Method and system to assess pulmonary hypertension using phase space tomography and machine learning
Phase space tomography methods and systems to facilitate the analysis and evaluation of complex, quasi-periodic system by generating computed phase-space tomographic images and mathematical features as a representation of the dynamics of the quasi-periodic cardiac systems. The computed phase-space tomographic images can be presented to a physician to assist in the assessment of presence or non-presence of disease. In some implementations, the phase space tomographic images are used as input to a trained neural network classifier configured to assess for presence or non-presence of pulmonary hypertension, including pulmonary arterial hypertension.
1. A method for non-invasively assessing presence or non-presence of pulmonary hypertension, the method comprising:
obtaining, by one or more processors, acquired data from a measurement of one more biophysical signals of a subject, wherein the acquired data is derived from measurements acquired via noninvasive equipment configured to measure properties of the heart; and
generating, by the one or more processors, a phase space model based on the acquired data, wherein the phase space model comprises a plurality of faces and a plurality of vertices generated from a topology analysis of at least one of a residue analysis and a noise analysis performed on the acquired data, wherein the topology analysis includes at least one of a volume metric, a number of distinct bodies, and/or a maximal color variation of the phase space model; and
generating, by the one or more processors, via machine learning operations, predictors linking the phase space model across a population of patients representing both positive and negative cases to detect the presence of myocardial tissue associated with pulmonary hypertension,
wherein the predictors are presented for an assessment of presence and/or non-presence of pulmonary hypertension.
2. The method of claim 1 , comprising:
determining, by the one or more processors, a machine-trained assessment of presence and/or non-presence of pulmonary hypertension using a trained neural network-based nonlinear classifier.
3. The method of claim 1 , further comprising:
generating, by the one or more processors, a set of tomographic images derived from a phase space model generated based on the acquired data, wherein at least one of the phase space model comprises a plurality of faces and a plurality of vertices, wherein the plurality of vertices are defined, in part, by fractional subspace derivative operations of low-energy subspace parameters generated directly or indirectly from the acquired data;
generating a contour data set for each tomographic image of the set of tomographic images, wherein the contour data set is presented for the assessment of presence and/or non-presence of pulmonary hypertension.
4. The method of claim 3 , wherein the contour data set is generated by:
sweeping, via the one or more processors, a moving window associated with a trained neural network-based nonlinear classifier on a pixel by pixel basis over, at least a portion of, a given tomographic image; and
combining, for a given pixel of the tomographic image, outputs of the swept moving window.
5. The method of claim 3 , comprising:
presenting, via a display of a remote computing system, the generated contour data set.
6. The method of claim 3 , comprising:
presenting, via a display of a remote computing system, the generated contour data set and a corresponding tomographic image used to generate the contour data set, wherein the generated contour data set is rendered as an overlay over the corresponding tomographic image.
7. The method of claim 3 , wherein the generated phase space model comprises a three-dimensional object defined by the plurality of faces and a plurality of vertices.
8. The method of claim 7 , wherein the plurality of vertices are generated as a point cloud in 3D space, wherein each point in the point cloud is associated with a fractional order of a fractional subspace derivative operation of the low-energy subspace parameters.
9. The method of claim 7 , where at least one of one or more color attribute parameters associated with a variance of a modeled channel signal is generated from a model-derived construction of the acquired data subtracted from a baseline-removed raw channel of the acquired data.
10. The method of claim 1 further comprising:
removing, by the one or more processors, a baseline wandering trend from the acquired data prior to generating the one or more phase space models.
11. The method of claim 3 further comprising:
performing a model-derive reconstruction operation of the acquired data to generate the low-energy subspace parameters, the low-energy subspace parameters comprising a plurality of basis functions and coefficients.
12. The method of claim 3 , wherein the low-energy subspace parameters consist of low-energy subsets of a plurality of basis functions and coefficients.
13. The method of claim 3 further comprising:
causing, by the one or more processors, generation of a visualization of generated phase space volumetric object as a three-dimensional object, wherein the three-dimensional object is rendered and displayed at a display of a computing device or report.
14. The method of claim 1 , wherein the acquired data comprises differential channel signals.
15. The method of claim 1 , wherein the assessment of presence and/or non-presence of pulmonary hypertension is selected from the group consisting of: an assessment of pulmonary arterial hypertension; an assessment of PH due to left heart disease; an assessment of PH relating to lung disease or hypoxia; an assessment of pulmonary hypertension includes chronic thromboembolic pulmonary hypertension; and an assessment of a rare disorder that leads to PH.
16. The method of claim 1 , wherein the acquired data comprise cardiac frequency information at a frequency selected from the group consisting of about 1 kHz, about 2 kHz, about 3 kHz, about 4 kHz, about 5 kHz, about 6 kHz, about 7 kHz, about 8 kHz, about 9 kHz, about 10 kHz, and greater than 10 kHz.
17. A system comprising:
a processor; and
a memory having instructions thereon, wherein the instructions when executed by the processor causes the processor to:
obtain acquired data from a measurement of one more biophysical signals of a subject, wherein the acquired data is derived from measurements acquired via noninvasive equipment configured to measure properties of the heart; and
generate a phase space model based on the acquired data, wherein the phase space model comprises a plurality of faces and a plurality of vertices generated from a topology analysis of at least one of a residue analysis and a noise analysis performed on the acquired data, wherein the topology analysis includes at least one of a volume metric, a number of distinct bodies, and/or a maximal color variation of the phase space model; and
generate, via machine learning operations, predictors linking the phase space model across a population of patients representing both positive and negative cases to detect the presence of myocardial tissue associated with pulmonary hypertension,
wherein the predictors are presented for an assessment of presence and/or non-presence of pulmonary hypertension.
18. A non-transitory computer readable medium having instructions stored thereon, wherein execution of the instructions by a processor causes the processor to:
obtain acquired data from a measurement of one more biophysical signals of a subject, wherein the acquired data is derived from measurements acquired via noninvasive equipment configured to measure properties of the heart; and
generate a phase space model based on the acquired data, wherein the phase space model comprises a plurality of faces and a plurality of vertices generated from a topology analysis of at least one of a residue analysis and a noise analysis performed on the acquired data, wherein the topology analysis includes at least one of a volume metric, a number of distinct bodies, and/or a maximal color variation of the phase space model; and
generate, via machine learning operations, predictors linking the phase space model across a population of patients representing both positive and negative cases to detect the presence of myocardial tissue associated with pulmonary hypertension,
wherein the predictors are presented for an assessment of presence and/or non-presence of pulmonary hypertension.