FROM A PIEZOELECTRIC DEVICE AND DETECTING STENOSIS

[0001] CROSS-REFERENCE TO RELATED APPLICATION

[0002] This application claims the benefit of U.S. Provisional Application Serial No. 62/350,576, filed June 15, 2016, the disclosure content of which is hereby incorporated by reference in its entirety.

[0003] FIELD OF INVENTION

[0004] The present application is generally related to devices and methods for reducing noise within a data sample, to allow for determination of stenosis of an artery in the arterial circulatory system, through mechanical, hardware, and software strategies.

[0005] BACKGROUND

[0006] Stroke is the major cause of adult neurological disability in the world. About eighty percent of all strokes occur from vessel blockage. Stroke is an enormous health burden on society. Ischemic Stroke is the most common cause of disability in adults and the third leading cause of mortality in developed countries. Around the world, stroke causes nine percent of all deaths (1 in 11) and is the second leading cause of death. According to the World Health

Organization fifteen million people suffer stroke annually. Of these five million die and another five million are permanently disabled. In the United States stroke is the fourth leading cause of death affecting eight hundred thousand people annually. Ischemic stroke, occurring due to insufficient blood supply to the brain, accounts for the largest number of strokes (88%), followed by intracerebral hemorrhage (9%) and subarachnoid hemorrhage (3%).

[0007] The primary cause of Ischemic stroke is atherosclerosis which is a long-term inflammatory disease beginning at the abluminal surface and eventually causes endothelial abnormalities. The thickening and hardening of the vessel wall eventually produces Atheros lotic plaques which are essentially composed of lipid fibrous tissue and inflammatory cells.

Progression of the plaque can lead to a narrowing of the lumen, i.e., stenosis. (The percentages of stenosis that will be quoted herein are by the NASCET standard of measuring stenosis). The superficial location of the carotids allows non-invasive methods to be used in detecting abnormal blood flow within them. Computational simulations and experimental flow visualizations both demonstrate marked differences in flow patterns distal to concentric and eccentric stenosis for moderate and high stenosis cases. This is one example of an important parameter for blood flow characteristics which is dependent upon more than just the degree of stenosis.

[0008] Roughly half of all strokes are caused by atherothromboembolism and most of these are extracranial atheromatous lesions, most often involving narrowing of the internal carotid arteries (ICAs). Symptomatic patients with severe stenosis (70-99%) benefit from carotid endarterectomy. It has been suggested that endarterectomy could also reduce the risk of stroke from moderate (50-69%) stenosis, therefore imaging of the carotid artery is indicated in patients with symptoms of cerebral ischemia.

[0009] Thankfully, modern surgical techniques have allowed advancement in treatment of stroke, and deaths from stroke have declined dramatically in the US. Stroke is now listed as the fourth leading cause of death rather than the third leading cause, because more people are dying from lung cancer than from stroke. The American Stroke Association commissioned a panel of doctors (a "Stroke Council"), chosen on the basis of recent work in their respective fields of expertise, to assess what factors have been influencing the decline in stroke mortality. This Council issued its conclusions as "A statement from the American Heart Association/ American Stroke Association" in 2008. The report was based upon systematic literature reviews, published clinical and epidemiological studies, morbidity and mortality reports, clinical and public health guidelines, authoritative statements, personal files, and expert opinion to summarize evidence. The document underwent extensive American Heart Association internal peer review, Stroke Council leadership review, and Scientific Statements Oversight Committee review before consideration and approval by the American Heart Association Science Advisory and

Coordinating Committee. The review declares that "The decline of stroke mortality over the past decades represents a major improvement in population health that is observed for both sexes and all racial/ethnic and age groups. The major decline in stroke mortality represents a reduction in years of potential lives lost."

[0010] The remarkable decline in stroke mortality was acknowledged as one of the ten great public health achievements in the twentieth century. This decline has continued over the past decade (2001 to 2010) and dropping stroke mortality was again identified as one of the ten great public health achievements of the first decade of the twenty-first century. The Stroke Council report states that stroke mortality in the U.S. has been falling faster than ischemic heart disease mortality for several decades now. Medications for blood pressure control have had a larger and more immediate impact on stroke than on heart disease. Public health officials consider the lowering of blood pressure and hypertension control as the major contributors to the decline of stroke.

[0011] Also mentioned as contributing to the decline of stroke have been smoking cessation programs, improved control of diabetes and of abnormal cholesterol levels, and better as well as faster treatment. The Stroke Council concluded that efforts in hypertension control initiated in the 1970's were the most substantial influence on the decline in stroke mortality. An interesting aspect of this extensive report is that Doppler ultrasound ("DUS") is not mentioned specifically, despite all of its improvements over the decades. This dovetails well with the fact that, even though DUS is widely utilized to estimate stenosis in the carotid artery, DUS lacks precision in that there is an inability to distinguish between some of the various sub-classifications of stenosis from each other, and generally, the DUS devices provide results with error bars which cross over entire decimal percentage subdivisions. As another example of this, DUS is not able to distinguish nearly as well the 50-69% "moderate" stenosis level as compared to other levels of stenosis.

[0012] Despite the recent gains in stroke treatment, there remains a massive hole in early detection and treatment of patients before, not after, they have experienced stroke. Any stroke, even small, frequently leads to a rapid reduction in quality of life and this morbidity is especially troublesome as improved devices and scanning of patients could remove and avoid many stroke occurrences, especially to patients that are generally deemed at a moderate or low risk.

[0013] Coronary disease, like that of the carotid artery, is also a significant risk. While age- adjusted mortality rates for coronary artery disease have declined steadily in the United States since the 1960's, heart disease remains the leading cause of death in the United States, with more than 600,000 deaths attributed to heart disease.

[0014] Heart attacks, like stroke, are often treated with certain blood thinning medications or invasive surgeries. Those who receive treatment before a major heart attack, enjoy longer life and reduced morbidity. Unfortunately, those who suffer heart attacks face arduous recovery, morbidity, or death, which may be preventable in many cases with early detection and treatment.

[0015] Accordingly, there is a need for treatment of these diseases. Described herein are devices, systems, and methods that can be utilized to detect blockage and to filter and eliminate noise from the detection systems to enable determination of blockage in these circulatory systems.

[0016] SUMMARY OF THE INVENTION

[0017] Accordingly, embodiments of the invention relate to methods and devices for reducing noise from a medical apparatus for measuring distinct sounds created by the flow of blood in diseased blood vessels that can then determine the percent of occlusion in such vessels and serves as a predictor for heart disease, stroke, and other arterial diseases. The devices primarily focus on active noise cancellation strategies, and passive noise attenuation strategies, while the methods focus on certain data analysis and filtering strategies to remove noise from the data.

[0018] A Y shaped array made of two components, an inner array and an outer array, comprising three openings, one at each of the end of the Y branches; said array made of a sound attenuating material, sufficient to reduce the ambient noises generated by the movement of the array; configured to said Y shaped array are three sensors, one positioned in each of the three openings; a diaphragm bellows membrane having a ring shape, an outer flange at the outer circumference, and an inner flange on the inner circumference of said ring; said outer circumference compressed between said inner and outer array in each opening; a sensor base configured having a locking groove to accept the inner flange between said a base housing and a locking cap; and a processing board; configured to said base housing is a disposable sensor assembly comprising a piezo sensor mounted onto a flange of a piezo cap, and comprising attachment means between said piezo cap and said housing.

[0019] A C-shaped yoke having a track like feature capable of securing to said track-like feature two or more sensor pods, wherein said sensor pods are secured via a track opening in the base of each of said sensor pod.

