SCIENCE CHINA Information Sciences, Volume 62 , Issue 10 : 202402(2019) https://doi.org/10.1007/s11432-019-9906-1

A flexible skin-mounted wireless acoustic device for bowel sounds monitoring and evaluation

More info
  • ReceivedMar 18, 2019
  • AcceptedMay 23, 2019
  • PublishedSep 2, 2019



This work was supported by National Basic Research Program of China (Grant No. 2015CB351900) and National Natural Science Foundation of China (Grant Nos. 11320101001, 11222220, 11625207).



The calculation of $f_r$ and $f_0$

The internal coincidence frequency $f_0$ of the gas in the pipe is defined by \begin{equation} f_0=\frac{f_r}{4}\left(\frac{c_2}{343}\right),\end{equation} where $c_2$ is the speed of sound inside the pipe and $f_r$ is the ring frequency, which is \begin{equation} f_r=\frac{c_L}{2\pi a},\end{equation} where $c_L$ refers to the compressional wave speed in the pipe wall. In this way $f_0$ (approximately 20000 Hz) turns out to be much higher than the frequency of peak bowel sound, $f_p$ (lower than 1000 Hz).

Preprocessing of bowel sound data

Considering the difference of digestion rates for nuts and oranges, part of recordings of bowel sounds are cut out and only 90 min' recordings of each measurement are retained. For nuts and oranges, corresponding recordings in the first 30 min and 10 min respectively are cut out.

Firstly, the digital sound stream is filtered using a digital low-pass infinite impulse response (IIR) filter (0–600 Hz) 1) 2) 3) for de-noising, to eliminate the possibilities of miscalculation for signals' features. Then sound signals are normalized to a scale of $-1$, 1 so that the amplitude of the signal is not affected by measurement sites and individual differences. After that, every bowel sound is extracted separately into 256-point segments (0.128 s). If the length of some certain sounds are larger than 256, then they are not considered in subsequent analysis. Besides, Hamming window is assigned to each individual segment.

After data segmentation, there are two types of features adopted for subsequent analysis: time-domain features and FFT results. In time domain, a single parameter is starting time, which is related to the time interval of occurrence of bowel sounds. For FFT results, the power of the frequency bands between 60 and 375 Hz is only chosen. Finally, each segment has 42 features.

Ranta R, Louis-Dorr V, Heinrich C, et al. Digestive activity evaluation by multichannel abdominal sounds analysis. IEEE Trans Biomed Eng, 2010, 57: 1507–1519.

Yoshino H, Abe Y, Yoshino T, et al. Clinical application of spectral analysis of bowel sounds in intestinal obstruction. Dis Colon Rectum, 1990, 33: 753–757.

Kim K-S, Seo J-H, Song C-G. Non-invasive algorithm for bowel motility estimation using a back-propagation neural network model of bowel sounds. Biomed Eng Online, 2011, 10: 69.


[1] Garner C G, Ehrenreich H. Non-invasive topographic analysis of intestinal activity in man on the basis of acustic phenomena. Res Exp Med, 1989, 189: 129-140 CrossRef Google Scholar

[2] Harari D, Norton C, Lockwood L. Treatment of constipation and fecal incontinence in stroke patients: randomized controlled trial.. Stroke, 2004, 35: 2549-2555 CrossRef PubMed Google Scholar

[3] Talley N J, Vakil N B, Moayyedi P. American gastroenterological association technical review on the evaluation of dyspepsia.. Gastroenterology, 2005, 129: 1756-1780 CrossRef PubMed Google Scholar

[4] Dagdeviren C, Javid F, Joe P. Flexible piezoelectric devices for gastrointestinal motility sensing.. Nat Biomed Eng, 2017, 1: 807-817 CrossRef PubMed Google Scholar

[5] Inderjeeth A J, Webberley K M, Muir J. The potential of computerised analysis of bowel sounds for diagnosis of gastrointestinal conditions: a systematic review.. Syst Rev, 2018, 7: 124 CrossRef PubMed Google Scholar

[6] Zhang Y, Jeon M, Rich L J. Non-invasive multimodal functional imaging of the intestine with frozen micellar naphthalocyanines. Nat Nanotech, 2014, 9: 631-638 CrossRef PubMed ADS Google Scholar

[7] Kim K S, Seo J H, Song C G. Non-invasive algorithm for bowel motility estimation using a back-propagation neural network model of bowel sounds.. Biomed Eng Online, 2011, 10: 69 CrossRef PubMed Google Scholar

