Intestinal motility quantitative evaluation method and system based on multi-divisional bowel sound monitoring

Abstract

The application provides an intestinal tract movement quantitative evaluation method and system based on multi-partition intestinal sound monitoring, and relates to the technical field of physiological acoustics signal monitoring.The application collects multi-path partition mixed sound signals through multiple abdominal contact acoustics heads arranged in different abdominal partitions, collects reference signals through a reference noise collection channel, and synchronously receives and writes time stamps by a central collector; estimates and offsets the noise in the partition mixed sound signals according to the reference signals to obtain partition intestinal sound signals; detects candidate intestinal sound events, correlates across channels and merges or retains according to the correlation results to obtain intestinal sound events; determines the occurrence area according to the correlation results, acoustic characteristics and corresponding abdominal partitions, and counts the intestinal sound events according to the occurrence area to obtain intestinal tract movement quantitative parameters.The application is beneficial to improving the anti-interference and regional quantitative accuracy of the body surface intestinal sound monitoring.

Description

Technical Field

[0001] This invention relates to the field of physiological acoustic signal monitoring technology, and in particular to a method and system for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring. Background Technology

[0002] The acoustic information generated during intestinal motility can be used to reflect the state of intestinal activity. Current methods for monitoring bowel sounds mostly rely on manual listening or single-point acoustic acquisition. The acquisition results are easily affected by the attachment location, monitoring duration, and usage environment, making it difficult to continuously reflect the distribution of bowel sound activity in different abdominal regions and hindering the formation of comparable quantitative results.

[0003] In actual monitoring environments, the amplitude of the acoustic signal on the abdominal surface is relatively weak, and environmental noise, friction from clothing, and body movement interference can easily overlap with the bowel sound signal. If only energy is judged from a single signal, problems such as missed bowel sounds, false noise detection, and repeated counting of the same sound at different acquisition locations can easily occur, affecting the stability and objectivity of bowel sound monitoring results.

[0004] Therefore, how to improve the anti-interference ability of bowel sound signals under non-invasive collection conditions on the body surface, establish the temporal consistency and event correspondence between different collection locations, and obtain quantitative parameters that can characterize the region and intensity of bowel sound occurrence has become a problem to be solved in the acoustic monitoring technology of intestinal activity. Summary of the Invention

[0005] To address the problems in existing bowel sound monitoring methods, such as single-point acquisition failing to reflect the distribution of bowel sound activity in different abdominal regions, interference from environmental noise and friction from the device itself on the acoustic signal, and the tendency for the same sound to be counted repeatedly at different acquisition locations, this invention provides a method and system for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring. This method acquires acoustic signals from different abdominal zones under surface acquisition conditions and combines them with reference signals for noise cancellation, event correlation, and determination of the occurrence area, thereby obtaining quantitative parameters of intestinal motility that characterize the intensity and regional distribution of bowel sound activity.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring, comprising: S101 collects multi-channel mixed acoustic signals by multiple abdominal contact acoustic heads located in different abdominal zones, collects reference signals by a reference noise acquisition channel, and is synchronously received and written to a timestamp by the central acquisition unit. S102, based on the reference signal, estimate and cancel the noise in the partitioned mixed sound signal to obtain the partitioned bowel sound signal; S103, detect candidate bowel sound events in the zoned bowel sound signals, correlate candidate bowel sound events across channels, and merge or retain candidate bowel sound events according to the correlation results to obtain bowel sound events; S104. The occurrence area is determined based on the correlation results, the acoustic characteristics of bowel sound events, and the corresponding abdominal regions. S105, according to the occurrence area, count bowel sound events to obtain quantitative parameters of intestinal motility.

[0007] Furthermore, multiple abdominal contact acoustic heads are respectively attached to at least two abdominal regions in the upper, middle, and lower abdomen; the reference signals include environmental noise signals collected by an environmental noise reference microphone and motion reference signals collected by a triaxial motion sensor.

[0008] Furthermore, the synchronous reception and writing of timestamps includes: multiple abdominal contact acoustic heads are connected to the multi-channel audio ADC of the central acquisition unit via flexible differential lines and share a sampling clock; the output of the multi-channel audio ADC, the ambient noise reference microphone, and the triaxial motion sensor are respectively connected to the main control of the central acquisition unit; the main control writes timestamps for each signal to obtain time-stamped partitioned mixed sound signals and reference signals.

[0009] Furthermore, noise cancellation includes: synchronizing and framing the reference signal and the partitioned mixed sound signal; using the current filtering parameters, estimating the noise in the partitioned mixed sound signal frame based on the reference signal frame to obtain a noise estimation frame; subtracting the noise estimation frame from the partitioned mixed sound signal frame to obtain the current partitioned bowel sound signal frame; determining the next filtering parameters based on the reference signal frame and the current partitioned bowel sound signal frame; using the next filtering parameters in the next synchronized signal frame, performing noise estimation, subtraction, and filtering parameter determination frame by frame to obtain the partitioned bowel sound signal.

[0010] Furthermore, the next filtering parameter is determined as follows: calculate the maximum value of the cross-correlation function amplitude of the reference signal frame and the synchronized partitioned mixed sound signal frame; if the maximum value is not lower than the correlation threshold, update the current filtering parameter according to the least mean square criterion based on the reference signal frame and the current partitioned bowel sound signal frame; if the maximum value is lower than the correlation threshold, keep the current filtering parameter; use the updated or kept current filtering parameter as the next filtering parameter.

[0011] Furthermore, the detection and cross-channel association of bowel sound events include: screening candidate signal frames based on short-time energy or the proportion of bowel sound frequency band energy, merging consecutive candidate signal frames into candidate bowel sound events; calculating the occurrence time difference and spectral similarity of candidate bowel sound events in different abdominal regions, merging candidate bowel sound events whose time difference does not exceed the time threshold and whose spectral similarity is not lower than the similarity threshold, and retaining the remaining candidate bowel sound events to obtain bowel sound events.