[0020] An array comprising an array body, and three sensor pods; said array body comprising an inner array half and an outer array half, each inner and outer half comprising two arms and a neck; and three openings defined at each end of the arms and neck; said openings defined to accept a diaphragm bellows membrane, wherein said diaphragm bellows membrane comprises an outer flange to be accepted between said inner array half and outer array half; and a disposable sensor pod comprising a disposable piezo assembly and a sensor base, said disposable piezo assembly comprising: a circular piezo cap comprising a top and a bottom an inner face and an outer face, with an opening between the top and bottom with the opening larger at the top than the opening at the bottom; a flange positioned on the inner face of the opening; a piezo having a top a bottom and a perimeter support; said piezo disposed of within said opening, with the bottom of the perimeter support engaged to an adhered to said flange; a Printed Circuit Board having a ring shape and an outer diameter to fit within the opening and engaged to the bottom of said flange; and on said inner face one-half of an attachment means for securing said disposably piezo assembly to said sensor base; and said sensor base comprising a diaphragm bellows membrane a printed circuit board housing, a printed circuit board, and a cap; said diaphragm bellows membrane being a ring shape having an outer flange on an outer circumference of said ring, and an inner flange on an inner circumference of said ring; said outer flange engaging between said inner array half and said outer array half in each of said three openings, and said inner flange engaging between said cap and said printed circuit board housing; said printed circuit board housing comprising a bell shape, having a narrow bottom and a wide top, with an opening between the top and bottom, a locking groove on said narrow bottom to engage said inner flange; and an attachment means a the top of the top; said printed circuit board fitting within said opening.

[0021] A passive noise attenuating sensor pod comprising a disposable piezo assembly and a sensor base, said disposable piezo assembly comprising: a circular piezo cap comprising a top and a bottom an inner face and an outer face, with an opening between the top and bottom with the opening larger at the top than the opening at the bottom; a flange positioned on the inner face of the opening; a piezo having a top a bottom and a perimeter support, and a noise attenuating barrier positioned around the top of the opening of the circular piezo cap, creating a second seal around a surface for detecting stenosis; said piezo disposed of within said opening, with the bottom of the perimeter support engaged to an adhered to said flange; a Printed Circuit Board having a ring shape and an outer diameter to fit within the opening and engaged to the bottom of said flange; and on said inner face one-half of an attachment means for securing said disposably piezo assembly to said sensor base; and said sensor base comprising a diaphragm bellows membrane a printed circuit board housing, a printed circuit board, and a cap; said diaphragm bellows membrane being a ring shape having an outer flange on an outer circumference of said ring, and an inner flange on an inner circumference of said ring; said outer flange engaging between said inner array half and said outer array half in each of said three openings, and said inner flange engaging between said cap and said printed circuit board housing; said printed circuit board housing comprising a bell shape, having a narrow bottom and a wide top, with an opening between the top and bottom, a locking groove on said narrow bottom to engage said inner flange; and an attachment means a the top of the top; said printed circuit board fitting within said opening.

[0022] An active noise cancelling method comprising; a first sensor placed adjacent to a skin surface and second sensor disposed of away from said skin surface; detecting sounds, simultaneously in said first and second sensor; processing said sounds from analog to digital and subtracting said digital sounds from said second sensor from said first sensor.

[0023] An active noise cancelling method comprising: a first sensor placed adjacent to a skin surface and a second sensor disposed of away from said skin surface; detecting sounds simultaneously in said first and second sensors; processing the sounds received in said second sensor and phase shifting said sounds by 180 degrees and emitting a proportional phase shifted sound.

[0024] A method of de-noising data collected from a sensor comprising receiving analog data from a first sensor; amplifying said analog data; converting the analog data to digital;

performing a wavelet analysis through removal of sounds in the range of l-70Hz.

[0025] A method of de-noising data collected from a sensor comprising receiving analog data from a first sensor; amplifying said analog data; converting the analog data to digital;

performing a wavelet analysis through removal of sounds in the range of l-70Hz; performing a method selected from the group consisting of Burg's Method, Welch's method, or combinations thereof, and generating a Power Spectral Density.

[0026] A method of reducing noise received at a sensor comprising: placing a sensor adjacent to the skin surface of a patient; wherein said sensor comprises a disposable piezo assembly and a sensor base, said disposable piezo assembly comprising: a circular piezo cap comprising a top and a bottom an inner face and an outer face, with an opening between the top and bottom with the opening larger at the top than the opening at the bottom; a flange positioned on the inner face of the opening; a piezo having a top a bottom and a perimeter support, and a noise attenuating barrier positioned around the top of the opening of the circular piezo cap, creating a second seal around a surface for detecting stenosis; said piezo disposed of within said opening, with the bottom of the perimeter support engaged to an adhered to said flange; a Printed Circuit Board having a ring shape and an outer diameter to fit within the opening and engaged to the bottom of said flange; and on said inner face one-half of an attachment means for securing said disposably piezo assembly to said sensor base; and said sensor base comprising a diaphragm bellows membrane a printed circuit board housing, a printed circuit board, and a cap; said diaphragm bellows membrane being a ring shape having an outer flange on an outer

circumference of said ring, and an inner flange on an inner circumference of said ring; said outer flange engaging between said inner array half and said outer array half in each of said three openings, and said inner flange engaging between said cap and said printed circuit board housing; said printed circuit board housing comprising a bell shape, having a narrow bottom and a wide top, with an opening between the top and bottom, a locking groove on said narrow bottom to engage said inner flange; and an attachment means a the top of the top; said printed circuit board fitting within said opening; placing a second sensor away from said skin surface; detecting sounds, simultaneously in said first and second sensor; processing said sounds from analog to digital and subtracting said digital sounds from said second sensor from said first sensor.

[0027] In a further embodiment, taking a method of reducing noise at a sensor from above and further by receiving analog data from a first sensor; amplifying said analog data; converting the analog data to digital; performing a wavelet analysis through removal of sounds in the range of l-70Hz. In a further embodiment, further subjecting the data to Burg's method, Welch's method or both.

[0028] A further embodiment is directed to A method for eliminating noise from a data sample comprising passive noise cancellation, active noise cancellation, and a software based filtering process; said passive noise cancellation comprises collecting data from a piezo sensor that is sound isolated by a noise attenuating material surrounding said piezo sensor and forming a connection to the surface to be sampled; isolating said piezo sensor on a device comprising a membrane; said active noise cancellation comprises utilizing a second sensor adjacent to said piezo sensor to detect ambient sounds and subtracting said ambient sounds detected from said second sensor from said data; performing a wavelet analysis on said data; and performing a method selected from the group consisting Burg's method, Welch's method, and combinations thereof. The method wherein said membrane is a diaphragm bellows membrane. The method wherein said diaphragm bellows membrane is ring shaped having an outer circumference and an inner circumference; and an outer flange on the outer circumference and an inner flange on said inner circumference. The method wherein said inner flange is connected to a sensor pod comprising said piezo sensor.

[0029] Methods of determining stenosis include a new data adaptive filter based on wavelets that improves the ability of determining specific sounds measured by piezoelectric units by filtering out the unwanted sound frequencies such as the background noise in the input signal. The process of removing the background noise in the input signal is very complicated and challenging. Sources of the noise are many. Some can be prevented by our highly engineered and sensitive sensor. Others are unavoidable such as human voices or the ambient sounds in the room where the recording was taken. This type of noise is stationary but it is more challenging to detect sounds that are non- stationary such as the patient movement, or unexpected interruptions related to breathing, sneezing, or coughing. So many methods have been explored and the wavelets remain an effective tool for filtering out the unwanted sound frequencies, and after analyzing thousands of samples of the human artery sound data, our objective has been achieved by identifying a class of the wavelets that works very effectively to de-noise the signal for the next procedure which is based on Fast Fourier Transform to extract the desired sound spectrum for quantifying the degrees or percent of partially occluded arteries.

[0030] Brief Description of the Figures

[0031] FIG. 1 depicts example sensor pads.

[0032] FIG. 2 depicts a passive cancellation device with "over-the-ear" like construction, to block ambient noise from the sensor.

