Diagnostic Equipment

The diagnostic device uses a high-sensitivity, low-noise acceleration sensor to detect high-frequency muscle sound signals, overcoming the challenge of noise interference and enabling effective diagnosis of neurodegenerative diseases.

JP7576846B2Active Publication Date: 2024-11-01INSTITUTE OF SCIENCE TOKYO
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
JP2021552365
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-03-31
Filing Date
2020-10-09
Publication Date
2024-11-01
Estimated Expiration
2040-10-09

AI Technical Summary

Technical Problem

Muscle sound signals, particularly those above 15 Hz, are often buried under noise and difficult to separate from tremors, making it challenging to utilize them for diagnosing neurodegenerative diseases like Parkinson's disease.

Method used

A diagnostic device using a high-sensitivity, low-noise acceleration sensor to measure high-frequency muscle sound signals, with a noise density of 100 μG/√Hz or less, allowing detection of previously undetectable high-frequency components.

Benefits of technology

Enables accurate diagnosis of neurodegenerative diseases by analyzing high-frequency muscle sound signals, providing a new diagnostic support system that was not previously possible.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007576846000001
    Figure 0007576846000001
  • Figure 0007576846000002
    Figure 0007576846000002
  • Figure 0007576846000003
    Figure 0007576846000003
Patent Text Reader

Abstract

A diagnosis apparatus (100) is provided with an acceleration sensor (102) and a processing device (104). The acceleration sensor (102) can be attached to a subject (2) and has a noise density of 100 µG / √Hz or less. The processing device (104) diagnoses the subject (2) on the basis of a high frequency component included in a muscle sound signal which is obtained from an output from the acceleration sensor (102) and having a higher frequency than a predetermined frequency.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present disclosure relates to diagnostic devices. [Background technology]

[0002] The weak mechanical vibrations generated by muscles are called muscle sounds, and have been measured using microphones and accelerometers. However, muscle sound signals are very weak and easily affected by noise. Therefore, muscle sound signals have traditionally been identified as components observed in the low frequency band below 15 Hz in the state of maximum muscle output. On the other hand, high frequency bands above 15 Hz are buried under the noise floor, and muscle sound signals have not been found in this region. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Jankovic, J. Parkinson's disease: clinical features and diagnosis. J. Neurol. Neurosurg. Psychiatry 79, 368-376 (2008). [Non-Patent Document 2] Post, B., Merkus, MP, De Haan, RJ, Speelman, JD & CARPA Study Group. Prognostic factors for the progression of Parkinson's disease: a systematic review. Mov. Disord. 22, 1839-1851 (2007). [Non-Patent Document 3] Fahn, S. Clinical aspects of Parkinson disease. In Parkinson's Disease: Molecular and Therapeutic Insights form Model Systems, eds. R. Nass, S. Przedborski, (Elsevier Inc., Amsterdam, 2008), pp. 3-48. [Non-Patent Document 4] Jankovic, J., Rajput, AH, McDermott, MP & Perl, DP The evolution of diagnosis in early Parkinson disease. Arch. Neurol. 57, 369-372 (2000). [Non-Patent Document 5] Hausdorff, JM Gait dynamics in Parkinson's disease: common and distinct behavior among stride length, gait variability, and fractal-like scaling. Chaos 19, 026113 (2009). [Non-Patent Document 6] Hass, CJ et al. Quantitative normative gait data in a large cohort of ambulatory persons with Parkinson's disease. PLoS One 7, e42337 (2012). [Non-Patent Document 7] Ibitoye, MO et al. Mechanomyography and muscle function assessment: A review of current state and prospects. Clin. Biomech. 29, 691-704 (2016). [Non-Patent Document 8] Orizio, C. et al. The surface Mechanomyogram as a tool to describe the influence of fatigue on biceps brachii motor unit activation strategy. Historical basis and novel evidence. Eur. J. Appl. Physiol, 90, 326-336 (2003).

Non-Patent Document 9

Non-Patent Document 10

Non-Patent Document 11

Non-Patent Document 12

Non-Patent Document 13

[0004] Muscle sound signals are generally measured at maximum muscle output, and tremors and shaking occur simultaneously. The frequency of this tremor overlaps with the low-frequency band of muscle sound signals around 10 Hz that was previously measurable, making it difficult to separate them. This makes it difficult to use muscle sound signals to diagnose neurodegenerative diseases.

