Lung disease detection by sound analysis techniques
1.
2. Lung Sound characteristic
Vesicular Breath Sound Abnormal Breath Sound Adventitous Breath Sound
Tracheal
(High pitch)
Vesicular
(Low pitch)
Bronchial
(High pitch)
Bronchovesicular
(Low pitch)
Absent / Decreased
Normal Breath
Bronchial Sound
in Abnormal
Continuous
Lung Sound
Discontinuous
Lung Sound
Wheeze
Dominant frequency range:- 400Hz or more
Duration:- 250ms or more ; High pitch
Sensor location:- any location over lung/ Trachea or
most of chest wall area
Disorder:- Asthma
Ronchi
Dominant frequency range:- 200Hz or more
Duration:- 250ms or more ; Low pitch
Sensor location:- any location over lung/ Trachea or
most of chest wall area
Disorder :- COPD ,Acute or severe
Crakles/Rales
Dominant frequency
Range:-200 to 2000Hz
Coarse ckrakles
Duration:- 100ms or more ;
Low pitch; Location:- anterior
of posterior of chest wall
Fine crakles
Duration:- 100ms or more;
High pitch
Sensor location:- posterior
chest wall
3. Wave form of Crackle
• IDW or initial deflection width represents
the duration between the beginning of the
crackle and the first deflection.
• 2CD (two-cycle duration) is the duration
from the beginning of the crackle to the
date at which the waveform did two
complete cycles
• It is accepted that the duration of a crackle
is lower than 20 ms and the frequency
range is between 100 and 200 Hz
• Crackles are generated by small airways
snapping open on inspiration. Therefore,
they are predominantly inspiratory. The
difference between the course and
fine crackles is believed to come from the
size of the airway snapping open (larger
airways, deeper pitched, courser crackles)
Wave form of Wheeze
• A wheeze is a continuous, coarse, whistling
sound produced in the respiratory airways
during breathing.
• For wheezes to occur, some part of the
respiratory tree must be narrowed or
obstructed (for example narrowing of the
lower respiratory tract in an asthmatic
attack), or airflow velocity within the
respiratory tree must be heightened.
• Rhonchi are rattling, continuous and
low-pitched breath sounds that are
often hear to be like snoring.
• It is caused by thick secretions in large
airways as air passes by.
• Rhonchi can be heard in patients with
pneumonia, chronic bronchitis, cystic
fibrosis or COPD (chronic obstructive
pulmonary disease). Coughing will
often clear rhonchi.
• They are lower in pitch than wheezes
and have a snoring quality. They also
have a sinusoidal pattern on waveform,
but the number of deflections per unit
time is less than that of wheezes as they
are of lower frequency.
Wave form of Rhonchi
4. Figure:- Location on posterior of lung to receive
the most pure sound from lung
Location of the Chest sound receiving devices
5. Lung sound Processing
Steps:-
1.) Receiving of Sound from different sources.
2.) Extraction of Noise and Conversion of sound into signals/message.
3.) Processing of sound in computer in line.
4.) Separated sound of lung and heart shown on computer display.
Step 1 Step 2 Step 3 Step 4
6. 1.)Receiving of Sound from different sources
Heart sound wave
Lung sound wave
Noise sound wave
Received sound wave contain sound
from three sources:- Heart, lung and
Noise
20Hz to 150Hz
Noise
Microphone as
sound receiver
Main sources of Sound
25Hz to 1500Hz
7. 2.)Extraction of Noise
The firs section of the step is to extract the lung sound from mixture of heart sound and lung sound :-
Microphone:- Receive
sound
Amplifier:- Amplifies
sound
Low pass filter
Computer in line
Set at cut-off of 2000Hz
Output message
Input message
Right chest sound
Left chest sound
8. 3.)Processing of sound in computer In line
Computer Sound card Processing Algorithm
Output sound signals of right chest
and left chest when the noise sound
waves are filtered by the low pass filter
Graph of sounds produce by
heart and lung separately
Right chest sound
Left chest sound
9. Graph shown on display
4.)Separated sound of lung and heart shown on computer display
Lungs
Heart
10. Sound Sample WavePad Audio Editor (Free Version)
Wave Graph of Sound Sample
No. of small sections of
sound are created and
observed that which type
of waveform are formed
and process in Database
through which result will
display on screen
11. Algorithm for separation of the lung sound from mixture of sound
ICA:- Independent component analysis technique
Let us consider two sources of signal form both lung X and X
Then X (t) = A S (t) + A S (t) (1)
X (t) = A S (t) + A S (t) (2)
Where a are correlation coefficients, the problem is to get the original signals out of the recorded ones, but separately. So we
need to identify original sources S and S using only the recorded mixed signals X and X . A more simplified writing of equations
(1) and (2) is;
X=A.S
Where X is Mixing matrix, and S is recorded signal matrix and S, is original sound source matrix.