[0012] Furthermore, the determination of the occurrence region includes: for the bowel sound events obtained after merging, determining the arrival time difference, signal energy difference, and spectral characteristics based on the corresponding signals of the bowel sound events in different abdominal partitions, and sorting the abdominal partitions respectively; when the first partition of each sort is the same, the first partition is determined as the occurrence region; when the first partitions of each sort are different, the occurrence region is determined according to the sorting results; for the bowel sound events obtained after retention, the abdominal partition corresponding to the signal channel where the bowel sound event is located is determined as the occurrence region.

[0013] Furthermore, the sorting of abdominal regions and the determination of the occurrence region based on the sorting results include: determining the arrival order based on the time delay of the peak value of the cross-correlation function, determining the energy level based on the acoustic energy integral, and determining the spectral characteristics based on the energy proportion of the bowel sound band; assigning ranks according to arrival from early to late, energy from high to low, and energy proportion of the bowel sound band from high to low; adding the ranks of each abdominal region in the sorting, and determining one or more abdominal regions with the smallest sum of ranks as the occurrence region.

[0014] Furthermore, the intestinal motility quantification parameters are obtained by: determining the effective monitoring period for each abdominal region based on the channel detachment status and motion artifacts; retaining a count for the same bowel sound event in each occurrence region; and calculating at least one of the event rate, duration percentage, and average sound energy for each occurrence region according to the effective monitoring period to obtain the intestinal motility quantification parameters.

[0015] Secondly, the present invention also provides a system for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring, comprising multiple abdominal contact acoustic heads, a reference noise acquisition channel, and a central acquisition unit. The central acquisition unit is equipped with a noise cancellation module, an event detection module, an event association module, a region determination module, and a parameter generation module. Multiple abdominal contact acoustic heads are respectively attached to different abdominal zones and acquire multi-zone mixed acoustic signals. The reference noise acquisition channel is used to acquire reference signals. The central acquisition unit is connected to multiple abdominal contact acoustic heads and the reference noise acquisition channel to synchronously receive each signal and write a timestamp. The noise cancellation module is used to estimate and cancel noise in the mixed acoustic signals of the zones based on the reference signal to obtain zoned bowel sound signals. The event detection module is used to detect candidate bowel sound events in the zoned bowel sound signals. The event association module is used to associate candidate bowel sound events across channels and merge or retain candidate bowel sound events according to the association results to obtain bowel sound events. The region determination module is used to determine the occurrence region based on the association results, the acoustic characteristics of the bowel sound events, and the corresponding abdominal zone. The parameter generation module is used to statistically analyze bowel sound events according to the occurrence region to obtain quantitative parameters of intestinal motility.

[0016] The beneficial effects of this invention compared to the prior art are: By simultaneously acquiring mixed acoustic signals from multiple abdominal contact acoustic heads located in different abdominal regions and writing timestamps into them by a central acquisition unit, bowel sound activity in different abdominal regions can be compared under the same time reference, which is beneficial for reflecting the distribution of bowel sound activity in different areas of the abdomen.

[0017] By synchronously acquiring reference signals through the reference noise acquisition channel, and estimating and canceling noise in the mixed acoustic signals of the zones based on the reference signals, it is beneficial to reduce the influence of ambient noise, wearing friction, and body movement interference on the zoned bowel sound signals, thereby improving the stability of the body surface acoustic monitoring results.

[0018] By detecting candidate bowel sound events and performing cross-channel correlation, interrelated candidate bowel sound events are merged, while the remaining candidate bowel sound events are retained. This helps to reduce the repeated counting of the same sound at different acquisition locations, while also preserving local acoustic events that occur within a single abdominal region.

[0019] By combining correlation results, acoustic features, and corresponding abdominal regions to determine the occurrence area, and statistically analyzing bowel sound events according to the occurrence area, quantitative parameters of intestinal motility such as event rate, duration percentage, and average acoustic energy can be generated, which is beneficial for forming comparable results of bowel sound activity intensity and regional distribution. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring.

[0021] Figure 2 A schematic diagram showing the arrangement of the acoustic head for abdominal contact.

[0022] Figure 3 This is a schematic diagram of the multi-reference input noise cancellation process.

[0023] Figure 4 This is a schematic diagram of a system for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring. Detailed Implementation

[0024] This embodiment provides a method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring. Intestinal motility quantification parameters are obtained through abdominal surface acoustic acquisition, synchronous acquisition of reference signals, noise cancellation, bowel sound event detection, cross-channel correlation, determination of occurrence regions, and zonal statistics. The occurrence region represents the abdominal zone determined based on the surface acoustic response. The intestinal motility quantification parameters are used to represent the bowel sound activity and distribution in different abdominal zones during the monitoring period, outputting acoustic events and statistical parameters.

[0025] like Figure 1As shown, this method includes: S101, acquiring multi-channel mixed acoustic signals from multiple abdominal contact acoustic heads located in different abdominal zones, acquiring reference signals through a reference noise acquisition channel, and synchronously receiving and writing timestamps by a central acquisition unit; S102, estimating and canceling noise in the mixed acoustic signals based on the reference signals to obtain zoned bowel sound signals; S103, detecting candidate bowel sound events in the zoned bowel sound signals, associating candidate bowel sound events across channels, merging or retaining candidate bowel sound events according to the association results to obtain bowel sound events; S104, determining the occurrence area based on the association results, the acoustic characteristics of the bowel sound events, and the corresponding abdominal zone; S105, statistically analyzing bowel sound events according to the occurrence area to obtain quantitative parameters of intestinal motility.