[0005] The present disclosure has been made in consideration of the above-mentioned circumstances, and one exemplary purpose of an aspect thereof is to provide a diagnostic device based on muscle sound signals. [Means for solving the problem]

[0006] An embodiment of the present disclosure relates to a diagnostic device. The diagnostic device includes an acceleration sensor that can be attached to a subject, and a processing device that diagnoses the subject based on high-frequency components (high-frequency bands) higher than a predetermined frequency contained in a muscle sound signal obtained from the output of the acceleration sensor. The acceleration sensor has low-noise characteristics with a noise density lower than 100 μG / √Hz.

[0007] By using a high-sensitivity, low-noise acceleration sensor, it is possible to measure high-frequency components of muscle sound signals, whose existence was previously unknown due to being buried in the noise floor. Then, by analyzing these high-frequency components, it becomes possible to diagnose, for example, neurodegenerative diseases, or other diseases or impaired motor function. It can be said that the present disclosure is novel in that it recognizes the existence of muscle sound signals in a high-frequency band whose existence was previously unknown, and finds that these high-frequency muscle sound signals can be used for diagnosis. Therefore, the acceleration sensor only needs to be highly sensitive and low-noise enough to detect these high-frequency muscle sound signals, and the specific value of 100 μG / √Hz is merely an example, and does not exclude the use of sensors with a noise density of, for example, 300 μG / √Hz or 200 μG / √Hz in this technology.

[0008] The predetermined frequency may be set between 10 and 30 Hz, and more preferably between 15 and 20 Hz.

[0009] The muscle sound signal may be measured under non-resting conditions in which the subject exerts force.

[0010] The force generated under non-resting conditions may be 50% or less of the maximum output. This allows the muscle output during measurement to be significantly reduced, thereby preventing the generation of low-frequency noise due to tremor and enabling the measurement of muscle sound signals originating from non-tremor. It is preferably 20% or less of the maximum output. The lower the force generated under non-resting conditions, the more the effect of low-frequency tremor is reduced, making it possible to measure low-frequency muscle sound signals with a high S / N ratio.

[0011] The non-resting condition may be a state in which the subject is holding an object with his / her fingers. This condition places a small burden on the subject and has excellent reproducibility.

[0012] The non-resting condition may be a state in which the subject performs lateral finger lift.

[0013] The non-resting condition may be a state in which an external weight is applied to the subject.

[0014] The muscle sound signal may be measured under resting conditions where the subject is not producing any force.

[0015] The processing device may perform diagnosis based on low-frequency components (low-frequency band) that are lower than a predetermined frequency and are included in the muscle sound signal, in addition to the high-frequency components.

[0016] The acceleration sensor may be attached to the superficial muscles or the flexible muscle group of the forearm on the subject's dominant hand. Alternatively, the acceleration sensor may be attached to the superficial muscles of the palm of the subject's dominant hand. These are muscle groups on the superficial side, and are suitable for measuring muscle sounds at the skin surface.

[0017] Any combination of the above components or conversion of the present disclosure between methods, devices, etc. are also valid aspects of the present disclosure. Effect of the Invention

