Non-invasive method and system for estimating arterial flow characteristics
The present disclosure facilitates the evaluation of wide-band phase gradient information of the heart tissue to assess, e.g., the presence of heart ischemic heart disease. Notably, the present disclosure provides an improved and efficient method to identify and risk stratify coronary stenosis of the heart using a high resolution and wide-band cardiac gradient obtained from the patient. The patient data are derived from the cardiac gradient waveforms across one or more leads, in some embodiments, resulting in high-dimensional data and long cardiac gradient records that exhibit complex nonlinear variability. Space-time analysis, via numeric wavelet operators, is used to study the morphology of the cardiac gradient data as a phase space dataset by extracting dynamical and geometrical properties from the phase space dataset.
1. A system for detecting one or more cardiac pathologies of a human subject, comprising:
an analysis system configured with computer readable instructions to retrieve an electrophysiological signal data set that (i) is associated with the subject and (ii) has been acquired by measuring equipment configured to non-invasively capture, via one or more sensors, the signal data set as a long cardiac gradient record that exhibits complex non-linear variability, the analysis system using the signal data set to detect the one or more cardiac pathologies of the subject,
wherein the detection operation comprises:
generating, by a processor executing a transformation operation, an intermediate residue data set from the signal data set, wherein the intermediate residue data set comprises an alpha shape or a Delaunay triangle mesh data object of the signal data set; and
generating, by the processor, one or more geometric features and/or dynamical properties of the alpha shape or the Delaunay triangle mesh, wherein the geometric features and/or the dynamical properties include at least one of:
a 3D volume value,
a void value,
a surface area value,
a principal curvature direction value, and
a Betti number value,
of the alpha shape or the Delaunay triangle mesh,
wherein the generated one or more geometric features and/or dynamical properties of the alpha shape or the Delaunay triangle mesh are used as variables representative of the subject in a machine learning operation to detect the one or more cardiac pathologies of the subject.
2. The system of claim 1 , wherein the intermediate residue data set is generated by:
generating, by the processor, a model of the signal data set;
determining, by the processor, a residue data object between the model and the signal data set, wherein the alpha shape or a Delaunay triangle mesh data object is generated from the residue data object of the signal data set.
3. The system of claim 1 , wherein the signal data set includes cardiac data acquired via two or more channels, each of a biopotential signal acquired over a plurality of heart beats as the long cardiac gradient record.
4. The system of claim 3 , wherein the signal data set is acquired for at least 210 seconds, and wherein a portion of the long cardiac gradient record is modeled to generate the residue data object.
5. The system of claim 1 , wherein the signal data set comprises a plurality of wide-band gradient signals simultaneously captured via a plurality of channels, including the two or more channels, and wherein a temporal lag or skew between each of the channels is less than about 10 femtoseconds.
6. The system of claim 5 , wherein each of the plurality of wide-band gradient signals is unfiltered prior to, and during, the transformation operation, to generate the signal data set.
7. The system of claim 5 , wherein each of the plurality of wide-band gradient signals comprises cardiac data that in a frequency domain have frequency components greater than about 1 kHz.
8. The system of claim 5 , wherein each of the plurality of wide-band gradient signals
comprises cardiac data that in a frequency domain have frequency components of at least 8 kHz.
9. The system of claim 1 , wherein the one or more sensors comprise at least one of a surface electrode, an intracardiac electrode, and a non-contact electrode.
10. The system of claim 1 , wherein the machine learning operation comprises an artificial neural network algorithm or a regression random forest algorithm.
11. The system of claim 1 , wherein the measuring equipment is configured to transmit the signal data set over a network to the analysis system.
12. The system of claim 11 , further comprising:
a storage area network, wherein the measuring equipment is configured to transmit the signal data set over the network to the storage area network, and wherein the analysis system is configured to retrieve the acquired signal data set from the storage area network for analysis.