[0026] In S101, such as Figure 2 As shown, multiple abdominal contact acoustic heads are attached to at least two abdominal regions in the upper, middle, and lower abdomen. This embodiment uses four abdominal contact acoustic heads, attached to the upper left, upper right, lower left, and lower right abdomen, respectively. The central acquisition unit stores the channel number, abdominal region, attachment position, and attachment time, establishing a one-to-one correspondence table between signal channels and abdominal regions. The attachment position avoids incisions, stomas, drainage tube fixation areas, and areas of skin damage. After attachment, the center position of the abdominal contact acoustic head is read using a positioning template, and the straight-line distance between the two center positions is calculated and written to the central acquisition unit in meters. When the positioning template outputs readings in millimeters or centimeters, the readings are divided by 1000 or 100 respectively before storage.

[0027] The abdominal contact acoustic head includes a medical adhesive ring, a silicone contact ring, a coupling disk, a sealed low-volume acoustic cavity, a differential analog MEMS microphone, and flexible differential lines. The coupling disk conforms to the abdominal wall and receives body surface vibrations. The sealed low-volume acoustic cavity generates sound pressure that varies with body surface vibrations. The differential analog MEMS microphone converts the sound pressure changes into differential electrical signals. The flexible differential lines transmit the differential electrical signals, reducing common-mode electromagnetic interference in cable transmission. Each acoustic channel uses a fixed acquisition gain, and automatic gain control is disabled. During factory calibration, each abdominal contact acoustic head is sequentially installed in the same coupling fixture, and the root mean square (RMS) value of the signal is recorded under the same excitation input. The amplitude correction value for the other signal channels is obtained by dividing the RMS value of the first signal channel by the RMS values ​​of the other signal channels. The central acquisition unit saves the amplitude correction value and keeps it unchanged during monitoring. If the calibration signal shows clipping or a RMS value of zero, the connection and excitation input are rechecked, and the calibration result is not used. Fixed gain and amplitude correction are used to reduce the impact of differences in acquisition channel sensitivity on partition energy comparison.

[0028] The signals acquired by the abdominal contact acoustic head include bowel sounds, ambient sound coupling components, and wearing friction components, forming a zoned mixed sound signal. A reference noise acquisition channel is used to acquire reference signals, including ambient noise signals acquired by an ambient noise reference microphone and motion reference signals acquired by a triaxial motion sensor. The ambient noise reference microphone is mounted on the outward-facing side of the central acquisition unit's housing and is used to record speech, equipment alarm sounds, and collision sounds. The triaxial motion sensor is mounted on the circuit board of the central acquisition unit, which is fixed to the wearing area, and is used to record motion changes accompanying body position changes, cable traction, or equipment collisions. The motion reference signal is used to estimate the noise components related to wearing motion and to identify motion artifacts.

[0029] Synchronous reception and timestamp writing include: multiple abdominal contact acoustic heads connected to a central acquisition unit via flexible differential lines, sharing a common sampling clock; the output of the multi-channel audio ADC, the ambient noise reference microphone, and the triaxial motion sensor are respectively connected to the main control unit of the central acquisition unit; the main control unit writes timestamps for each signal, resulting in a time-stamped zoned mixed sound signal and a reference signal. The multi-channel audio ADC uses a synchronous sampling method, with an audio sampling rate of [missing information]. The output data width is 24 bits. An ambient noise reference microphone is connected to the main controller via a PDM interface, converting the data into 16kHz pulse code modulation data. The three-axis motion sensor is connected to the main controller via an SPI interface, outputting three-axis motion data at 100Hz. The main controller determines the audio sampling time based on the audio sample count and the motion sampling time based on the motion data sampling timestamp. When the audio sampling time falls between two adjacent motion sampling times, the proportion of the time between the audio sampling time and the previous motion sampling time to the interval between the two motion sampling times is first calculated. This proportion is then multiplied by the difference between the two motion sampling values ​​and added to the previous motion sampling value to obtain the corresponding motion reference value. The next motion sampling value is buffered during processing, and the corresponding audio data is released after interpolation. Interpolation is used for time alignment; the effective frequency information of the motion reference signal is still determined by the sampling settings of the three-axis motion sensor.

[0030] Before formal monitoring, a 30-second static attachment calibration segment is collected, simultaneously saving the mixed acoustic signal of the zone, the ambient noise signal, the motion reference signal, and the equipment status. During the acquisition period, the abdominal contact acoustic head must remain attached, with no talking, collisions, cable pulling, or changes in body position. Any time periods with the above-mentioned recordings or acquisition anomalies are removed from the static attachment calibration segment. The main control unit simultaneously checks the continuity of sampling counts; if the sampling count is interrupted, data packets are missing, or the number of sampling points within a data packet is inconsistent with the configuration, the corresponding time period is marked as an acquisition anomaly. The static attachment calibration segment is used to determine the background correlation threshold, event detection threshold, and motion threshold; the calibration period is not included in the formal monitoring duration.

[0031] In S102, such as Figure 3As shown, noise cancellation includes: synchronizing and framing the reference signal and the partitioned mixed sound signal; using the current filtering parameters, estimating the noise in the partitioned mixed sound signal frame based on the reference signal frame to obtain a noise estimation frame; subtracting the noise estimation frame from the partitioned mixed sound signal frame to obtain the current partitioned bowel sound signal frame; determining the next filtering parameters based on the reference signal frame and the current partitioned bowel sound signal frame; using the next filtering parameters in the next synchronized signal frame, performing noise estimation, subtraction, and filtering parameter determination frame by frame to obtain the partitioned bowel sound signal.

[0032] Before frame division, divide the 24-bit signed audio sample value by The zoned mixed audio signal is multiplied by the corresponding amplitude correction value. The three-axis motion sample values ​​are converted to acceleration based on the sensitivity of the three-axis motion sensor, and then divided by the acceleration value corresponding to the configured range. Both audio and motion data use floating-point numbers in subsequent calculations. The reference signal and the zoned mixed audio signal are divided into 1-second synchronization signal frames according to the same start timestamp, each frame containing... Number of sampling points. Number of reference inputs is This includes one ambient noise signal and three motion reference signals. Let 'a' represent the channel number of the zone-mixed acoustic signal, 'm' represent the reference input number, 'p' represent the filter tap number, and 'n' represent the sampling point number within the current synchronization signal frame. Each zone-mixed acoustic signal is configured with four sets of filter parameters, each set containing... One tap; during the initial processing, all filter parameters are set to zero.

