Method and device for swallowing impairment detection

• US 9,687,191 B2
• Filed: 01/18/2012
• Issued: 06/27/2017
• Est. Priority Date: 01/18/2011
• Status: Active Grant

• Current Assignee: Holland Bloorview Kids Rehabilitation Hospital, University Health Network
• Original Assignee: Holland Bloorview Kids Rehabilitation Hospital, University Health Network
• Inventors: Chau, Thomas T. K., Steele, Catriona, Sejdic, Ervin, Maruzzo, Bruno C.
• Primary Examiner(s): Henson, Devin
• Assistant Examiner(s): Kremer, Matthew
• Application Number:
  US13/980,510
• Publication Number:
  US 20140228714A1
• Time in Patent Office: 1,987 Days
• Field of Search: None
• CPC Class Codes: A61B 5/1121 Determining geometric values, e.g. centre of rotation or angular range of movement
• CPC Class Codes: A61B 5/1123 Discriminating type of movement, e.g. walking or running A61B5/1116, A61B5/112 take precedence
• CPC Class Codes: A61B 5/1126 using a particular sensing technique
• CPC Class Codes: A61B 5/4205 Evaluating swallowing
• CPC Class Codes: A61B 5/6843 Monitoring or controlling sensor contact pressure
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: A61B 5/7282 Event detection, e.g. detecting unique waveforms indicative of a medical condition A61B5/4094, A61B5/4818, A61B5/349, A61B5/0823 take precedence
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

A method and an apparatus for swallowing impairment detection acquire dual axis accelerometry data representative of one or more swallowing events executed by a candidate. Upon feature extraction and classification, vibrational data acquired in respect of each swallowing event is classified as indicative of one of normal or possibly impaired swallowing. Computer-readable media comprising statements and instructions for implementation by a processing device are also described in facilitating swallowing impairment detection respective to candidate swallowing events.

Claims

1. A device for use in identifying a possible swallowing impairment in a candidate during execution of a swallowing event, the device comprising:

a dual axis accelerometer configured for alignment along an anterior-posterior (A-P) axis and a superior-inferior (S-I) axis of the candidate'"'"'s throat and to acquire axis-specific vibrational data representative of the swallowing event; and

a processing module that is a local or remote computing device operatively coupled to the accelerometer, the processing module configured for processing the axis-specific data to (i) extract therefrom distinct axis-specific features representative of the swallowing event, the distinct axis-specific features comprising a log energy of vibrational data acquired along the A-P axis and absent a log energy of vibrational data acquired along the S-I axis, the distinct axis-specific features further comprising an entropy of vibrational data acquired along the S-I axis and absent an entropy of vibrational data acquired along the A-P axis, (ii) perform a comparison of the distinct axis-specific features against preset classification criteria defined for each of swallowing safety and swallowing efficiency, and (iii) classify the vibrational data as one of a first classification or a second classification based on the comparison of the distinct axis-specific features against the preset classification criteria defined for each of the swallowing safety and the swallowing efficiency.

2. The device of claim 1, wherein the second classification is indicative of at least one of a swallowing safety impairment or a swallowing efficiency impairment.

3. The device of claim 1, wherein the second classification is indicative of penetration or aspiration, the processing module configured to further classify the swallowing event as indicative of one of a first event and a second event based on the comparison of the distinct axis-specific features against the preset classification criteria defined for each of the swallowing safety and the swallowing efficiency.

4. The device of claim 1, wherein the log energy of the vibrational data acquired along the A-P axis data comprises a log energy of each level of an 18 level sym8 wavelet decomposition of the vibrational data acquired along the A-P axis, and the entropy of the vibrational data acquired along the S-I axis data comprises an entropy of each level of a 12 level sym8 wavelet decomposition of the vibrational data acquired along the S-I axis.

5. The device of claim 4, wherein the distinct axis-specific features comprise a weighted sum of the log energy of each level of the 18 level sym8 wavelet decomposition of the vibrational data acquired along the A-P axis, and a weighted sum of the entropy of each level of the 12 level sym8 wavelet decomposition of the vibrational data acquired along the S-I axis, wherein weights in each of the weighted sums are defined by preset feature reduction parameters.

6. The device of claim 1, wherein the preset classification criteria are defined by features previously extracted and classified from a known training data set.

7. The device of claim 1, wherein the distinct axis-specific features are classified as a function of distances of values of the distinct axis-specific features from values of the preset classification criteria.

8. The device of claim 7, wherein the distinct axis-specific features are classified as the function of the distances via discriminant analysis using Mahalanobis distances with stratified covariance estimates.

9. The device of claim 1, wherein the classification of the vibrational data as one of the first classification or the second classification comprises identifying, based on preset feature reduction parameters, predominant components of the distinct axis-specific features and the classification of the vibrational data is based on the distinct axis-specific features and the predominant components thereof.

10. The device of claim 1, configured to identify the second classification during execution of multiple successive swallowing events, by the processing module classifying the vibrational data acquired in respect of each of the successive swallowing events as indicative of one of the first classification or the second classification.

11. The device of claim 10, wherein the processing module is configured for automatically segmenting the vibrational data acquired in respect of each of the successive swallowing events for independent processing and classification.

12. The device of claim 10, comprising a user interface for selectively segmenting the vibrational data acquired in respect of each of the successive swallowing events for independent processing and classification.

13. The device of claim 10, comprising a display, the processing module further configured to process the vibrational data acquired in respect of each of the successive swallowing events, and output the one of the first classification or the second classification on the display in accordance with a preset swallowing impairment assessment protocol.