[0033] FIG. 3 depicts an exploded view of a piezoelectric sensor which is attached to a diaphragm bellows membrane to isolate it from ambient noise.

[0034] FIG. 4A depicts an electronic view of subtracting ambient noise from a received signal.

[0035] FIG. 4B depicts a flow-chart of subtraction of ambient noise from a signal.

[0036] FIG. 4C depicts a flow-chart of an active noise cancellation process.

[0037] FIG. 5A depicting a double piezo assembly.

[0038] FIG. 5B depicting an alternative double piezo assembly in an exploded view.

[0039] FIG. 5C depicting a parallel piezo assembly.

[0040] FIG. 5D depicting a microphone on an array.

[0041] FIG. 5E depicting a microphone on a base. [0042] FIG. 5F depicting a microphone on a cart.

[0043] FIG. 6 depicts a sensor pod assembly with sound attenuating materials.

[0044] FIGS. 7 A, 7B, and 7C depicts a flow-chart data collection, with 7 A wired, 7B wireless from a single module, and 7C wireless from multiple modules.

[0045] FIG. 8 depicts a flow-chart of data processing for stenosis determination

[0046] FIG. 9 depicts a chart showing a frequency chart.

[0047] FIG. 10 depicts certain raw data from three channels.

[0048] FIG. 11 depicts a ten second channel plot.

[0049] FIG. 12 depicts a PSD periodogram

[0050] FIG. 13 depicts Welch's Power Spectral Density estimate.

[0051] FIG. 14 depicts additional data plot of Welch's method.

[0052] FIG. 15 depicts Burg's method of smoothing.

[0053] FIG. 16 depicts Reflection Coefficients.

[0054] FIG. 17 depicts a PSD before denoising.

[0055] FIG. 18 depicts a PSD before denoising.

[0056] FIG. 19 depicts a PSD before denoising.

[0057] FIG. 20 depicts a Burg's Power Spectral Density Estimate.

[0058] FIG. 21 depicts a Parametric PSD after denoising, depicting peaks.

[0059] FIG. 22 depicts a perturbation representative in an artery.

[0060] FIG. 23 depicts a vortex ring flow.

[0061] FIG. 24 depicts flow in a circular tube with vortex jets.

[0062] Detailed Description of the Preferred Embodiment

[0063] The word "about" is intended to encompass a range of values +/- 10% of a specified numerical value.

[0064] All references cited herein are hereby incorporated by reference in their entirety.

[0065] Current protocols in medicine regard DUS as the preferred non-invasive imaging method for initial testing of carotid artery stenosis. However, DUS, as well as magnetic resonance angiography ("MRA"), is concerned with measuring flow, not vessel diameter directly. A chart or the equivalent is needed to transform from flow to diameter stenosis, such as Spencer's curve. This is because the conversion from flow to diameter varies significantly as flow slows at the upper end of the degree of stenosis diagnosis, a reversal of the relation of flow to diameter at more moderate or mild degrees of stenosis. Unfortunately, these curves prove to lack precision and the estimates they provide for stenosis, are just that, estimates, and these estimates are frequently wrong. Accordingly, use of DUS or MRA is not sufficient for stenosis determination and requires a secondary evaluation, often by invasive procedures, where high levels of stenosis are suspected. Even worse, there are incorrect predictions for hundreds of patients that have little stenosis, but are nonetheless subjected to invasive procedures. Finally, there are of course errors in failing to accurately determine high levels of stenosis and patients at risk. Accordingly, current protocols have both false positive and false negative errors.

[0066] Herein, we describe a device and methods for detecting stenosis with a non-invasive device that measures specific sounds emanating from the arterial circulatory system and provides the ability to calculate stenosis therefore. For example, the device utilizes at least one highly sensitive listening device, which uses a piezoelectric component, placed adjacent the carotid artery. The piezoelectric device detects nearly any energy being transmitted from within the confines of the space, of course, including those emanating from the carotid artery itself. Indeed, a particular solution that needed to be solved for the device to function is the elimination of the ambient noise that is detected by the body, the electrical features in a room, the patient, other persons in the room, and other ambient noise.

[0067] Applicant's device is directly sensitive towards coherent flow structures called large asymmetric ring vortices which are directly related to plaque build-up; therefore signs of flow having a direct correlation with stroke. Because of the sensitivity of the device and the methods described herein, the correlation and determination of stenosis is operator independent. This is in direct contrast to prior art devices. For example, DUS as well as other non-invasive

measurements of degree of stenosis require human judgment, with variability present among identical tests, as reviewed and determined by different persons.

[0068] Large ring vortex motion in the post stenotic region, generates sound of much higher intensity, at a low acoustic frequency than that produced by smaller eddies and other forms of turbulence within the artery. The large ring vortices constitute a coherent disturbance causing oscillations at the artery wall of discrete frequencies due to circumferential velocities within the ring, which are perpendicular to the axially directed velocities. These ring vortices only appear in the blood flow in the post stenotic. This occurs in the carotids during the systolic portion of the heart cycle. There is a spread or broadening of frequencies surrounding the discrete ones in a nearly bell curve shape in the intensity signal. Oscillations in blood motions that are circumferential as well as some of the intensity of radial oscillations which are perpendicular to the wall are associated with ring vortex motions. These, in turn, create pressure variations at the wall, which produce sound at the arterial surface which can then be picked up by a microphone at the skin surface.

[0069] A key issue in hearing the low intensity sounds is utilizing a device that is sensitive enough to accurately detect sounds with a low amplitude of these sounds. As described herein, in order to de-noise data received from such a device, one or more of the following protocols are required. Each protocol can be used alone, or in combination with others, so as to efficiently and accurately determine stenosis in the arterial circulatory system. Only after numerous iterations were we able to make a device having the necessary features to detect the sounds we were seeking and to block and remove sounds unrelated to the stenosis which we are measuring.

[0070] Accordingly, after reviewing several options for highly efficient listening devices, we settled upon a piezoelectric sensor ("piezo"), which is able to detect very small vibrations and sounds that occur in the neck which cannot be detected or interpreted even with the use of a stethoscope. Indeed, the sounds that we are detecting are between about 20 and 3000 Hz and have a low amplitude based on the low sound generated by the large ring vortices being detected. Of course, a device having such sensitivity to sound and vibrations presents numerous issues, in that it detects literally any sound or vibration associated with ordinary life. For example, in the process of placing the device on a patient, ambient sounds that become noise include a sound of fans, footsteps, breathing, computer sounds, phone rings, and vibrations from phones. Certainly, all other ambient noises at a doctor's office will also be detected. However, even the heartbeat, movement of the patient, movement of the technician holding the sensing device, deep breaths, sighs, etc. likely produce enough sound or vibrations to be detected by the device. Thus, during a test, using the piezo device, numerous sounds, not only the sounds of interest, but ambient sounds are detected and must be removed from the data. Accordingly, the data goes through numerous different filtering processes to remove the various background noises before analysis can be performed on the, now de-noised data, so that stenosis can be predicted.

[0071] The noises that we are particularly measuring are these subtle vortexes. These vortexes are created as wall pressure fluctuations distal to a constriction (stenosis) in rigid or elastic pipes, or in arteries, reveal the presence of low-frequency maxima. These fluctuations are found to be associated with large-scale, medium-scale, or small-scale vortices (also called "eddies" if small), that are strong in the region distal to the constriction (called "stenosis" when in an artery).

[0072] Normal blood flow in a heathy patient causes certain sounds which are detectable by our device. Patients which have stenosis in the carotid arteries will often have another 2 or 3 additional sounds that can be picked up by our device. Depending on the amount of stenosis and how many stenosed areas the sound will change. The carotid artery has a branch which feeds two main areas in the head. One main branch going to the brain and the other branch going to the face. The area that we test for is where the carotid artery branches into these two areas. Thus depending if there is stenosis in one branch or two can lead to multiple sounds being picked up. Because these sounds/vibrations are at such a low level it is vital to make sure as much external noise is eliminated as possible. Even small noises in the 20-3000 Hz range can overwhelm the noises we are looking for making noise elimination critical.