[0033] For each reference input, the current sampling point and the previous 127 sampling points are read, multiplied by the corresponding filter parameters respectively, and summed to obtain the noise component estimate corresponding to the reference input. The four noise component estimates are added together to obtain the noise estimate for the same sampling point. Noise estimation uses: ; In the formula, This is the noise estimate for channel a. The sampled value of input m at the corresponding tap position is used as a reference. This represents the current filtering parameters. All noise estimates within a synchronization frame constitute the noise estimation frame. When the tap position is before the current synchronization frame, the reference sample value from the end of the previous synchronization frame is read from the continuous buffer; when historical sample values ​​are missing for the first processing iteration, the missing portion is set to zero.

[0034] Subtract the mixed acoustic signal frame and the noise estimation frame from each zone using the same sampling number to obtain the current zone's bowel sound signal frame. The sampled values ​​within the frame are then used... The three-axis motion data, after being filtered by their respective parameters, participate in noise component estimation. The final subtraction results in two objects: a partitioned mixed acoustic signal frame and a noise estimation frame. Noise estimation uses the current and historical sampling points of the reference input, and the object to be canceled is the noise component that the reference input can characterize. Residual motion interference after filtering continues to participate in motion artifact determination.

[0035] The next filtering parameter is determined as follows: Calculate the maximum value of the cross-correlation function amplitude between the reference signal frame and the synchronized partitioned mixed sound signal frame; if the maximum value is not lower than the correlation threshold, update the current filtering parameter according to the least mean square criterion based on the reference signal frame and the current partitioned bowel sound signal frame; if the maximum value is lower than the correlation threshold, maintain the current filtering parameter; use the updated or maintained current filtering parameter as the next filtering parameter. In this embodiment, judgments are performed separately for each partitioned mixed sound signal channel and each reference input. The time delay retrieval range for cross-correlation calculation is −8ms to +8ms; at each time delay position, using the common sampling points after shifting, subtract the mean of the two signals in the common sampling interval, calculate the cumulative sum, and divide by the square root product of the sum of squares of the two signals. Take the absolute value of the calculation result, and then compare the absolute values ​​at each time delay position to obtain the maximum value of the cross-correlation function amplitude. Reference inputs with a denominator of zero do not participate in the parameter update for this frame. Time delay retrieval is used to determine the degree of synchronization change between the reference input and the partitioned mixed sound signal, without increasing the historical sampling length of noise estimation.

[0036] The correlation thresholds are determined separately for each combination of the zoned mixed acoustic signal channel and the reference input. For the static attachment calibration segment, the same framing and cross-correlation calculations are performed, and the maximum amplitude values ​​of the cross-correlation function obtained for the same combination in each background frame are sorted in ascending order. When the number of effective background frames is B, the value of the [number]th frame after sorting is taken. Each value is used as a corresponding threshold, and rounding up ensures that the arrangement position is an integer. Background frames with zero denominators are not included in the arrangement of the corresponding combination; if there are fewer than 20 valid background frames, a static calibration segment is re-acquired. Each reference input is judged and its corresponding filter parameters are updated separately, so that the update criteria for the environmental noise reference input and the update criteria for the motion reference input are executed separately.

[0037] For the reference input used in the parameter update, a block-based minimum mean square update formula is adopted: ; In the formula, For the next filtering parameters, This is the step size used in this frame. The summation process multiplies each sample value of the current partition's bowel sound signal frame with the reference sample value at the corresponding tap position, and then sums all the products. Historical sample values ​​are provided by a continuous buffer; missing historical sample values ​​are set to zero for the first processing. Reference inputs not yet updated to parameters retain the current filter parameters.

[0038] The step size is determined according to the actual filtered input. The central acquisition unit reads all tap samples from the four reference inputs at each sampling position, calculates the sum of squares of all sampled values, and then averages the sum of squares over a synchronization signal frame to obtain the reference input power. The base step size in this embodiment is 0.01; hour, Take 0.01 and The minimum value in; All filtering parameters are maintained. The reference input power includes all reference inputs and all taps, and the step size is updated according to the actual input amplitude. The processing results of each frame are arranged continuously according to the timestamp to obtain the zoned bowel sound signal. No parameter updates are performed on abnormal frames; when resuming continuous acquisition, the corresponding historical buffer is cleared, missing parts are set to zero, and the invalid flag is retained for abnormal frames. After completing the background correlation threshold calculation, the static attachment calibration segment is processed first, and then the formal monitoring data is processed continuously.

[0039] In S103, the detection and cross-channel association of bowel sound events include: screening candidate signal frames based on short-time energy or the proportion of bowel sound frequency band energy, merging consecutive candidate signal frames into candidate bowel sound events; calculating the occurrence time difference and spectral similarity of candidate bowel sound events in different abdominal regions, merging candidate bowel sound events whose time difference does not exceed the time threshold and whose spectral similarity is not lower than the similarity threshold, and retaining the remaining candidate bowel sound events to obtain bowel sound events.

[0040] The borborygmus signal was detected using a 100ms analysis window, with a 50ms frame shift between adjacent analysis windows, corresponding to 1600 and 800 sampling points respectively. Each analysis window used a Hanning window, and the short-time energy was the sum of squares of the windowed sample values. A 1600-point Discrete Fourier Transform was performed on the windowed signal to calculate the square of the amplitude at each discrete frequency point. The sum of squares of the amplitudes in the 20Hz to 2500Hz range was divided by the sum of squares of the amplitudes in the 0Hz to 8kHz range to obtain the energy proportion of the borborygmus audio band. When the denominator was zero, the energy proportion of the borborygmus audio band was zero, and analysis windows with zero short-time energy were not considered candidate signal frames. If an analysis window contained abnormally collected sampling points, the analysis window was not included in event detection.