[0018] Certain aspects of the present disclosure enable diagnosis based on muscle sound signals. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is a diagram showing a diagnostic device according to an embodiment. [Diagram 2] FIG. 1 is a diagram showing a state in which an acceleration sensor is attached to a subject. [Diagram 3] 3(a) and (b) show the spectra of muscle sounds measured for a patient and a healthy subject. [Figure 4]FIG. 13 is a diagram showing a spectrum of muscle sounds measured by a high-sensitivity acceleration sensor. [Diagram 5] FIG. 5(a) is a diagram showing an acceleration sensor unit used in the experiment, and FIG. 5(b) is a diagram showing how the sensor unit is attached to the human body. [Figure 6] 6(a) to (c) are diagrams showing resting conditions and wrist flexion conditions. [Figure 7] FIG. 1 shows an example of measurement results obtained under resting conditions in a healthy elderly person. [Figure 8] FIG. 1 shows an example of measurement results obtained on a healthy elderly person under wrist flexion conditions. [Figure 9] FIG. 1 shows an example of measurement results obtained under resting conditions for a healthy young person. [Figure 10] FIG. 1 shows an example of measurement results obtained for a healthy young person under wrist flexion conditions. [Figure 11] FIG. 13 is a diagram showing the acceleration spectra measured when a commercially available sensor and a high-sensitivity sensor are placed on a horizontal surface for one minute. [Figure 12] FIG. 1 shows the spectrum of acceleration of a PD patient measured by a high-sensitivity sensor under resting and wrist flexion conditions. [Figure 13] FIG. 13 shows the acceleration spectrum of a healthy elderly person measured by a high-sensitivity sensor under resting and wrist flexion conditions. [Figure 14] 13 is a spectrum of acceleration of a healthy young subject measured by a highly sensitive sensor under resting and wrist flexion conditions. [Figure 15] FIG. 1 shows spectra of a healthy elderly subject and a healthy young subject. [Figure 16] Figure 16(a) shows the energy (spectral area) in the frequency band above 15 Hz for healthy elderly subjects (HE) and healthy young subjects (HY), and Figure 16(b) shows the intermediate frequencies for healthy elderly subjects (HE) and healthy young subjects (HY). [Figure 17] 17(a) to (c) are diagrams for explaining the conditions of rest, outer weight application, and outer finger lifting. [Figure 18] 18(a) to (c) are diagrams showing spectra obtained for a PD patient and a healthy elderly person under three conditions: rest, lateral weight application, and lateral finger lifting. [Figure 19] 19(a) to (c) are diagrams showing spectra obtained by a high-sensitivity sensor for a PD patient and a healthy elderly person under three conditions: rest, lateral weight application, and lateral finger lifting. [Figure 20] 20(a) and (b) show the spectra obtained by a high-sensitivity sensor for a PD patient and a healthy elderly person. [Figure 21] 21(a) and (b) are diagrams showing the pinch condition and the rest condition. [Figure 22] A summary of participants in Experiment 3. [Figure 23] FIG. 4 is a diagram showing the mounting position of a sensor. [Figure 24] FIG. 1 is a diagram illustrating an analysis method. [Diagram 25] 25(a) and (b) show the average spectra of healthy elderly subjects and PD patients. [Figure 26] 26(a) and (b) are diagrams showing energy in the low frequency band and the high frequency band. [Figure 27] FIG. 1 shows spectra of PD patients and healthy elderly subjects measured under resting conditions. [Figure 28] 28(a) and (b) are diagrams showing energy in the low frequency band and the high frequency band. [Figure 29] FIG. 1 shows the spectra of a PD patient and a healthy elderly person measured under pinch conditions. [Diagram 30] 30(a) and (b) are diagrams showing energy in the low frequency band and the high frequency band. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] The present invention will be described below based on preferred embodiments with reference to the drawings. The same or equivalent components, parts, and processes shown in each drawing are given the same reference numerals, and duplicated descriptions are omitted as appropriate. In addition, the embodiments are illustrative rather than limiting the invention, and all features and combinations thereof described in the embodiments are not necessarily essential to the invention.

[0021] 1 is a diagram showing a diagnostic device 100 according to an embodiment. The diagnostic device 100 includes an acceleration sensor 102 and a processing device 104.

[0022] The acceleration sensor 102 is attached to the subject (patient) 2. The acceleration sensor 102 has high sensitivity characteristics with sensitivity on the order of mG to μG or nG, and low noise characteristics with a noise density of 100 μG / √Hz or less. The acceleration sensor 102 may be attached to a location where a change in the muscle sound signal is likely to occur when measuring the muscle sound signal, and typically, the acceleration sensor 102 may be attached to the upper limb 4 of the subject 2. More specifically, the acceleration sensor 102 may be attached to a muscle on the surface side of the upper limb 4, which is a muscle that allows muscle sound measurement at the skin surface, and may be attached to the superficial muscle and radial muscle group of the forearm, or the superficial muscle of the palm. Examples of the superficial muscle and radial muscle group of the forearm include the pronator teres, flexor carpi radialis, palmaris longus, flexor digitorum superficialis, flexor carpi ulnaris, brachioradialis, extensor carpi radialis longus, extensor carpi radialis brevis, extensor digitorum minimi, and extensor carpi ulnaris. Examples of superficial muscles of the palm include the abductor pollicis brevis, flexor pollicis brevis, opponens pollicis, abductor digiti minimi, flexor digiti minimi brevis, and opponens digiti minimi.