[0073] With regard to flow and the noises created therein, some of the fluid-flow energy enters into the vortex motions distal to a constriction, which then results in an increase in the wall pressure amplitude, above that of turbulence alone, at the lower frequency end of the wall pressure power spectrum. These maxima are nearly Gaussian-shaped bell curves situated atop a broad, nearly flat spectrum at low frequencies that is due to turbulence within the pipe or artery. The maxima are always found at lower frequencies than the so-called "break" frequency characteristic of the turbulence spectrum where the latter changes quite abruptly from nearly flat to steep declining in intensity (when the logarithm of signal intensity is plotted versus a logarithmic frequency scale).

[0074] Interestingly, measuring these maxima and plotting the power spectrum provides for a visual image of stenosis in an artery. Indeed, we have determined that by plotting the power spectrum on the y axis and amplitude on the x-axis, we can effectively determine the percentage of stenosis in the carotid arteries of a patient.

[0075] These maxima (generally two in number) are the main features in the frequency power spectrum at low frequencies generated by the wall pressure fluctuations when there is a constriction as compared to the situation of no constriction yet fully developed turbulence. [0076] In order to analyze this data, we have developed devices and invented several methodologies and processes that reduce or eliminate extraneous noise from the data samples, to enable further spectrum analysis downfield.

[0077] The device eliminates noise in several ways. One by using sound barriers/dampening material to eliminate external noise as much as possible as well as noise caused by the patient moving; i.e. passive noise cancelling. We also eliminate or cancel ambient noise with active noise cancelling strategies, whether generating opposing waves or subtracting ambient noise; finally, we de-noise the received data by methodologies related to data processing using

Wavelet, Welch's and Burg's methods. Ultimately, we plot peaks on a PSD and calculate stenosis of an area of interest in the arterial circulatory system through comparing these peaks on the PSD.

[0078] Passive noise cancelling strategies and methodologies

[0079] A first set of strategies includes mechanical strategies to eliminate or reduce noise. We can also consider these strategies to be passive noise cancelling strategies.

[0080] For example, in preferred embodiments, the yoke 5, as depicted in FIG. 5D is made of a plastic or a polymer. Construction of a yoke with as few components as possible is intended, as additional components create joints that may cause ambient noise to the system. We typically use unibody constructed devices, molded into a form, or devices having an inner and outer portion, thereby allowing some materials to be compressed within said device, and for insertion of wires, batteries, processors, memory, and the like, into the array. In embodiments where multi-body construction is used, it is preferable that mechanisms are in place to ensure proper stability and to prevent unnecessary vibrations and sound due to the construction. This can be achieved through appropriate materials and fixing mechanisms, including the use of dampening materials when connecting two or more components together on the yoke 1. The yoke 5 may further optionally include sound cancelling materials disposed of in or on the yoke 5. This provides that movement of the yoke 5 or of the patient while the yoke 5 is on the patient, will prevent unnecessary noises that may disrupt the sound received by the piezos.

[0081] FIGS. 1 and 2 depict disposable sensor pads 18. These pads 18 serve as the first line of active noise canceling, where the pads 18 have a durometer and shape to allow for secure contact with the skin of a patient, which blocks some ambient noise from entry to the piezo sensor 90. The sensor pad 18 is placed on the piezo 90 and positioned such that a flat side of the pad is in contact with the piezo 90 and the obverse side is in contact with the skin of the patient.

Particular designs, such as those in FIG. 1 are angled on the skin facing side to create a good seal against the skin. The sensor pads are angled at the skin facing surface, such that on the left hand side, the curvature on the bottom right engages to an angled structure to ensure a good acoustic fit. By contrast, the sensor pad on the right hand side of the page comprises a dual concave structure, to fit around a structure that is rounded. In each case, there is a proper fit, and so the sensor pod must be able to rotate to allow the sensor to be properly fit against the skin to achieve a proper acoustic contact for data collection. Cross-sectional views of the left and right sensor pads are depicted for clarity. The sensor pads 18 further direct sound and vibrations from the patient's skin to the piezo and results in sound and data that eliminates some noise from the signal.

[0082] In further embodiments, it is advantageous to utilize gel on the skin of a patient that assists in forming a temporary seal between the pad and the skin of the patient. Certain oil and water based gels or liquids are useful in assisting with the seal.

[0083] FIG. 2 adds a further feature, which is an external noise attenuating material 219 that compresses around the sensor pad 18. The external noise attenuating material 219 is like an "over-the-ear" headphone, which blocks ambient noise from the ear. In the similar manner, the external noise attenuating material 219 surrounds the sensor pad 18 and blocks some of the ambient noise.

[0084] The sensor pod itself, therefore, must also attenuate and block out some of the ambient noise. This can be achieved through several features that are depicted in detail in FIG. 3. FIG. 3 depicts an exploded view of a sensor pod, beginning with the piezo 90 which is attached to the sensor cap 100 with an adhesive 92. The piezo 90 fits within a recess at the top of the sensor or piezo cap 100, and sits on a flange on the opening in piezo cap 100. The piezo cap 100 is made of a plastic material having a density to attenuate and reduce penetration of sound waves. Accordingly, sound will travel from a sensor pad 18 placed onto the top surface of the piezo 90, but will be limited from the bottom surface or from the side of the piezo, due to the construction of the sensor cap 100 and the remaining components. Higher density materials have greater sound attenuating properties, so appropriate density plastics can be selected around the piezo 90 to reduce ambient noises. [0085] A second adhesive 92 connects to the Printed Circuit Board 105, and several PCB contacts 106 contact the spring pins 111 on the PCB processor board 110 to make electronic connections. A processing unit 112 is defined on the bottom of the PCB processor board and comprises a battery, memory, and a processor. Alternatively, a battery may be centrally located, and the processing unit may be centrally located. The Piezo cap 100 contains a groove 101 to receive a quarter-turn locking feature 116 that is located on the PCB housing 115. This housing, like the PCB cap 100 attenuates and reduces ambient noise penetration to the piezo 90. A screw 113 secures the PCB housing 115 to a diaphragm bellows membrane 120, which allows movement of the entire sensor pod in directions in the lateral and longitudinal axis. Accordingly, when a device is placed against a surface, the sensor pod will be able to move away from the surface, or laterally to create a better fit towards the skin of the patient. Furthermore, this diaphragm bellows membrane 120, being non-rigid, will reduce the transfer of vibration and movement from a person holding a device containing the sensor pod, such as an array. A locking mechanism 121 secures the inner portion of the diaphragm bellows membrane 120 between the locking groove 117 and the locking cap 125.

[0086] Accordingly, an embodiment of the disclosure comprises passive noise cancellation strategies comprising a sensor pod (features 85 and 86 together) comprising a disposable piezo cap 85, having a piezo 90, a Piezo cap 100 having noise attenuating properties, and a PCB house assembly 86 having a PCB board 110, a diaphragm bellows membrane 120, and a PCB housing 115. A locking feature on the PCB housing 115 connects to the Piezo cap 100 to secure them together. The rear of the PCB house assembly 86 comprises a diaphragm bellows membrane 120 that allows for movement of the components to isolate them from ambient noise and vibrations. The device may further comprise a noise attenuating material 219 disposed of around the sensor pad 18 to passively waves from the piezo sensor 90.

[0087] FIG. 6 further details a sample piezo utilizing sound attenuating materials. The sensor pad 18 is positioned on the sensor 13, with attenuating materials 661, 662, 663, 664, 665, 666, 667, and 668 surrounding the sensor 13. By use of these materials, we can surround the sensor 13 with attenuating materials and reduce the ambient noise that is received at the sensor. Appropriate low and high density materials can be use, sound baffling materials and the like.