[0041] The energy threshold and frequency band proportion threshold are determined using the noise-cancelled zoned bowel sound signals from the static attachment calibration segment. For each signal channel, the short-time energy of the effective analysis window is sorted in ascending order and the median is taken to obtain the energy baseline value; the absolute value of the difference between each short-time energy and the energy baseline value is calculated, and the median of the absolute values ​​is taken to obtain the energy discrete value. The energy threshold is the energy baseline value plus three times the energy discrete value. Using the same method, the frequency band baseline value and frequency band discrete value are calculated for the energy proportion of the bowel sound band, and the frequency band proportion threshold is the frequency band baseline value plus the frequency band discrete value. When the number of sorted segments is even, the median is the average of the two middle values. Three times and one time are the detection sensitivity settings selected in this embodiment. When there are fewer than 60 effective analysis windows, another 30-second static attachment calibration segment is collected; when the cumulative number of effective analysis windows is still less than 60, the monitoring is stopped and an attachment check prompt is output.

[0042] If the short-time energy of the analysis window is not lower than the energy threshold, or the energy proportion of the bowel sound band is not lower than the frequency band proportion threshold, the analysis window is used as a candidate signal frame. Consecutive candidate signal frames are merged into one candidate bowel sound event; non-candidate signal frames or those interrupted by acquisition abnormalities are merged. The start time of a candidate bowel sound event is the start timestamp of the first candidate signal frame, the end time is the end timestamp of the last candidate signal frame, and the occurrence time is the center timestamp of the analysis window with the highest short-time energy within the event; if the short-time energies are the same, the analysis window with the earlier time is selected. Each candidate bowel sound event saves its event number, signal channel number, start time, end time, and occurrence time.

[0043] The time compensation amount and similarity threshold used for cross-channel association were determined during the equipment calibration phase. During calibration, the abdominal contact acoustic head was positioned on the tissue-equivalent acoustic phantom according to the positioning template, and mechanical excitation was applied at each placement position using an electric vibrator. The driving signals were 100Hz, 400Hz, and 1000Hz sinusoidal signals, each lasting 100ms, with a 2s interval between adjacent excitation start times. Each frequency was repeated 10 times at each excitation position. The excitation number, position, driving frequency, and trigger time for each excitation were saved, and each signal channel was recorded synchronously. The same framing, noise cancellation, and candidate bowel sound event detection procedures were used for calibration acquisition. Cross-channel candidate bowel sound events belonging to the same excitation were grouped into homologous event pairs; cross-channel candidate bowel sound events belonging to excitations of different driving frequencies were grouped into heterologous event pairs. Excitations of the same frequency occurring at different times were retained with excitation numbers, and event differentiation was mainly based on time conditions. Mechanical excitation was used to generate traceable calibration input and was not used as a verification result for the accuracy of human bowel sound region discrimination.

[0044] The time threshold between different signal channels is calculated using the following formula: ; In the formula, b is the number of another signal channel that is different from channel a; This is the time threshold between two signal channels, in seconds. The distance between the centers of the two abdominal contact acoustic heads, in meters; The nominal sound velocity used for soft tissue sound propagation time calculation is 1540 m / s, which is used in medical ultrasound propagation time conversion in this embodiment. To analyze the number of sampling points for the frame shift in the analysis window; This represents the synchronization tolerance corresponding to one audio sampling period. This is the compensation time for the event occurrence. For each pair of homogeneous events corresponding to the channel combination, calculate the absolute value of the time difference between occurrences, subtract the propagation time, frame shift time, and synchronization tolerance calculated based on the calibrated distance, and take the maximum value between all remaining values ​​and zero as the compensation time. When saving the compensation time, the positioning template and sampling configuration number are also saved. Recalibration is required after any changes to the layout or sampling configuration. The time threshold is used for candidate event pairing, and the compensation time is used to account for the time offset caused by differences in waveform and detection position during the calibration process.

[0045] Spectral similarity is calculated using the cosine similarity between the power spectral vectors of two candidate bowel sound events. Let u and v represent the event numbers, respectively. The larger of the sampling point counts for both events is taken, and then the least square power of the larger sampling point count is selected as the common transform point number. Both events are windowed according to their event length, padded with zeros to the common transform point number, and then subjected to a Discrete Fourier Transform with the same number of points. The squared amplitudes of the same discrete frequency points within the 20Hz to 2500Hz range are extracted to form... and , where q is the frequency point number within the common frequency grid. The spectral similarity is: ; In the formula, Let be the spectral similarity between candidate bowel sound events u and v. The spectral similarity is zero when the sum of the squares of any power spectral vector is zero; otherwise, the spectral similarity ranges from 0 to 1. The number of common transformation points ensures that corresponding elements in the two power spectral vectors use the same physical frequency.

[0046] Similarity thresholds are calculated using both homologous and heterologous event pairs. Values ​​between 0 and 1, with intervals of 0.01, are sequentially used as candidate thresholds. For each candidate threshold, the number of homologous event pairs with spectral similarity no lower than the candidate threshold is counted, and this count is divided by the total number of homologous event pairs to obtain the homologous event pair retention ratio. Simultaneously, the number of heterologous event pairs with spectral similarity no lower than the candidate threshold is also counted. Among the candidate thresholds with a homologous event pair retention ratio of no less than 95%, the candidate threshold with the fewest heterologous event pairs is selected; if multiple values ​​exist, the smallest value is selected. 95% is the homologous event pair retention ratio setting used in the calibration procedure. If the number of homologous or heterologous event pairs is zero, no similarity threshold is generated for this calibration. If the time threshold, compensation time, or similarity threshold is not generated, cross-channel association and regional quantization output are not initiated. Spectral similarity and time difference are used together for event pairing; different events with similar spectra continue to undergo time-conditional testing.