13. A method comprising:
retrieving, by a processor, from a storage area network, an electrophysiological signal data set that (i) is associated with a human subject and (ii) has been acquired by measuring equipment configured to non-invasively capture, via one or more sensors, the signal data set as a long cardiac gradient record that exhibits complex non-linear variability; and
detecting one or more cardiac pathologies of the subject using the signal data set, wherein the detection operation comprises:
generating, by the processor executing a transformation operation, an intermediate residue data set from the signal data set, wherein the intermediate residue data set comprises an alpha shape or a Delaunay triangle mesh data object of the signal data set; and
generating, by the processor, one or more geometric features and/or the dynamical properties of the alpha shape or the Delaunay triangle mesh, wherein the geometric features and/or the dynamical properties include at least one of:
a 3D volume value,
a void value,
a surface area value,
a principal curvature direction value and
a Betti number value,
of the alpha shape or the Delaunay triangle mesh,
wherein the generated one or more geometric features and/or dynamical properties of the alpha shape or the Delaunay triangle mesh are used as variables representative of the subject in a machine learning operation to detect the one or more cardiac pathologies of the subject.
14. The method of claim 13 , wherein the wherein the intermediate residue data set is generated by:
generating, by the processor, a model of the signal data set;
determining, by the processor, a residue data object between the model and the signal data set, wherein the alpha shape or a Delaunay triangle mesh data object is generated from the residue data object of the signal data set.
15. The method of claim 13 , wherein the signal data set includes cardiac data acquired via at least one channel, each of a biopotential signal acquired over a plurality of heart beats as the long cardiac gradient record.
16. The method of claim 15 , wherein the electrophysiological signal data set are acquired for at least 210 seconds, and wherein a portion of the long cardiac gradient record are modeled to generate the residue data object.
17. The method of claim 13 , wherein the signal data set comprises a plurality of wide-band gradient signals simultaneously captured via a plurality of channels, including the at least one channel, and wherein a temporal lag or skew between each of the channels is less than about 10 femtoseconds.
18. The method of claim 17 , wherein each of the plurality of wide-band gradient signals is unfiltered prior to, and during, the transformation operation, to generate the signal data set.
19. The method of claim 17 , wherein each of the plurality of wide-band gradient signals comprises cardiac data that in a frequency domain have frequency components greater than about 1 kHz.
20. The method of claim 17 , wherein each of the plurality of wide-band gradient signals comprises cardiac data that in a frequency domain have frequency components of at least 8 kHz.
21. A system for detecting one or more pathologies of a human subject, comprising:
an analysis system configured with computer readable instructions to retrieve an electrophysiological signal data set that (i) is associated with the subject and (ii) has been acquired by measuring equipment configured to non-invasively capture, via one or more sensors, the signal data set as a long cardiac gradient record that exhibits complex non-linear variability, the analysis system using the signal data set to detect the one or more pathologies of the subject,
wherein the detection operation comprises:
generating, by a processor executing a transformation operation, an intermediate residue data set from the signal data set, wherein the intermediate residue data set comprises an alpha shape or a Delaunay triangle mesh data object of the signal data set;
generating, by the processor, one or more geometric features and/or dynamical properties of the alpha shape or the Delaunay triangle mesh, wherein the geometric features and/or the dynamical properties include at least one of:
a 3D volume value,
a void value,
a surface area value,
a principal curvature direction value, and
a Betti number value,
of the alpha shape or the Delaunay triangle mesh,
determining, by the processor, presence or non-presence of abnormal arterial characteristics, wherein the determination comprises an application of the generated one or more geometric features and/or dynamical properties of the alpha shape or the Delaunay triangle mesh in a machine learning operation; and
storing, by the processor, in a computer readable medium, an indication of the determined presence or non-presence of ischemic myocardium, wherein the indication is subsequently retrieved and used to diagnose the one or more pathologies of the subject.
22. The system of claim 21 , wherein the intermediate residue data set is generated by:
generating, by the processor, a model of the signal data set;
determining, by the processor, a residue data object between the model and the signal data set, wherein the alpha shape or a Delaunay triangle mesh data object is generated from the residue data object of the signal data set.
23. The system of claim 21 , wherein the signal data set includes cardiac data acquired via two or more channels, each of a biopotential signal acquired over a plurality of heart beats as the long cardiac gradient record.