14. The device of claim 1, comprising a pressure sensor or pressure sensitive film positioned so to monitor a pressure applied by the accelerometer when positioned on the candidate'"'"'s throat to increase data acquisition accuracy.

15. A method for classifying cervical dual-axis accelerometry data acquired in respect of a candidate swallowing event to identify a possible swallowing impairment, the method comprising:

receiving as input axis-specific vibrational data from an anterior-posterior (A-P) axis and a superior-inferior (S-I) axis and representative of the swallowing event, a processing module that is a local or remote computing device operatively coupled to a dual axis accelerometer receives the axis-specific vibrational data from the dual axis accelerometer;

extracting distinct axis-specific features representative of the swallowing event from the axis-specific vibrational data, the processing module extracts the distinct axis-specific features, the distinct axis-specific features comprising a log energy of vibrational data acquired along the A-P axis and absent a log energy of vibrational data acquired along the S-I axis, the distinct axis-specific features further comprising an entropy of vibrational data acquired along the S-I axis and absent an entropy of vibrational data acquired along the A-P axis;

comparing the distinct axis-specific features with preset classification criteria defined as a function of the distinct axis-specific features, the processing module compares the distinct axis-specific features with the preset classification criteria; and

outputting, based on the comparing step, classification of the vibrational data as one of a first classification or a second classification, the processing module outputs the classification.

16. The method of claim 15, wherein the second classification is indicative of one or more of unsafe swallowing and inefficient swallowing.

17. The method of claim 15, wherein the second classification is indicative of at least one of a penetration risk and an aspiration risk.

18. The method of claim 15, wherein the log energy of the vibrational data acquired along of the A-P axis comprises a log energy of each level of an 18 level sym8 wavelet decomposition of the vibrational data acquired along the A-P axis, and the entropy of the S-I axis data comprises an entropy of each level of a 12 level sym8 wavelet decomposition of the vibrational data acquired along the S-I axis.

19. The method of claim 18, comprising reducing the distinct axis-specific features by associating respective weights to each level of the 18 level sym8 wavelet decomposition of the vibrational data acquired along the A-P axis and each level of the 12 level sym8 wavelet decomposition of the vibrational data acquired along the S-I axis.

20. The method of claim 15, wherein the comparing step comprises calculating distances of values of the distinct axis-specific features from values of the preset classification criteria and selecting a most likely classification as a function of the distances.

21. The method of claim 20, wherein the calculating of the distances comprises performing a discriminant analysis of the distinct axis-specific features using Mahalanobis distances with stratified covariance estimates.

22. The method of claim 15, wherein the receiving step comprises receiving as input axis-specific vibrational data representative of successive swallowing events, the method comprising segmenting the vibrational data into event-specific data, and repeating the extracting and comparing steps for each of the event-specific data.

23. The method of claim 22, wherein the segmenting step is implemented automatically.

24. The method of claim 22, wherein the outputting step comprises outputting the classification of the vibrational data as one of the first classification or the second classification in accordance with a preset swallowing impairment assessment protocol.

25. A method for identifying a possible swallowing impairment in a candidate via execution of one or more preset swallowing events, comprising:

recording dual-axis vibrational data along an anterior-posterior (A-P) axis and a superior-inferior (S-I) axis, the dual-axis vibrational data is representative of the one or more preset swallowing events, a processing module that is a local or remote computing device operatively coupled to a dual axis accelerometer records the dual-axis vibrational data from the dual axis accelerometer;

extracting distinct axis-specific time-frequency domain features from the dual-axis vibrational data, the processing module extracts the distinct axis-specific time-frequency domain features, the distinct axis-specific time-frequency domain features comprising a log energy of vibrational data acquired along the A-P axis and absent a log energy of vibrational data acquired along the S-I axis, the distinct axis-specific time-frequency domain features further comprising an entropy of vibrational data acquired along the S-I axis and absent an entropy of vibrational data acquired along the A-P axis; and

classifying the distinct axis-specific time-frequency domain features as one of a first classification or a second classification, the processing module classifies the distinct axis-specific time-frequency domain features.

26. The method of claim 25, further comprising, during the recording step, having the candidate execute the one or more swallowing events.

27. The method of claim 25, comprising, for execution of two or more preset swallowing events, selectively recording the dual-axis vibrational data for each event independently to provide event-specific data, and implementing the extracting and classifying steps on the event-specific data for each of the events.

28. The method of claim 27, comprising identifying the candidate as exhibiting a potential swallowing impairment upon classifying the distinct axis-specific time-frequency domain features of at least one of the two or more preset swallowing events as indicative of possible swallowing impairment.

29. The method of claim 28, wherein the potential swallowing impairment comprises unsafe swallowing, the method comprising identifying the candidate as a low aspiration risk candidate upon classifying at least three successive swallowing events as not indicative of possible swallowing impairment.

30. The method of claim 25, comprising, for execution of two or more preset swallowing events, successively recording the dual-axis vibrational data for each of the events, and comprising automatically segmenting the successively recorded data to provide event-specific data, and implementing the extracting and classifying steps on the event-specific data for each of the events.

31. The method of claim 25, wherein the second classification is indicative of at least one impairment selected from the group consisting of penetration, aspiration, unsafe swallowing and inefficient swallowing.

32. The method of claim 25, wherein the second classification is indicative of one or more of unsafe swallowing and inefficient swallowing, the classifying step comprising comparing the distinct axis-specific time-frequency domain features against respective preset classification criteria defined for each of swallowing safety and swallowing efficiency.

33. The method of claim 25, comprising testing a dual-axis vibrational data quality for consistency with preset quality characteristics.