[0023] The processing device 104 diagnoses the subject 2 based on high-frequency components higher than a predetermined frequency from the muscle sound signal obtained from the output of the acceleration sensor 102. The predetermined frequency can be set between 10 Hz and 30 Hz. For example, the processing device 104 can be implemented as a combination of a computer and a software program, and can include a classifier obtained by machine learning. This classifier is implemented with a trained model generated using muscle sound signals obtained from patients with a specific disease and healthy individuals as training data (teacher data). The classifier may be generated using a support vector machine (SVM) or deep learning. Note that the feature of the present invention is what information is used to diagnose various diseases (i.e., what is used as the input of the classifier), and the implementation and learning method of the classifier are not essential to the present invention, and publicly known technologies may be used.

[0024] In the case of the diagnostic device 100 for diagnosing Parkinson's disease (PD), the classifier may output the Hoehn & Yahr severity classification, or the functional impairment classification.

[0025] Furthermore, the processing device 104 may extract low frequency components lower than a predetermined frequency from the muscle sound signal, and diagnose the subject 2 based on a combination of the high frequency components and the low frequency components.

[0026] "Diagnosis" here is not limited to whether or not subject 2 is suffering from a particular pathology, but also includes cases where a diagnosis is simply made of a decline in the function of the upper limbs (including the hands, wrists, forearms and upper arms).

[0027] The above is the configuration of the diagnostic device 100. The inventors have discovered that two types of muscle sounds can be detected by measuring muscle sounds using a high-sensitivity acceleration sensor 102 on the order of mG to μG. One is a new muscle sound signal with high-frequency components in the range of about 20 to 40 Hz. By lowering the noise floor by increasing the sensitivity of the acceleration sensor 102, it becomes possible to measure high-frequency components that were buried under the conventional noise floor. As will be explained in the experiment below, when these high-frequency components are measured on patients with certain types of pathology, there are cases where a significant difference is observed between them and healthy subjects. Therefore, diagnosis is possible based on these high-frequency components.

[0028] The other is a non-tremor-derived muscle sound signal that has a peak in the low frequency band around 10 Hz, which has been known for some time. In other words, by increasing the sensitivity of the acceleration sensor 102, it is possible to significantly reduce the muscle output during measurement, thereby preventing the generation of low frequency band noise due to tremor and making it possible to measure non-tremor-derived muscle sound signals. There are also cases where significant differences in this low frequency component are observed between patients with certain types of pathology and healthy individuals. Therefore, diagnosis is possible based on this low frequency component.

[0029] The above is the configuration of the diagnostic device 100 and the principle of measurement. When such muscle sound measurements were performed on patients with neurodegenerative diseases such as Parkinson's disease, it was confirmed that the high-frequency muscle sound peaks observed in healthy young people and healthy elderly people disappeared in the patients. In other words, by detecting the presence or absence of muscle sound signals in this high-frequency band, a new diagnostic support system that did not exist in the past can be constructed. Below, the results of an experiment on patients with neurodegenerative diseases such as Parkinson's disease will be described.

[0030] I Experiment 1 FIG. 2 is a diagram showing an acceleration sensor attached to a subject. Details of the acceleration sensor will be described later. In the experiment, two types of acceleration sensor were used: a commercially available acceleration sensor and an acceleration sensor developed in the Masu Laboratory of Tokyo Institute of Technology (referred to as a high-sensitivity acceleration sensor or simply as a high-sensitivity sensor in this specification). The newly developed high-sensitivity acceleration sensor is a capacitance-type acceleration sensor with a 1-axis Au weight. In this experiment, the muscle sound signal was measured under non-resting conditions in which the subject performed an outward finger-raising movement. Note that in some experiments described below, the force generated by the subject under non-resting conditions was set to 20% or less of its maximum output.