[0088] Active noise cancelling strategies and methodologies [0089] In addition to the passive noise cancelling features of the sensor pod assembly, a further strategy for reducing noise to the piezo includes active cancellation of noise. Active noise cancellation can be produced through several different strategies. A first strategy utilizes a second microphone or piezoelectric device to measure ambient noise. For example, in FIGS. 5 (A-F), different variations of this strategy are provided. An overview of these strategies is depicted in flow charts of FIGS. 4 A, 4B, and 4C

[0090] FIG. 4A depicts an electronic diagram depicting a signal received 330, ambient noise 331 and a subtraction 332, wherein the ambient noise 331 is literally removed from the received signal 330 to generate the subtracted signal 332. FIG. 4B provides a further flow-chart of this concept. For example, box 340 defines reading the analog sounds from the ambient room, converting these to digital 342, converting to a frequency domain 343. In parallel, the analog signals are received from the carotid artery 341, or another artery of the circulatory system, converted to digital 342, converted to frequency domain 343, and then the ambient room sounds are subtracted from the sounds from the artery 344. The different in sound is then converted back to time domain 345, and the data is processed 346 to calculate occlusion or stenosis of the artery being reviewed.

[0091] FIG. 4C depicts an active cancellation flow chart. A sensor reads analog sounds from and ambient room 351. Parallel sensor reads analog sounds from the carotid artery 350. Each sound is amplified to a desired volume in 352. Signal from the ambient room 351 is phase shifted 180 degrees 353, and the phase shifted sound 353 is emitted 354. Sounds are received by a microphone 355 and converted to digital signals. This effectively removes the ambient sound 351 from the digital signal processed from the carotid 350.

[0092] FIG. 5 A depicts a paired piezoelectric device, having a first piezo 90, a board 110, positioned between the first piezo 90 and a second piezo 150. The first piezo 90 would engage to a disposable pad 18 and be placed against the skin of the patient. The sounds from the patient would be detected through the disposable pad 18 and by the first piezo 90. The first piezo 90 would also pick up ambient noise, as well as noise and harmonics from power lines, in the 60Hz frequency. The purpose of the second piezo 150 is to detect these same ambient noises as the first piezo 90, but to not detect (or to detect at a much lower intensity) the sounds from the arterial circulatory system being investigated. The sounds from the second piezo 150 can then be compared to the sounds from the first piezo 90 to identify and eliminate background sounds from those from the arterial circulatory system. The subtraction process is depicted in flow-charts of FIG. 4.

[0093] FIG. 5B depicts a further exploded view of FIG. 5 A, and includes additional components. The piezo 90 engages to the piezo cap 100 with an adhesive 92 on a flange in the piezo cap 100. An adhesive 92 attaches the PCB contact board to the PCB board 110. Below the PCB board, is a second piezo 150, with is attached to the PCB board with a wiring harness 91. Both piezos can be contacted with a PCB board 105, and contact pins, as depicted in FIG. 3. The second piezo 150, being isolated by the PCB board 110 then detects ambient sounds and not the sounds from the patient.

[0094] Cancellation and subtraction of sound can be accomplished in two ways. First, the sounds from the second piezo can be inversed and literally subtracted from the first piezo.

Second, the sounds can be eliminated in analog by sending in a negative background signal which eliminates the sound. The prior art details several noise cancelling headphones, which use an external microphone to detect sound. This sound is then processed by a computing system with the device, and identifies and generates an out of phase sound, being out of phase by 180 degrees. This, when combined with the external sound, effectively cancels out the sounds that are received. Either method is functional, though the subtraction method may be preferable in certain embodiments.

[0095] FIGS. 5C, 5D, 5E, and 5F each detail a slightly different strategy for identifying ambient sounds for active cancellation. For example. FIG. 5C depicts a parallel piezo setup, comprising a base chip 26 and a first piezo 24 and a second piezo 25, arranged in parallel. This setup will allow for detection of stenosis along a linear path and determining of position of an occlusion between the two piezo sensors. This occurs as each piezo will detect the same sounds, but receive them at slightly different times. This allows for positional identification of the underlying blockage. Furthermore, one piezo may be contacted with the sensor pad 18 and a second not, thus allowing for subtraction strategies.

[0096] FIG. 5D depicts an array 5 comprising three sensor pods 1, and a microphone 27 on the body of the array. In this manner, the microphone 27 can pick up ambient sounds, but will be separated from the sounds of the arterial circulatory system that is being investigated. The microphone 27 can be any ordinary microphone or can be a copy of the piezo that is each of the sensor pods 1 so that the sounds can be closely matched. [0097] FIG. 5E depicts a microphone or piezo 28 depicted on a base 300. FIG. 5F depicts a microphone 30 or piezo on a cart 32 device.

[0098] A particular method comprises a method of reducing noise to a sensor comprising: engaging a first sensor to a patient and a second sensor to ambient air, adjacent to said first sensor; detecting noises from said patient and simultaneously detecting noises from ambient air with said second sensor; subtracting the noise from said second sensor from the data from said first sensor, which will remove the ambient noise from the data from the first sensor.

[0099] A particular method utilizes a phase change detected from a sensor to modify the sounds received at an adjacent sensor; a first sensor placed on a patient to detect sounds from the patient; a second sensor placed adjacent to said first sensor but shielded from the sounds of the patient; performing a phase change on the sounds received in said second sensor and emitting a proportional sound in said phase change.

[00100] Analysis based noise filtration methods

[00101] Active and passive cancellation can provide for a dramatic reduction in the amount of noise that ends up in a set of collected data. However, even with these background strategies to reduce and eliminate noise, detection of low frequency sounds can often be understood as looking at sounds that are "in the weeds." Accordingly, further processing may be necessary, in certain embodiments, to collect data, amplify the data and perform certain analysis using a computer to clarify the data for best analysis.

[00102] Spectrum analysis, also referred to as frequency domain analysis or Power Spectral Density ("PSD") estimation, is the technical process of decomposing a complex signal into simpler parts. As described above, many physical processes are best described as a sum of many individual frequency components. Any process that quantifies the various amounts (e.g.

amplitudes, powers, intensities, or phases), versus frequency can be called spectrum analysis.

[00103] Spectrum analysis can be performed on the entire signal. Alternatively, a signal can be broken into short segments (sometimes called frames), and spectrum analysis may be applied to these individual segments. Periodic functions (such as sin(t) are particularly well- suited for this sub-division when t (time) includes several cycles. General mathematical techniques for analyzing non-periodic functions fall into the category of Fourier analysis.

[00104] The Fourier transform of a function produces a frequency spectrum which contains all of the information about the original signal, but in a different form. This means that the original function can be completely reconstructed {synthesized) by an inverse Fourier transform. For perfect reconstruction, the spectrum analyzer must preserve both

the amplitude and phase of each frequency component. These two pieces of information can be represented as a 2-dimensional vector, as a complex number, or as magnitude (amplitude) and phase in polar coordinates (i.e., as a phasor). A common technique in signal processing is to consider the squared amplitude, or power. In this case the resulting plot is referred to as a power spectrum.

[00105] In practice, nearly all software and electronic devices that generate frequency spectra apply a Fast Fourier Transform ("FFT"), which is a specific mathematical approximation to the full integral solution. Formally stated, the FFT is a method for computing the discrete Fourier transform of a sampled signal.

[00106] Because of reversibility, the FFT is called a representation of the function, in terms of frequency instead of time; thus, it is a frequency domain representation. Linear operations that could be performed in the time domain have counterparts that can often be performed more easily in the frequency domain. Frequency analysis also simplifies the understanding and interpretation of the effects of various time-domain operations, both linear and non-linear. For instance, only non-linear or time-variant operations can create new

frequencies in the frequency spectrum.

[00107] The Fourier transform of a stochastic (random) waveform (noise) is also random.