[8] Ching S S, Tan Y K. Spectral analysis of bowel sounds in intestinal obstruction using an electronic stethoscope.. WJG, 2012, 18: 4585 CrossRef PubMed Google Scholar

[9] Craine B L, Silpa M, O'Toole C J. Computerized Auscultation Applied to Irritable Bowel Syndrome. Digestive Dis Sci, 1999, 44: 1887-1892 CrossRef Google Scholar

[10] Ranta R, Louis-Dorr V, Heinrich C. Digestive activity evaluation by multichannel abdominal sounds analysis.. IEEE Trans Biomed Eng, 2010, 57: 1507-1519 CrossRef PubMed Google Scholar

[11] Baid H. A critical review of auscultating bowel sounds.. British J Nursing, 2009, 18: 1125-1129 CrossRef PubMed Google Scholar

[12] Feinberg A N, Feinberg L A, Atay O K. Gastrointestinal care of children and adolescents with developmental disabilities.. Pediatric Clinics North Am, 2008, 55: 1343-1358 CrossRef PubMed Google Scholar

[13] Craine B L, Silpa M L, O'Toole C J. Enterotachogram Analysis to Distinguish Irritable Bowel Syndrome from Crohn's Disease. Digestive Dis Sci, 2001, 46: 1974-1979 CrossRef Google Scholar

[14] Yoshino H, Abe Y, Yoshino T. Clinical application of spectral analysis of bowel sounds in intestinal obstruction. Dis Colon Rectum, 1990, 33: 753-757 CrossRef Google Scholar

[15] Zaborski D, Halczak M, Grzesiak W. Recording and Analysis of Bowel Sounds.. EJOHG, 2015, 5: 67-73 CrossRef PubMed Google Scholar

[16] Vasseur C, Devroede G, Dalle D. Postprandial Bowel Sounds. IEEE Trans Biomed Eng, 1975, BME-22: 443-448 CrossRef Google Scholar

[17] Drossman D A. The functional gastrointestinal disorders and the Rome II process. Gut, 1999, 45 Suppl 2: II1-5. Google Scholar

[18] Gerald H, Paul E. Stress and Gastrointestinal Motility in Humans: A Review of the Literature. Neurogastroenterol Motil, 1991, 3(4): 245-54. Google Scholar

[19] Rekanos I T, Hadjileontiadis L J. An iterative kurtosis-based technique for the detection of nonstationary bioacoustic signals. Signal Processing, 2006, 86: 3787-3795 CrossRef Google Scholar

[20] Hadjileontiadis L J, Liatsos C N, Mavrogiannis C C. Enhancement of bowel sounds by wavelet-based filtering.. IEEE Trans Biomed Eng, 2000, 47: 876-886 CrossRef PubMed Google Scholar

[21] Craine B L, Silpa M L, O'Toole C J. Two-Dimensional Positional Mapping of Gastrointestinal Sounds in Control and Functional Bowel Syndrome Patients. Digestive Dis Sci, 2002, 47: 1290-1296 CrossRef Google Scholar

[22] Gao W, Emaminejad S, Nyein H Y Y. Fully integrated wearable sensor arrays for multiplexed in situ perspiration analysis. Nature, 2016, 529: 509-514 CrossRef PubMed ADS Google Scholar

[23] Gao G P, Hu B, Tian X L. Experimental study of a wearable aperture-coupled patch antenna for wireless body area network. Microw Opt Technol Lett, 2017, 59: 761-766 CrossRef Google Scholar

[24] Karthik V, Rao T R. SAR investigations on the exposure compliance of wearable wireless devices using infrared thermography.. Bioelectromagnetics, 2018, 39: 451-459 CrossRef PubMed Google Scholar

[25] Cai S, Han Z, Wang F. Review on flexible photonics/electronics integrated devices and fabrication strategy. Sci China Inf Sci, 2018, 61: 060410 CrossRef Google Scholar

[26] Chen S H, Pan T S, Yan Z C, et al. Flexible ultra-wideband rectangle monopole antenna with O-slot insertion design. Sci China Inf Sci, 2018, 61: 060414. Google Scholar

[27] Xu C Q, Liu Y, Yang Y T. An intelligent partitioning approach of the system-on-chip for flexible and stretchable systems. Sci China Inf Sci, 2018, 61: 060415. Google Scholar