[0031] Figures 3(a) and (b) show the muscle sound spectrum (vertical axis: 10 dB / div) measured for patients and healthy subjects. (i) shows the measurement result using a commercially available acceleration sensor, and (ii) shows the measurement result using a high-sensitivity acceleration sensor. These measurement results revealed for the first time the existence of two types of muscle sounds that were previously buried in noise. One is a new muscle sound signal with high-frequency components that peaks around 20 to 40 Hz, and the other is a non-tremor-derived muscle sound signal with a peak in the low-frequency band around 10 Hz. This is due to the lowering of the noise floor by increasing the sensitivity of the sensor, and the drastic reduction in muscle output during measurement to prevent the generation of low-frequency noise due to tremor.

[0032] Figure 4 shows the muscle sound spectrum (vertical axis: 5 dB / div) measured by a high-sensitivity acceleration sensor. When muscle sound measurements were performed on patients with neurodegenerative diseases such as Parkinson's disease using a high-sensitivity acceleration sensor, it was found that the muscle sound peaks in the high frequency band around 20 to 40 Hz observed in healthy subjects disappeared in the patients.

[0033] These experimental results show that by measuring muscle sound signals in this high-frequency band, it is possible to build an early diagnosis support system that did not previously exist.

[0034] II Experiment 2 1. Introduction Parkinson's disease (PD) is a progressive, permanent neurological disorder caused by the gradual degeneration of dopaminergic neurons in the substantia nigra. PD is accompanied by movement and muscle dysfunction (Refs. 1, 2). For example, typical movement-related symptoms such as bradykinesia, akinesia, muscle rigidity, and resting tremor have been reported to appear in PD (Ref. 3). These symptoms significantly reduce the quality of life of patients. Because PD is degenerative and permanent, it is important to diagnose and evaluate its severity as early as possible (Refs. 1, 4). Many PD patients suffer from a writing disorder called micrographia, in which handwriting becomes abnormally small. In another example, many PD patients have difficulty using tableware due to muscle dysfunction.

[0035] Recently, to assess the mechanical activity of skeletal muscles, there has been focus on muscle sound or mechanomyography (MMG), in which micro-muscle vibrations from muscle contractions are measured by small, lightweight condenser microphones, accelerometers, or piezoelectric contact sensors on the skin (Ref. 7).

[0036] For example, the amplitude of MMG correlates with the power generated by the muscle (Reference 8). Normal tremor in healthy individuals at rest is observed as an MMG of about 10 Hz with spectral components (Burne 1984, in Gallasch 1997). MMG has different characteristics from another method of evaluating muscle function, electromyography (EMG). EMG represents the electronic signal that activates the motor units. The correlation between MMG amplitude and muscle force seen in fresh muscles disappears in fatigued muscles, but the correlation between EMG amplitude and muscle force is shown in both fresh and fatigued muscles (Reference 8). Therefore, MMG is expected to compensate for and complement the shortcomings of EMG (Reference 7).

[0037] Only a few studies have been reported investigating MMG in PD patients. For example, MMG during maximal or near-maximal force use in PD patients was investigated. Brown showed that stronger MMG at approximately 10 Hz and weaker MMG at 40-50 Hz were observed in the wrist extensor muscles of the forearm of PD patients during strong flexion and extension of the wrist compared to healthy non-medicated individuals. This change in MMG in PD patients was associated with a disappearance of the electromyogram at 40-50 Hz, which is observed in healthy individuals during the highest muscle contractions (Ref. 9).

[0038] Marciak et al. measured the upper limb muscles of PD patients while holding a submaximal load (2 kg) on ​​the wrist and found that higher MMG was observed at around 5 Hz compared to healthy subjects (Reference 10). This higher MMG is thought to be caused by the action tremor of PD, which is often observed during submaximal muscle contractions.

[0039] However, MMG during weak force application in PD patients is still unclear, which may be related to the difficulty of using micrographs and tableware. The reason is the technical limitations of MMG measurement. Low MMG accompanied by weak muscle contractions is difficult to detect due to the limited resolution of MMG sensors, such as small and lightweight microphones and accelerometers. In particular, high-frequency MMG was not measurable because the high-frequency MMG was too small compared to the low-frequency MMG. For example, the power spectral density (PSD) of MMG above 15 Hz is 100 to 1000 times smaller than the power spectral density (PSD) of about 5 Hz in PD patients holding submaximal loads (Ref. 10).