Some kind of averaging is required in order to create a clear picture of the underlying frequency content (frequency distribution). Typically, the data is divided into time-segments of a chosen duration, where time is long enough to include several cycles of typical frequencies, and transforms are performed on each one. Then the magnitude or (usually) squared-magnitude components of the transforms are summed into an average transform. This is a very common operation performed on digitally sampled time-domain data, using the discrete Fourier transform. This type of processing is called Welch's method or Entropy Maximum (Burg) method. These methods are known and understood by a person of ordinary skill in the art. When the result is flat, it is commonly referred to as white noise. However, such processing techniques often reveal spectral content even among data which appear noisy in the time domain.

[00108] Accordingly, by taking a piezoelectric unit, capable of measuring sounds and vibrations at low amplitude and within a particular frequency range, we can measure the wall pressure fluctuations due to stenosis. Accordingly, the sensitive piezoelectric devices combined with amplifiers are placed onto the skin above the carotid artery and the piezoelectric device detects these sounds. The detected sounds are then passed through analog to digital converters before reaching a computer in which further amplification and an analysis of the signal occurs.

[00109] In the case of the arterial circulatory system, the piezo is placed on the skin above the artery in the region of a suspected stenosis. In the case of a carotid artery the placement would be on the neck, slightly below the ear. The particular placement of the piezo and the location of the stenosis is suggested by Fredberg and Borisyuk. Indeed, in an artery, between the stenosis and the region where turbulence has significantly decayed, the intensities can be rather large, where the wall can be subjected to large fluctuating stresses imposed by the turbulent blood flow. [Fredberg 1974] The distance over which this occurs is estimated to be about 12D downstream, where D is the normal diameter of the carotid artery. Borisyuk [2010]. For a typical internal carotid D of 0.5 cm, that distance would be of the order of several cm.

[00110] Detection of vortices generated due to flow in the carotid artery produce low intensity sounds that are related to development of stenosis in an artery. These low intensity sounds are sometimes difficult to detect and to pull out of the mass of noise being generated by the body. Accordingly, a highly specialized detection device using piezo devices for arteries that are near the surface. In the relevant frequency range of 20 Hz to about 3000 Hz the wavelengths are long compare to other lengths, such as artery length or thickness of tissue between the artery and the skin. In this case the surface is still within the "near field" of a wave (much closer than one wavelength), in which case the tissue acts as an incompressible medium. The energy in the near field of a wave is attached to the source and cannot propagate away. Thus there is no net energy flux out from the source. Because near-field pressure fluctuations cannot propagate away, they are generally called "pseudo- sound".

[00111] Borisyuk [2010] has been able to relate the shape of the power spectrum at the surface to the vortex structures in the blood flow distal to a constriction. He divides the region distal to a constriction into three: Region I. The flow separation region, in which a jet flow of higher velocity, in the center, acts separately from the slower flow outside the jet. Region II. The flow reattachment region. The two regions, I and II, constitute the "most disturbed flow region". The length of the first two regions, I plus II, based upon extensive calculations, Borisyuk estimates to be less than 7D, where D is the normal diameter of the artery. Here, stenosis may be detected in several different arteries in the arterial circulatory system. For example, detection may be directed towards detecting stenosis in the Internal Carotid Artery (ICA) in an adult, in which D is approximately 0.7 cm but the internal carotid is typically 0.5 cm. Therefore, the total length of the regions spoken of, I and II, would be at most about 3.5 cm. Region III is the region of flow stabilization where flow develops into the less turbulent flow farther upstream. This region extends from at most, 7D to 12D, or in the case of the ICA, at most from about 3.5 cm to about 6 cm.

[00112] Conservation of fluid requires that v = V (D/d) ^A 2. Let lower case v be the flow velocity inside the constriction and capital V the flow velocity past the constriction. Let d be the diameter of the flow inside the constriction. Borisyuk suggests estimates of two characteristic ring vortex frequencies. The first, fl, of vortices inside the jet, with typical size d; the second, f2, of vortices between the jet and the outer wall, with typical size, D.

[00113] Accordingly, Borisyuk provides for a broad disclosure that certain structures in the blood generate flow patterns. Based on these flow patterns, and separated into three regions, Borisyuk estimates characteristics of vortex frequencies. However, these estimations provide only a rough estimate as to a vortex structure.

[00114] Accordingly, our method for determining stenosis consists in connecting the frequencies associated with largest intensities in the spectral domain to three frequencies, fl thru f2 in order to obtain estimates of percentage stenosis of the artery, (l-d/D)xl00.

[00115] Description of Ring Vortices

[00116] FIG. 23 is the side view of a ring vortex showing the rotation of the core, the velocity of the motion of the center of the core (u'), and the diameter of the vortex (d). In a carotid artery, the diameter of the vortices are initially equal to the diameter of the stenosed region. This is followed by a second region in which the diameter is equal to the inside diameter of the artery. Note that the core is thin compared to the radius of the entire ring. Inside the core, the blood molecules rotate as shown by FIG. 23 in circular or near circular (elliptical) motion around the center of the core. A blood molecule farther from the center rotates at higher velocity than one which is closer to the center. This is similar to a solid disk. The rotational motion is coherent, which maintains the same angular velocity without friction between particles at different distances from the center. This solid like motion eliminates internal frictional, dissipative forces, which if they existed would diminish the energy of the rotation quite rapidly. In such a case, the vortices would not travel nearly as far, turning to full turbulence at shorter distance of motion.

[00117] The ring vortices are produced equidistant from each other at a distance between them equal to their diameter as they move downstream, as illustrated in Fig (Ring Vortex Figure 2) which shows the formation of ring vortices upon the exit of air from a long tube. In this well- known experiment, air is being blown from a cylinder due to the motion of a piston within the cylinder. As the air departs the cylinder at sufficient velocity, ring vortices in the emerging air are formed and remain at the same diameter and distances between adjacent vortices for the entire distance that they travel. They will later dissipate into smaller eddies, which is called full turbulence. As the ring vortices pass the flame, the high speed of the air within the core of the ring vortices will blow out the flame. The air ring vortices are sufficiently stable to travel a distance of 10-20 times the distance between the individual rings. The arrows above and below the cylinder shows that air spreads out as it leaves the cylinder because there are no containing walls. Yet the diameter of the rings does not increase as they move toward the flame. Within the carotid artery, the medium is blood rather than air but the behavior is the same if the Reynolds number is the same. In the artery blood is not free to expand beyond the size of the artery, however, the size of the vortices in the flow of blood remains the same diameter as the orifice (stenosis) opening, even though the size of the artery is larger than the diameter of the vortices.

[00118] In FIG. 24 which illustrates vortices in air the size of the vortices is a small percentage larger than the size of the cylinder opening. In the flow of blood in which the flow is restricted to the size of the artery rather than being free to expand, the size of the vortices is the same as the size of the jet emerging from the stenosed section of the artery. Note that the most recent vortex formed is at a distance of approximately one vortex diameter from the orifice. A microphone placed to the side of the vortex flow will measure sound at a frequency given by the frequency in which the vortices pass in front of the microphone. Sound is produced by the vortices because the rapid motion of molecules inside the ring is highly organized, that is non- random, which causes lower pressure at the surface of each individual vortex ring. This lower pressure at the surface of the vortex ring followed by a higher pressure between the vortex rings, causes sound to be transmitted to the microphone. This is the same principle as occurs in the passage of ring vortices within a blood vessel [Mollo-Christensen, Kolpin, and Marticcelli, "Experiments on jet flows and jet noise far-field spectra and directivity patterns," Journal of Fluid Mechanics 1964, Vol. 18, Iss. 2, 285-301]. Note the sound is produced in a direction perpendicular to the motion of the ring vortices, along the axis of the artery.