[0047] Multiple event merging is performed in a predetermined order. First, all candidate bowel sound events are sorted in ascending order of occurrence time. If occurrence times are the same, they are sorted in ascending order of signal channel number. Within the same channel, they are further sorted in ascending order of candidate bowel sound event number. The candidate bowel sound events that are ranked first and not yet grouped are selected to form an event group. The remaining ungrouped events are then checked sequentially. If the occurrence time difference between the event to be checked and each event in the event group does not exceed the corresponding time threshold, the spectral similarity is not lower than the similarity threshold, and the signal channel containing the event to be checked does not yet exist in the event group, the merged event time interval is determined according to the earliest start time and latest end time after joining. The sampling integrity of each member in the event group and the signal channel containing the event to be checked is checked within the merged event time interval. If no abnormal sampling points are found in any signal channel, the event to be checked is added to the event group. If any abnormal sampling point exists in any signal channel, the event to be checked remains ungrouped. After completing this round of checks, a bowel sound event number is assigned to the event group, and the process is repeated from the remaining ungrouped events until all candidate bowel sound events are grouped.

[0048] Event groups containing two or more signal channels are merged into a single bowel sound event; event groups containing only one candidate bowel sound event retain the candidate event and treat it as a single bowel sound event. The merged event saves the channel numbers and event time information of all members, with the time interval being the earliest start time to the latest end time of each member's event. Each pair of members within a group undergoes time difference and spectral similarity checks. Only one member from the same signal channel is retained in an event group to avoid merging multiple consecutive sounds due to indirect association. The correspondence record between each member and the bowel sound event number is used for subsequent region determination and deduplication counting.

[0049] In S104, the determination of the occurrence region includes: for the bowel sound events obtained by merging, determining the arrival time difference, signal energy difference, and spectral characteristics based on the corresponding signals of the bowel sound events in different abdominal partitions, and sorting the abdominal partitions respectively; when the first partition of each sort is the same, the first partition is determined as the occurrence region; when the first partitions of each sort are different, the occurrence region is determined based on the sorting results; for the bowel sound events obtained by retention, the abdominal partition corresponding to the signal channel where the bowel sound event is located is determined as the occurrence region.

[0050] For the merged bowel sound events, the zoned bowel sound signals within the merged event time interval are extracted from the signal channels of each member. The signal channel of the member whose occurrence time is earlier is selected as the reference channel. If the occurrence times are the same, the channel number of the earlier member is selected, and the relative arrival time of the reference channel is set to zero. For other signal channels, the event signal is moved point by point within the corresponding time threshold, and the amplitude of the cross-correlation function with the event signal of the reference channel is calculated. The movement time corresponding to the peak value of the amplitude is used to determine the relative arrival time. When the signal to be compared forms a peak after moving in the direction of decreasing time, the relative arrival time is taken as positive; when it forms a peak after moving in the direction of increasing time, the relative arrival time is taken as negative. When the same peak value appears at multiple time delay positions, the time delay with the smallest absolute value is taken first. If the absolute values ​​are still the same, the time delay with the smallest numerical value is taken. The relative arrival time from small to large corresponds to the arrival from early to late. The reference channel selection and time delay sign rules apply to all merged events.

[0051] The acoustic energy integral is calculated using the sum of squares of the sampled values ​​of the bowel sound signals within the event time interval as the discrete value of the signal energy. The difference in signal energy between two abdominal regions is the difference in their corresponding acoustic energy integrals; when the difference is positive, the energy of the preceding abdominal region is ranked first, and when the difference is zero, they are ranked equally. All channels use the same sampling rate and fixed gain. The acoustic energy integral is used for relative comparisons between events and between abdominal regions, and the output uses digital signal energy units. The spectral characteristics are calculated using the energy proportion of the bowel sound frequency band, calculated by dividing the sum of squares of the spectral amplitudes of the event signal in the range of 20Hz to 2500Hz by the sum of squares of the spectral amplitudes in the range of 0Hz to 8kHz, with zero taken as the denominator. Time, energy, and spectral indicators together describe the surface acoustic response of different abdominal regions.

[0052] The sorting of abdominal regions and the determination of the occurrence region based on the sorting results include: determining the arrival order based on the time delay of the peak value of the cross-correlation function, determining the energy level based on the acoustic energy integral, and determining the spectral characteristics based on the energy proportion of the bowel sound band; assigning ranks according to arrival time from earliest to latest, energy from highest to lowest, and energy proportion of the bowel sound band from highest to lowest; adding the ranks of each abdominal region in each sorting, and determining one or more abdominal regions with the smallest sum of ranks as the occurrence region. Each sorting starts from rank 1, with regions having the same value using the same rank, and the rank of the next different value increments by 1. When the first set of regions in the three sortings is the same, the abdominal regions within that set are determined as the occurrence region; when the first set of regions are different, the sum of ranks is used to determine the occurrence region. When the occurrence region consists of multiple abdominal regions, all region numbers are output. The sorting results are used for acoustic attribution at the abdominal region level, and the surface attachment location, coupling state, and differences in tissue propagation are all conditions for using the region determination results.

[0053] For bowel sound events that have been preserved, the central data acquisition unit reads the signal channel number and looks up the abdominal region in a one-to-one correspondence table between signal channels and abdominal regions. The abdominal region found in the lookup is then identified as the occurrence area. Preserved events retain their original event number, time interval, and signal channel number, and are then included in the regional statistics. Events detected by a single channel are acoustically assigned based on their detection location, while events detected by multiple channels are acoustically assigned based on acoustic response comparison. Both types of events generate occurrence areas and event records.

[0054] In S105, the intestinal motility quantification parameters are obtained by: determining the effective monitoring period for each abdominal region based on the channel detachment status and motion artifacts; retaining a count for the same bowel sound event in each occurrence region; and calculating at least one of the event rate, duration percentage, and average sound energy for each occurrence region according to the effective monitoring period to obtain the intestinal motility quantification parameters.