[28] Mamun K A A, McFarlane N. Integrated real time bowel sound detector for artificial pancreas systems. Sens Bio-Sens Res, 2016, 7: 84-89 CrossRef Google Scholar

[29] Zwikker C, Kosten C W. Sound absorbing materials. New York: Elsevier Publishing Company, Inc. Google Scholar

[30] Bies D A, Hansen C H. Engineering noise control : theory and practice: Spon Press. Google Scholar

[31] Fahy F J, Gardonio P. Sound and Structural Vibration: Radiation, Transmission and Response: Elsevier Science. Google Scholar

[32] Georgoulis B. Bowel sounds. Proc R Soc Med, 1967, 60(9): 917-20. Google Scholar

[33] Yamaguchi K, Yamaguchi T, Odaka T. Evaluation of gastrointestinal motility by computerized analysis of abdominal auscultation findings.. J Gastroenterol Hepatol, 2006, 21: 510-514 CrossRef PubMed Google Scholar

[34] Bingham S A, Day N E, Luben R. Dietary fibre in food and protection against colorectal cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC): an observational study. Lancet, 2003, 361: 1496-1501 CrossRef Google Scholar

[35] Fadelu T, Zhang S, Niedzwiecki D. Nut Consumption and Survival in Patients With Stage III Colon Cancer: Results From CALGB 89803 (Alliance).. JCO, 2018, 36: 1112-1120 CrossRef PubMed Google Scholar

  • Figure 1

    (Color online) Working principle and design of the flexible acoustic device. (a) Schematic of the flexible skin-mounted wireless acoustic device for bowel sound detection. (b) Illustration of the principle of the device for evaluations of bowel problems, based on a machine learning model of different intestinal conditions, such as weak peristalsis, or peristalsis with poisonous contents. (c) Exploded view diagram of the overall design structure of the device. (d) Layout of device's circuit. The red dotted rectangle separates MEMS microphone chip and the main circuit component area to reinforce the flexible board's bending performance. (e) Small rechargeable battery that can be bent on a rod with a diameter of 2 cm. protect łinebreak (f) Waveform display in mobile application. (g) Bent device using the fingers. (h) Device mounted on the curved abdominal surface.

  • Figure 2

    (Color online) Design of resonator. (a) Top and cross-sectional views of the resonator structure; (b) a pipe model ideally transformed from the channel section of the resonator; (c) curves of acoustic resistance vs. pipe radius and (d) transmission loss vs. wall thickness for sound frequencies of 50, 150 and 1000 Hz; (e) a skin-shell-air model depicting the transmission of a sound wave through the shell; (f) sound transmission coefficient vs. shell thickness curves for the sound frequencies of 50, 150 and 1000 Hz; (g) sound intensity streamlines in the resonator in lateral view at the sound frequency of 150 Hz; (h) comparison of the received sound signal amplitude when the device faces the same sounds with and without a resonator.

  • Figure 3

    (Color online) Comparison between the flexible acoustic device and commercial e-stethoscope for experimental characterization during abdominal breathing. (a) Example of a recorded bowel sound waveform. Each wave is a bowel sound pulse. (b) Comparison of simultaneous recordings using a commercial e-stethoscope and our device. (c) Photographs of the attached flexible acoustic device during an abdominal respiratory cycle. (d) Plots of measured signals during inhalation and exhalation. In the middle image, the black line between the red line (inhalation) and blue line (exhalation) represents the end-inspiratory pause. Magnified views of the two bowel sounds during inhalation (left) and exhalation (right). (e) The approximate maximum bending curvature of the acoustic device.

  • Figure 4

    (Color online) Long time evaluation of bowel sounds. (a) Bowel sound variations for near five hours after food intake; (b) variation of number of peaks (in histogram) and short-time energy (broken line) of bowel sounds over time; protectłinebreak (c) spectrogram of bowel sounds (darker color means higher spectral power along the y-axis) using short-time Fourier transform.

  • Figure 5

    (Color online) Clinical tests of the device. (a) Typical waveform lines of bowel sounds from one normal subject (blue line) and two patients with MIO (red line) and paralytic ileus (pink line); (b) corresponding power envelope curves associated with the three studied cases.

  • Figure 6

    (Color online) Collecting and classification of bowel sounds. (a) FFT results of bowel sound segments at different times from four volunteers following the consumption of nuts and oranges; (b) recognition accuracy outcomes for two types of bowel sounds in the 10-fold cross-validation.