24. The system of claim 23 , wherein the signal data set is acquired for at least 210 seconds, and wherein a portion of the long cardiac gradient record is modeled to generate the residue data object.
25. The system of claim 21 , wherein the signal data set comprises a plurality of wide-band gradient signals simultaneously captured via a plurality of channels, including the two or more channels, and wherein a temporal lag or skew between each of the channels is less than about 10 femtoseconds.
26. The system of claim 25 , wherein each of the plurality of wide-band gradient signals is unfiltered prior to, and during, the transformation operation, to generate the signal data set.
27. The system of claim 25 , wherein each of the plurality of wide-band gradient signals comprises cardiac data that in a frequency domain have frequency components greater than about 1 kHz.
28. The system of claim 25 , wherein each of the plurality of wide-band gradient signals
comprises cardiac data that in a frequency domain have frequency components of at least 8 kHz.
29. The system of claim 21 , wherein the one or more sensors comprise at least one of a surface electrode, an intracardiac electrode, and a non-contact electrode.
30. The system of claim 21 , wherein the machine learning operation comprises an artificial neural network algorithm or a regression random forest algorithm.
31. The system of claim 21 , wherein the measuring equipment is configured to transmit the signal data set over a network to the analysis system.
32. The system of claim 31 , further comprising:
a storage area network, wherein the measuring equipment is configured to transmit the signal data set over the network to the storage area network, and wherein the analysis system is configured to retrieve the acquired signal data set from the storage area network for analysis.
33. A method comprising:
retrieving, by a processor, from a storage area network, an electrophysiological signal data set that (i) is associated with a human subject and (ii) has been acquired by measuring equipment configured to non-invasively capture, via one or more sensors, the signal data set as a long cardiac gradient record that exhibits complex non-linear variability; and
detecting one or more cardiac pathologies of the subject using the signal data set, wherein the detection operation comprises:
generating, by the processor executing a transformation operation, an intermediate residue data set from the signal data set, wherein the intermediate residue data set comprises an alpha shape or a Delaunay triangle mesh data object of the signal data set;
generating, by the processor, one or more geometric features and/or the dynamical properties of the alpha shape or the Delaunay triangle mesh, wherein the geometric features and/or the dynamical properties include at least one of:
a 3D volume value,
a void value,
a surface area value,
a principal curvature direction value and
a Betti number value,
of the alpha shape or the Delaunay triangle mesh,
determining, by the processor, presence or non-presence of abnormal arterial characteristics, wherein the determination comprises an application of the generated one or more geometric features and/or dynamical properties of the alpha shape or the Delaunay triangle mesh in a machine learning operation; and
storing, by the processor, in a computer readable medium, an indication of the determined presence or non-presence of ischemic myocardium, wherein the indication is subsequently retrieved and used to diagnose the one or more pathologies of the subject.
34. The method of claim 33 , wherein the wherein the intermediate residue data set is generated by:
generating, by the processor, a model of the signal data set;
determining, by the processor, a residue data object between the model and the signal data set, wherein the alpha shape or a Delaunay triangle mesh data object is generated from the residue data object of the signal data set.
35. The method of claim 33 , wherein the signal data set includes cardiac data acquired via at least one channel, each of a biopotential signal acquired over a plurality of heart beats as the long cardiac gradient record.
36. The method of claim 35 , wherein the electrophysiological signal data set are acquired for at least 210 seconds, and wherein a portion of the long cardiac gradient record are modeled to generate the residue data object.
37. The method of claim 33 , wherein the signal data set comprises a plurality of wide-band gradient signals simultaneously captured via a plurality of channels, including the at least one channel, and wherein a temporal lag or skew between each of the channels is less than about 10 femtoseconds.
38. The method of claim 37 , wherein each of the plurality of wide-band gradient signals is unfiltered prior to, and during, the transformation operation, to generate the signal data set.
39. The method of claim 37 , wherein each of the plurality of wide-band gradient signals comprises cardiac data that in a frequency domain have frequency components greater than about 1 kHz.
40. The method of claim 37 , wherein each of the plurality of wide-band gradient signals comprises cardiac data that in a frequency domain have frequency components of at least 8 kHz.