[0055] The channel detachment status is determined based on acoustic channel input fault detection, sampling continuity checks, and manual removal records. When an input fault detection report indicates an open or short circuit, interrupted sampling count, or missing data packet, the corresponding signal channel's synchronization signal frame is marked invalid. The manual removal record includes the removal time and signal channel number. From the removal time until the reattachment time of the record, the corresponding signal channel remains marked invalid; if no reattachment record exists before the end of monitoring, the invalid period continues until the end of monitoring. Synchronization signal frames intersecting with invalid periods are marked invalid. The attachment record is updated after the abdominal contact acoustic head is reattached. Motion artifacts are determined using 100Hz raw triaxial motion data: for each motion sampling moment, the sum of the squares of the triaxial accelerations is taken as the square root to obtain the triaxial resultant acceleration; the maximum value of the triaxial resultant acceleration within each synchronization signal frame is subtracted from the minimum value to obtain the change in resultant acceleration. The change in resultant acceleration of each frame within the static attachment calibration segment is sorted in ascending order, and the resultant acceleration of the next frame is taken. A numerical value is used as the motion threshold, and C is the number of valid calibration frames. When the resultant acceleration change of a formal monitoring frame exceeds the motion threshold, it is marked as a motion artifact frame. Synchronization signal frames with missing motion data are also marked as invalid. The valid monitoring period is the time period retained after excluding calibration periods, periods corresponding to channel detachment, periods of abnormal acquisition, and periods of motion artifacts.

[0056] For each occurrence area, the intersection of the bowel sound event time interval and the corresponding effective monitoring period of the abdominal zone is calculated. Each consecutive time interval within the intersection is considered a valid event time interval. Events with an empty intersection are not included in the statistics for that area; events with a non-empty intersection are counted once according to their bowel sound event number, and the same event number is retained even when the intersection is divided into multiple time intervals. Duration only uses the effective event time interval, and acoustic energy only uses the sampled bowel sound signal values ​​of the corresponding signal channel within the effective event time interval. When a bowel sound event corresponds to multiple occurrence areas, the intersection and counting are performed separately for each occurrence area, and the acoustic energy integral of the corresponding signal channel for each occurrence area is used. Event counts, durations, and acoustic energy use the same effective monitoring period to avoid invalid time intervals being included in the statistics even after being removed from the monitoring duration.

[0057] by Indicates the area number where the event occurred. This represents the number of events in the non-empty intersection after deduplication by event number. This indicates the effective monitoring duration for the corresponding abdominal region, in seconds. This represents the total duration after merging overlapping intervals of all valid event time periods, in seconds. This represents the sum of the acoustic energy integrals of each counted event within the valid event time period. Event rate. Duration percentage and average sound energy Calculate according to the following formula: ; When the effective monitoring duration is zero, a status flag is output, and no values ​​are generated for event rate, duration percentage, and average acoustic energy. and At this time, the event rate, duration percentage, and average acoustic energy are all set to zero. The event rate is in seconds, the duration percentage ranges from 0 to 1, and the average acoustic energy uses digital signal energy units consistent with the acoustic energy integral. The output also includes the abdominal zone, effective monitoring duration, and number of events for verification of the zone statistics results.

[0058] like Figure 4As shown, this embodiment also provides a quantitative assessment system for intestinal motility based on multi-zone bowel sound monitoring, including multiple abdominal contact acoustic heads, a reference noise acquisition channel, and a central acquisition unit. The central acquisition unit is equipped with a noise cancellation module, an event detection module, an event association module, a region determination module, and a parameter generation module. The multiple abdominal contact acoustic heads are used to attach to different abdominal zones and acquire multi-zone mixed acoustic signals; the reference noise acquisition channel is used to acquire reference signals; the central acquisition unit is connected to the multiple abdominal contact acoustic heads and the reference noise acquisition channel to synchronously receive each signal and write a timestamp.

[0059] The central acquisition unit includes a multi-channel audio ADC, a main controller, and a memory. The multi-channel audio ADC receives the differential electrical signal output from the abdominal contact acoustic head, while the main controller receives the outputs from the multi-channel audio ADC, an ambient noise reference microphone, and a triaxial motion sensor. The memory stores a one-to-one correspondence table between signal channels and abdominal zones, sampling configurations, amplitude correction values, calibration parameters, raw signals, and event records. A noise cancellation module estimates and cancels noise in the mixed acoustic signals of each zone based on the reference signal, obtaining the zoned bowel sound signals. An event detection module detects candidate bowel sound events within the zoned bowel sound signals. An event association module associates candidate bowel sound events across channels and merges or retains them based on the association results, obtaining the bowel sound events. A region determination module determines the occurrence region based on the association results, the acoustic characteristics of the bowel sound events, and the corresponding abdominal zone. A parameter generation module statistically analyzes bowel sound events according to their occurrence regions, obtaining quantitative parameters of intestinal motility.

[0060] Each functional module is implemented by the program in the main controller's execution memory. During system operation, the main controller first receives multiple signals with sampling counts and timestamps, completing time alignment, calibration parameter reading, and synchronous framing. The noise cancellation module generates zoned bowel sound signals frame by frame and updates or maintains the filtering parameters for each group. The event detection module forms candidate bowel sound events, and the event association module generates bowel sound event numbers and member records according to a determined grouping order. The region determination module outputs the occurrence region, and the parameter generation module completes deduplication counting, duration calculation, and acoustic energy summarization according to the effective monitoring period, outputting at least one of the event rate, duration percentage, and average acoustic energy. Hardware acquisition, event processing, and zone statistics use consistent timestamps, signal channel numbers, and bowel sound event numbers to form monitoring results that can be verified item by item.

