp. 1-5, 2009.
[2]
P. Sankaran, K. Chandrasekaran, A. H. Baig, and C.
L. Moll, "Development of a tele- stethoscope: Indian
perspective," in , 2010, pp. 237-239.
[3]
W. J. gang, S. Xiao-Bo, W. Ping, H. Wei, and D.
Cui-Lian, "Remote Heart Sound Monitoring
System," in , 2005, pp. 2138-2140.
[4]
D. Callahan, J. Waugh, G. A. Mathew, and W. M.
Granger, "Stethoscopes: what are we hearing?,"
Biomed Instrum Technol, vol. 41, no. 4, pp. 318-
323, 2007.
[5]
S. Applebaum, "Stethoscope Use without Behind-
The-Ear Hearing-Aid Removal," Association of
Medical Professionals with Hearing Losses, 2003.
[6]
P. D. Welsby, G. Parry, and D. Smith, "The
stethoscope: some preliminary investigations,"
Postgrad Med, vol. 79, p. 695–698, 2003.
[7]
Panasonic, "WM-61A - Omnidirectional Back
Electret Condenser Microphone Cartridge
Datasheet".
[8]
T. Corporation, "THAT 1510 - Low-Noise, High
Performance Audio Preamplifier IC Datasheet".
[9]
A. Devices, "AD1974 - 4 ADC with PLL, 192kHz,
24-Bit ADC Datasheet", 2007
[10]
A. Devices, "SigmaStudio™ Help Manual.", 2006.
[11]
A. Devices, "ADAU1701 - SigmaDSP® 28/56-Bit
Processor with Two ADCs and Four DACs
Datasheet", 2006.
[12]
B. A. Forouzan, "Data Communications and
Networking", 4th ed. McGraw-Hill
Science/Engineering/Math, 2007.