[00119] Johansen 1930, Figure 8 of "Flow through pipe orifices at low Reynolds numbers," Proceedings of the Royal Society A, vol. 126, 231-245 describes a photograph of blood flow below the critical value. This Reynolds Number (RE) equals vD/η = 600, where v is blood velocity, D is diameter of artery, and η is blood kinematic viscosity (equals 0.035cm ^A 2/s, at human temperature). Flow is from left to right. Note that there are no ring vortices yet formed since the velocity of the blood is too low as in the diastolic phase of the cardiac cycle and latter part of the systole. There are however small striations which occur at RE lower than the critical value, but vortex rings do not yet form.

[00120] At RE less than 800 or greater than 2100, ring vortices do not form. The closer to

800 while still remaining below 800, the more string-like motions are seen, as seen in Johansen. At greater than 2100, the vortices break-up into small eddies with random orientations. In Becker & Massaro, Figure 5, number 2 of "Vortex evolution in a round jet" Journal of Fluid Mechanics 1968, vol. 31, part 3, 435-448 shows 3 ring vortices emerging from an orifice. Note that the ring vortices without confining walls disintegrate into small eddies after only three ring vortices. Also note that the diameter of the ring vortices remains constant and the distance between adjacent vortices is equal to the diameter of a single vortex.

[00121] Johansen 1930, Figure 8 shows the blood flow pattern including ring vortices when the RE is 1000, which is above the critical value for ring vortices to be formed. The blood flow is from left to right, the transition region from smaller diameter vortices to larger occurs rapidly in less than the distance between two of the larger vortices. The centers of all vortices, small or large, travel at the same speed. We call the first region, with smaller diameter vortices, Region I. The region of the larger vortices we call Region II. Region III follows Region II, where the vortices have disintegrated into small eddies. Because the vortices in Region I are closer together a higher sound frequency is produced, which we call f2, than is produced by the larger vortices which have a larger distance between them which produce lower sound frequency, fl . The diameter of the small vortices matches the diameter of the stenosed region. The diameter of the large vortices matches the diameter of the blood vessel in the non-stenosed region. The ratio of the two frequencies is the same as the ratio of the diameters, from which percentage stenosis can be determined. Variations from one patient to another in diameter of artery, velocity of blood, blood viscosity, temperature, and other variables cancel when taking the ratio of the two frequencies. In each heart cycle, the velocity rises above critical value during systole, and drops below critical value during diastole. Typical values for the Internal Carotid Artery (ICA) at Peak Systolic Velocity (PSV) range from velocity of 64-77 cm/s and diameter of ICA between .511cm (for men) and .466cm (for women) yield RE equal to 852 (for men) and 1124 (for women), well within the range that produces ring vortex flow in the ICA. Also the ring vortices only appear during the deceleration phase of the systolic part of the heart cycle that is following the moment of peak systolic velocity. Using the formulas given on by Becker and Massaro [1968, pg 446], f*d/v=0.0122*Sqrt(RE), where v is the blood velocity, d is the diameter of the vortices, and f is the observed frequency seen at the microphone placed over the artery. Typical values of the solution of this equation at 50% stenosis yields fl=178 Hz and f2=356 Hz with a similar formula from other authors also quoted by Becker and Massaro [1968, pg 446], one obtains fl=236 Hz and f2=472 Hz. Different patients at 50% stenosis could have different values of frequency for the two peaks, but they will remain at the same proportionality.

[00122] If no fl appears in the PSD (between 60 and 260 Hz), there was insufficient energy in the flow emerging from the stenotic region for the vortices to reach Region II, in which the larger vortices appear, at the lower frequencies. This indicates the artery is heavily stenosed. If there is no f2, there is an insufficient amount of stenosis to create the smaller vortices (Region I) indicating a low level of stenosis (below 15%) as reported by Khalifa and Giddens

["Characterization and evolution of poststeotic flow disturbances," Journal of Biomechanics 1981. Vol. 14, No. 5, pg292] who report that below 25% reduction in area due to stenosis (which corresponds to a reduction of 13% in diameter), no signal is picked up. If there is neither fl nor f2, the indication is that there is a near blockage level of stenosis, as the vortices cannot be produced even when the velocity is sufficient to give RE between 800 and 2100.

[00123] The method has been implemented in a computer language we convert to binary, encrypted to be packaged as one whole product, software and hardware. The particular software used to run the data analysis can be determined by a person of ordinary skill in the art.

[00124] A particular embodiment comprises the following steps: A sensor device is placed on a patient and data is sampled from the patient and the sound/vibrations are converted from analog to digital. The data is streamed from the device with both of the sensors in one data stream. We break the data stream down into two streams, one for the left sensor and one for the right. We then begin the Wavelet analysis which takes out noise. After the Wavelet removes the noise a power spectral density analysis is done and we are given a power spectral density (PSD). This tells us what frequency noise is found within the data and how strong/powerful the noise is. Because the PSD gives transient noise smoothing the PSD must be done to correctly identify the strongest peaks within the data. After smoothing is done peaks are determined and based on the where the peaks are will determine the amount of stenosis or whether no stenosis is present. If there is one peak, No stenosis is present. If there are two or more peaks the patient has some stenosis present.

[00125] Wavelets have been frequently used in digital signal processing and are often known as small waves. A wavelet is a real-valued integral function ψ : R→ R satisfying Z ψ(ΐ) dt = 0. For practical applications, it has n vanishing moments: Z t ρψ(ΐ) dt = 0, p = 0, 1, . . . , n - 1. Consider the following family of dilations and translations of the wavelet function ψ defined by = 2-j/2v|/(2-j t - k), j, k = 0, ±1, ±2. The terms j and 2j are called the octave and the scale, respectively. By construction, this family consists of orthogonal basis functions in the sense that for a given time series or observed signal or simply data y(t), it can be written as the sum of these basis functions in a unique way: y(t) = X j where djk is the discrete wavelet transform (DWT) of y(t) given by djk = Z , k = 0, ±1, ±2. In practice, data is decomposed into its rough approximation at the chosen resolution level J (signal of interest) and details on a finite number of resolution levels j(< J). The latter will be considered as noise.

Denoising is equivalent to removing the details to allow for improved fit and prediction of peaks in a PSD plot.

[00126] An example of the process for calculation:

[00127] FIG. 8 details a flow-chart of the process for de-noising a sample after the passive and active noise cancellation steps. A first step is to read in data and separate it into different channels 70, based upon the number of piezo sensors. A single sensor will have only one channel, two sensors two channels, and three, as in FIG. 8, three channels, etc.

[00128] We next perform a wavelet analysis 71, to de-noise the data by removing low- frequency components 1-60 or 1-70 Hz. After the wavelet analysis we generate a Power Spectral Density (PSD) 73 using the denoised data, in combination with Welch and/or Burg's method. From this PSD plot, we detect a first spike, typically between 75-250 Hz, (74) though it can go as low as 60 Hz. Where lower peaks are present, the Wavelet is re-run to remove a lower set of data, so that the first peak is not obfuscated.

[00129] If a first spike is present between 75 and 250 Hz, we continue data acquisition

(74) . In certain embodiments, if there is no spike in this range, the sensor is adjusted (72) and the data acquisition process is re-started. Using this embodiment, we effectively build in a mechanism to ensure proper placement of the sensor, to make sure we have good quality data. However, other sounds may be utilized as a predetermined sound for ensuring proper placement in other embodiments.

[00130] Once we have a first spike between 75 and 250 Hz, a second spike is analyzed

(75) , as different from the first and less than 3000 Hz. (feature 75). If the second spike is not found in this range, we declare stenosis at less than 25%. If the second spike is in this range, then we can calculate stenosis by peak comparison using the formula. We use the formula (1- fl/f2)x 100%, where fl is the base frequency for the ring vortices in the artery (between 60 and 260 Hz) and f2 is the frequency from the restricted ring vortices (below 3000 Hz). If fl is not present, the artery is too stenosed to show a base ring vortex and therefore we conclude there is a very high level of stenosis. If f2 is not present then we conclude that there is insufficient stenosis to create a restricted ring vortex and thus we say there is a very low level of stenosis. If neither fl nor f2 are present, the patient is stenosed to the point where ring vortices can no longer form. This patient has extremely high stenosis and needs to see a specialist as soon as possible.