[0061] The above embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solutions based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring, characterized in that, include: S101 collects multi-channel mixed acoustic signals by multiple abdominal contact acoustic heads located in different abdominal zones, collects reference signals by a reference noise acquisition channel, and is synchronously received and written to a timestamp by the central acquisition unit. S102, based on the reference signal, estimate and cancel the noise in the partitioned mixed sound signal to obtain the partitioned bowel sound signal; S103, detect candidate bowel sound events in the zoned bowel sound signals, correlate candidate bowel sound events across channels, and merge or retain candidate bowel sound events according to the correlation results to obtain bowel sound events; S104. The occurrence area is determined based on the correlation results, the acoustic characteristics of bowel sound events, and the corresponding abdominal regions. S105, according to the occurrence area, count bowel sound events to obtain quantitative parameters of intestinal motility.

2. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 1, characterized in that, The plurality of abdominal contact acoustic heads are respectively attached to at least two abdominal regions in the upper, middle and lower abdomen; the reference signals include the environmental noise signal collected by the environmental noise reference microphone and the motion reference signal collected by the triaxial motion sensor.

3. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 2, characterized in that, The synchronous reception and writing of timestamps includes: multiple abdominal contact acoustic heads are connected to the multi-channel audio ADC of the central acquisition unit via flexible differential lines and share a sampling clock; the output of the multi-channel audio ADC, the ambient noise reference microphone and the triaxial motion sensor are respectively connected to the main control of the central acquisition unit; the main control writes timestamps for each signal to obtain a time-stamped partitioned mixed sound signal and a reference signal.

4. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 1, characterized in that, The noise cancellation includes: synchronizing and framing the reference signal and the partitioned mixed sound signal; using the current filtering parameters, estimating the noise in the partitioned mixed sound signal frame based on the reference signal frame to obtain a noise estimation frame; subtracting the noise estimation frame from the partitioned mixed sound signal frame to obtain the current partitioned bowel sound signal frame; determining the next filtering parameters based on the reference signal frame and the current partitioned bowel sound signal frame; using the next filtering parameters in the next synchronized signal frame, and performing noise estimation, subtraction, and filtering parameter determination frame by frame to obtain the partitioned bowel sound signal.

5. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 4, characterized in that, The next filtering parameter is determined as follows: calculate the maximum value of the cross-correlation function amplitude between the reference signal frame and the synchronized partitioned mixed sound signal frame; when the maximum value is not lower than the correlation threshold, update the current filtering parameter according to the least mean square criterion based on the reference signal frame and the current partitioned bowel sound signal frame; when the maximum value is lower than the correlation threshold, keep the current filtering parameter; use the updated or kept current filtering parameter as the next filtering parameter.

6. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 1, characterized in that, The detection and cross-channel association of bowel sound events include: screening candidate signal frames based on short-time energy or the proportion of bowel sound frequency band energy; merging consecutive candidate signal frames into candidate bowel sound events; calculating the occurrence time difference and spectral similarity of candidate bowel sound events in different abdominal regions; merging candidate bowel sound events whose time difference does not exceed the time threshold and whose spectral similarity is not lower than the similarity threshold; and retaining the remaining candidate bowel sound events to obtain bowel sound events.

7. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 1, characterized in that, The determination of the occurrence region includes: for the merged bowel sound events, determining the arrival time difference, signal energy difference, and spectral characteristics based on the corresponding signals of the bowel sound events in different abdominal regions, and sorting the abdominal regions respectively; when the first region of each sort is the same, the first region is determined as the occurrence region; when the first regions of each sort are different, the occurrence region is determined based on the sorting results; for the retained bowel sound events, the abdominal region corresponding to the signal channel where the bowel sound event is located is determined as the occurrence region.

8. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 7, characterized in that, The sorting of the abdominal regions and the determination of the occurrence region based on the sorting results include: determining the arrival order based on the time delay of the peak value of the cross-correlation function, determining the energy level based on the acoustic energy integral, and determining the spectral characteristics based on the energy ratio of the bowel sound band; assigning ranks according to arrival from early to late, energy from high to low, and energy ratio of the bowel sound band from high to low; adding the ranks of each abdominal region in the sorting, and determining one or more abdominal regions with the smallest sum of ranks as the occurrence region.

9. The method for quantitative assessment of intestinal motility based on multi-zone bowel sound monitoring according to claim 1, characterized in that, The intestinal motility quantification parameters are obtained by: determining the effective monitoring period for each abdominal region based on the channel detachment status and motion artifacts; retaining a count for the same bowel sound event in each occurrence region; and calculating at least one of the event rate, duration percentage, and average sound energy for each occurrence region according to the effective monitoring period to obtain the intestinal motility quantification parameters.

10. A quantitative assessment system for intestinal motility based on multi-zone bowel sound monitoring, characterized in that, It includes multiple abdominal contact acoustic heads, a reference noise acquisition channel, and a central acquisition unit. The central acquisition unit is equipped with a noise cancellation module, an event detection module, an event association module, a region determination module, and a parameter generation module. Multiple abdominal contact acoustic heads are used to attach to different abdominal zones and collect multi-zone mixed acoustic signals; The reference noise acquisition channel is used to acquire reference signals; The central acquisition unit is connected to multiple abdominal contact acoustic heads and reference noise acquisition channels to synchronously receive each signal and write timestamps. The noise cancellation module is used to estimate and cancel noise in the partitioned mixed sound signal based on the reference signal to obtain the partitioned bowel sound signal; The event detection module is used to detect candidate bowel sound events in the zoned bowel sound signal; The event association module is used to associate candidate bowel sound events across channels, and to merge or retain candidate bowel sound events according to the association results to obtain bowel sound events; The region determination module is used to determine the region of occurrence based on the correlation results, the acoustic characteristics of bowel sound events, and the corresponding abdominal regions; The parameter generation module is used to statistically analyze bowel sound events according to the occurrence area and obtain quantitative parameters of intestinal motility.

Images