S= WX
Where W is 1/A .
1 2
1 11 1 12 2
2 21 2221
ij
1 2 1 2
12. References
[1] Neeraj Sharma, Prashant Krishnan, Rohit Kumar, Shreyas Ramoji, Srikanth Raj Chetupalli, Nirmala R., Prasanta Kumar Ghosh, and Sriram Ganapathy “Coswara - A
Database of Breathing, Cough, and Voice Sounds for COVID-19 Diagnosis”.
[2] Malay Sarkar, Irappa Madabhavi1 , Narasimhalu Niranjan, Megha Dogra “Auscultation of the respiratory system”.
[3] Fatma Ayaria, Mekki Ksouria, Ali T. Alouan “Lung sound extraction from mixed lung and heart sounds FASTICA algorithm”.
[4] Abhishek Jain, Jithendra Vepa “Lung Sound Analysis for Wheeze Episode Detection”.
[5] Jen-Chien Chien1 ,Ming-Chuan Huang1 , Yue-Der Lin2 ,Fok-ching Chong “A Study of Heart Sound and Lung Sound Separation by Independent Component Analysis
Technique”.
[6] Chengyu Liu1, David Springer2, Qiao Li1, Benjamin Moody3, Ricardo Abad Juan4,5, Francisco J Chorro6, Francisco Castells5, José Millet Roig5, Ikaro Silva3,
Alistair E W Johnson3, Zeeshan Syed7, Samuel E Schmidt8, Chrysa D Papadaniil9, Leontios Hadjileontiadis9, Hosein Naseri10, Ali Moukadem11, Alain Dieterlen11,
Christian Brandt12, Hong Tang13, Maryam Samieinasab14, Mohammad Reza Samieinasab15, Reza Sameni14, Roger G Mark3 and Gari D Clifford, “An open access
database for the evaluation of heart sound algorithms”.
[7] Bruno M. Rocha1 , Dimitris Filos2 , Luís Mendes3,1, Gorkem Serbes4 , Sezer Ulukaya5,6, Yasemin P. Kahya6 , Nikša Jakovljević7 , Tatjana L. Turukalo7 , Ioannis M.
Vogiatzis2 , Eleni Perantoni2 , Evangelos Kaimakamis2 , Pantelis Natsiavas2 , Ana Oliveira8 , Cristina Jácome8 , Alda Marques8 , Nicos Maglaveras2,9, Rui Pedro Paiva1
, Ioanna Chouvarda2 , Paulo de Carvalh “An Open Access Database for the Evaluation of Respiratory Sound Classification Algorithms”
[8] Hong Wang1 , Le Yi Wang2 , Han Zheng3 , Razmig Haladjian4 , Meghan Wallo5 “Lung Sound/Noise Separation for Anesthesia Respiratory Monitoring”.
[9] V. Gross’, L: J. Hadjileontiadisz, Th. Penzel’, U. Koehler’, C. k‘ogelmeier’ “Multimedia Database “Marburg Respiratory Soun ds (MARS)“”.
[10] Rajkumar Palaniappan *, Kenneth Sundaraj, Nizam Uddin Ahamed “Machine learning in lung sound analysis: A systematic review”.