[0040] The diagnosis of PD is carried out using multiple perspectives, including movement disorders, CT, and the effect of L-dopa administration. However, the detection of movement disorders relied on qualitative evaluation by physician observation. Therefore, a major problem remains in the quantitative and objective evaluation of movement disorders. It is necessary to develop a system that can easily and quantitatively measure and evaluate movement disorders in PD. This study proposes a device and system that supports the diagnosis of PD using muscle sound measurement by a lightweight and highly sensitive acceleration sensor.

[0041] The purpose of this study is to investigate the MMG of PD patients while straining weak muscles using a highly sensitive, compact, lightweight acceleration sensor. The newly developed highly sensitive acceleration sensor is capable of detecting PSDs ten times smaller than commercially available ones (Reference 12). In this experiment, we investigated the MMG of PD patients, healthy elderly subjects, and healthy young subjects, and clarified the characteristics of the MMG of PD patients in response to weak contractions of the muscle belly of the extensor carpi radialis longus of the forearm under resting and wrist flexion conditions.

[0042] 2. Experimental Apparatus and Methods

[0043] The specifications of the high-sensitivity acceleration sensor used in the experiment are as follows: Noise density: 100μG / √Hz or less Sensitivity: 297fF / G Sampling rate: 85Hz or higher ·Weight: 10g or more

[0044] This high-sensitivity acceleration sensor is a MEMS acceleration sensor that detects acceleration as a change in capacitance, and its sensitivity is defined as the change in capacitance per unit of acceleration. Ultimately, the capacitance is converted into an electric signal such as a voltage signal by a capacitance detection circuit (C / V conversion circuit), and the signal is processed.

[0045] Figure 5(a) shows the acceleration sensor unit used in the experiment, and Figure 5(b) shows the attachment of the sensor unit 110 to the human body. The developed high-sensitivity acceleration sensor and a commercial acceleration sensor (MPU6050, InvenSence, USA, Ref. 13) were fixed on one module board and housed in a special case (see Figure 5(b)). The weight of the sensor and the special box is 9.0 g, and the width, depth, and height of the box are 20, 24, and 19 mm, respectively. The noise of the commercial and developed acceleration sensors are 400 μG / √Hz and 100 μG / √Hz, respectively. The noise floor of the PSD of the developed acceleration sensor in the acceleration spectral density is -80 dB, which is 1 / 10 times lower than the -70 dB of the commercial sensor. The sampling rates of the developed high-sensitivity acceleration sensor and the commercial acceleration sensor are 87.2 Hz and 90 Hz, respectively.

[0046] 3. Experimental Method The measurement method for MMG is as follows. Locomotor activity …Wrist extension or flexion …Hold the hand grip dynamometer and apply a certain amount of force External Load …holds constant weight ·break …without voluntary movement or external load

[0047] MMG measurements were conducted under two conditions: resting condition and wrist flexion condition. The acceleration sensor unit was fixed to the abdominal muscle of the extensor carpi radialis longus on the participant's dominant hand using medical paper tape. Figure 6(a)-(c) show the resting condition and wrist flexion condition. The microcontroller unit was attached to the upper arm using a special band. During the trial, the participant shall sit on a chair and place his / her forearm on a towel laid on a table. Under the resting condition, the participant is asked to open his / her fingers slightly and relax. Under the wrist flexion condition, the participant's hand shape and wrist flexion angle were set. First, the participant shall grasp an 8 cm diameter cylinder from above with all fingers and thumb as shown in Figure 6(c). Next, the wrist flexion angle shall be set to 150 degrees using a set square (Figure 6(b)). Finally, the experimenter shall pull the cylinder from the participant's hand. Immediately after the trial under the wrist flexion condition, the experimenter checked the flexion angle whether the participant maintained the angle. When the participant was unable to maintain the angle, another trial was conducted. During the test, participants were asked to close their eyes and remain motionless. Each condition was performed three times, resulting in a total of six trials for each participant. Measurements took 1 min for each trial. There was a break of at least 1 min between trials. To synchronize and calibrate the two accelerometers, participants were asked to move their arms slowly at the beginning of each trial. Participants rested their elbows on a towel and raised their forearms by approximately 30 degrees.