[00131] Example of data analysis

[00132] Read in data and look for extraordinary features. The step is important for reviewing if the device has followed protocol or not, and whether the subject has complied with the data acquisition procedures.

[00133] The function CVRData provides a pop-up menu asking a user to select data, followed with a graph plotting channels, selected from Left - channel 1, Right - channel 2, or center - channel 3. One or all channels can be selected.

[00134] The data of FIG. 10 depicts wherein y = CVRData. The variable y contains all three channels. Additional analysis in selecting channels is provided in a further step. The output of FIG. 10 was constructed from "plot3ch.m". The subject ID appears in the title of the last panel.

[00135] To select a channel to analyze, we look at the following aspects: [00136] Ch=l; note that Left or Ch=l, Right or Ch=2, and center or Ch-3.

[00137] Setup of basic parameters for data analysis. Variable x is one of the channels in the following formula x = y(ch:3 :length(y));

[00138] Fs is the sampling rate, wherein Fs=20,000;

[00139] One second record: the variable t is used for data visualization by plotting the first Fs or one second record of the channel values. Accordingly we can use the data:

[00140] t=(0:Fs)/Fs; subplot(l 11), plot(x(l:10*Fs)), title ('Ten second channel plot )

[00141] The resulting channel plot is depicted in FIG. 11.

[00142] A periodogram is generated. In general, one way of estimating the PSD of a process is to simply find the discrete-time Fourier transform of the samples of the process (usually done on a grid with an FFT) and appropriately scale the magnitude squared of the result. This estimate is called the periodogram.

[00143] Periodogram(x, hamming(length(x)), length(x), Fs); xlabel( Frequency (Hz) ').

[00144] FIG. 12 depicts the periodogram PSD estimate.

[00145] The number of frequencies plotted is 1 + half of length (x) and the unit is Hertz

(Hz).

[00146] Welch's Method can be used as an improved estimator of the PSD. Welch's Method, as known to a person of ordinary skill in the art, consists of dividing the time series data into (possibly overlapping) segments, computing a modified periodogram of each segment, and then averaging the PSD estimates. The result is Welch's PSD estimate.

[00147] The averaging of modified periodograms tends to decrease the variance of the estimate relative to a single periodogram estimate of the entire data record. Although overlap between segments introduces redundant information, this effect is diminished by the use of a nonrectangular window, which reduces the importance or weight given to the end samples of segments (the samples that overlap).

[00148] However, as mentioned above, the combined use of short data records and nonrectangular windows results in reduced resolution of the estimator. In summary, there is a tradeoff between variance reduction and resolution. Once can manipulate the parameters in Welch's method to obtain improved estimates relative to the periodogram, especially when the S R is low. This is illustrated in the following example: [00149] A signal such as x consisting of the left channel data pwelch(x); which is graphically represented in FIG. 13.

[00150] The graph of FIG. 13 depicts the normalized frequency.

[00151] Parameters to be specified with the Welch's method must be considered. The first parameter is the segment length. Default length is (x)/8. In code we use SGM= 100,000. The next parameter is percent of overlaps: novoerpals=50,000.

[00152] Through these elections we obtain Welch's overlapped segment averaging PSD estimate of the preceding signal. Use a segment length of 100,000 samples with 50 overlapped samples. Use l+length(x)/2 DFT points so that 100 Hz falls directly on a DFT bin. Input the sample rate to output a vector of frequencies in Hz. We can plot the result.

[00153] Example: [Pxx,F] = pwelch(x, sgm, noverlaps, [], Fs); plot (f, 10*logl0(Pxx)).

The result is the plot of FIG. 14.

[00154] We can further estimate PSD through autoregressive PSD estimate through use of

Burg's Method. Burg's Method is a parametric method for estimating PSD. Below returns a frequency vector, F, in cycles per unit time. The sampling frequency, Fs, is the number of sample per unit time. If the unit of time is seconds, then F is in cycles/second (Hz). For real- valued signals, F spans the interval [0,fs/2] when nfft is even and [0,fs/2] when nfft is odd.

[00155] The following formula assumes an AR(50) model to the data.

[00156] [Pxx,F] = pburg(x, 50, [], Fs); plot(F, 10*logl0(Pxx)). The result is plotted in

FIG. 15. A comparison between FIGS. 14 and 15 shows a much clearer set of peaks, allowing clearer determination of the stenosis.

[00157] We use AR(50) because we tested model orders starting from 5 through 50 and determined that AR(50) provided the cleanest data result.

[00158] Reflection Coefficients for Model Order Determination

[00159] The reflection coefficients are the partial autocorrelation coefficients scaled by -1.

The reflection coefficients indicate the time dependence between y(n) and y(n-k) after subtracting the prediction based on the intervening k-1 time steps.

[00160] Use of arburg to determine the reflection coefficients. Use the reflecting coefficients to determine an appropriate AR model order for the process and obtain an estimate of the process PSD. We use the following formula:

[00161] [a,e,k] = arburg(x,50); [00162] Stem(k, fi ' lled'); title ('Reflection Coefficients )' ; xlabel(" model Order )

[00163] FIG. 16 depicts the resultant Reflection Coefficients.

[00164] To find frequencies, we zoom into the data. Bf= 0.1000/129:3876

[00165] Plot(0: 1000/129:3876, 10*logl0(Pxx(l :51)))

[00166] Legend ('pburg PSD Estimate'); x label ('Frequency (Hz)'); y label

('Power/frequency (dB/Hz)'); title ('PSD before denoising'). The result is the data of FIG. 17.

[00167] We can then experiment with several choices of parameters in the Welch's PSD estimate, for example with 20 percent overlaps. Sgm=10, 000; noverlaps=2000; [Pxx,F] = pwelch9x, sgm, noverlaps, [], Fs); plot(F,10*logl0(Pxx)). This results in the plot of FIG. 18.

[00168] We can also test PSD by Welch with no overlaps:

[00169] Sgm= 10000; noverlaps=0; [Pxx,F] = pwelch(x, sgm, noverlaps, [], Fs);

[00170] Plot(F,10*logl0(Pxx)); xlabel('Frequency (Hz)'); ylabel('Magnitude (dB)'); title

('PSD before nenoising'). This results in the plot of FIG. 19.

[00171] If we zoon in the range of 2K Hz, with:

[00172] Uf= 2000; plot (Fl:uf), 10*logl0(Pxx)l:uj)))

[00173] xlabel('Frequency (Hz)'); ylab el ('Magnitude (dB)'); title ('PSD before nenoising'). This results in the plot of FIG. 20.

[00174] Finally, we can output with frequencies, for peak analysis with [Pxx, F] = pburg(Dl, 50, [], FsO' and zoom to within 2000Hz (though 3000 would be good as well).

[00175] Plot (0; 1000/129: 1938, 10*loglO(Pxx(l:26))) grid on;

[00176] Legend (p ' burg PSD estimate ')

[00177] xlabel('Frequency (Hz)'); ylab el ('Magnitude (dB/Hz)'); title ('Parametric PSD after denoising). This results in the plot of FIG. 21

[00178] We then allow the software to define the peaks. Once identified, the peaks can be used to calculate stenosis by (l-d/D)xl00.

[00179] Accordingly, we know that ambient noise is present in any data set and we know some of the sounds that are always present. Furthermore, we know the sounds that we are trying to detect and have determined that these sounds are at range 20-3000 Hz. We can remove other sounds introduced through these sensitive machines and concept is to provide a claim that covers the external and internal steps being applied to generate clean data.