[0048] 4. Statistical methods For the analysis of acceleration as muscle sounds, we used time and frequency domain features of acceleration, such as the amplitude and frequency characteristics of acceleration. The two acceleration sensor data were synchronized by matching the maximum value of the arm movement at the start of each trial. For the calibration of the two sensors, the acceleration of the developed sensor was divided by the maximum value of the acceleration of the developed sensor and multiplied by the acceleration value of the commercial sensor. To investigate the frequency characteristics of MMG, we calculated the power spectral density (PSD) using Welch's method. The first 5 seconds of the acceleration data were removed and 50 seconds of data were used.

[0049] Figure 7 shows an example of the measurement results obtained for a healthy elderly person under resting conditions. The top diagram shows the output waveform of the acceleration sensor, and the horizontal axis shows the number of samples versus time. The bottom diagram shows the power spectrum density of the part of the waveform in the top diagram enclosed by a rectangle.

[0050] Fig. 8 is a diagram showing an example of a measurement result obtained on a healthy elderly person under a condition of bending the wrist, Fig. 9 is a diagram showing an example of a measurement result obtained on a healthy young person under a resting condition, and Fig. 10 is a diagram showing an example of a measurement result obtained on a healthy young person under a condition of bending the wrist.

[0051] 5. Results and Discussion Figure 11 shows the acceleration spectra measured with a commercial sensor and a high-sensitivity acceleration sensor placed on a horizontal surface for one minute. The spectra obtained show the noise floor of each acceleration sensor. The noise floor of the high-sensitivity acceleration sensor is more than 10 dB lower than that of the commercial sensor.

[0052] FIG. 12 shows the acceleration spectrum of a PD patient measured by a high-sensitivity sensor under resting and wrist flexion conditions.

[0053] FIG. 13 shows the acceleration spectrum of a healthy elderly person measured by a high-sensitivity sensor under resting and wrist flexion conditions.

[0054] FIG. 14 shows the acceleration spectrum of a healthy young person measured by a high-sensitivity sensor under resting and wrist flexion conditions.

[0055] Figures 12 to 14 show the acceleration spectra measured by the high-sensitivity sensor under resting and wrist flexion conditions for PD patients, healthy elderly people, and healthy young people, respectively. For the high-sensitivity sensor, the acceleration spectra under resting and wrist flexion conditions are lower than the noise floor of the commercially available sensor in many frequency bands. This result makes it clear that the high-sensitivity sensor measures micromuscle sounds that cannot be observed by commercially available sensors, especially in the high-frequency band. No difference is observed between the spectra under resting and wrist flexion conditions for PD patients. On the other hand, it can be seen that the spectrum under the wrist flexion condition is higher than that of the healthy elderly and healthy young people under resting conditions, especially in the high-frequency band.

[0056] It can be said that this experimental result shows that in the case of PD patients, the high frequency band muscle sound signal is weakened. Non-Patent Document 14 discloses the knowledge that the activity of the subthalamic nucleus of the basal ganglia and the coherence of the myoelectric potential are weakened in PD patients. In other words, the diagnostic device according to this embodiment may be able to observe the weakening of the activity of the subthalamic nucleus of the basal ganglia and the coherence of the myoelectric potential as the weakening of the high frequency band muscle sound signal. Conventionally, this coherence could only be measured by deep brain stimulation therapy, which required surgically inserting electrodes into the basal ganglia of PD patients. It can be said that according to this embodiment, the coherence can be easily measured without performing a surgical procedure.

[0057] FIG. 15 shows the spectra of healthy elderly and healthy young subjects. It can be seen that the power in the high frequency band is increased under bending conditions compared to resting conditions for both healthy elderly and healthy young subjects. Therefore, the energy (spectral area) was calculated in the frequency band of 15 Hz or more, and statistically tested for the large power. FIG. 16(a) shows the energy (spectral area) in the frequency band of 15 Hz or more for healthy elderly subjects (HE) and healthy young subjects (HY). It can be seen that the energy in the high frequency band increases during bending for both healthy elderly and healthy young subjects (p<0.01). Furthermore, no significant difference was found between elderly and young subjects (p=0.947).

[0058] Since no high-frequency peaks are visible when averaging across individuals, we calculated the median power frequency (MdPF or MF) in the frequency band above 15 Hz, i.e., the frequency that divides the spectral area in half, and examined the shift in the spectrum in the frequency direction. Figure 16(b) shows the median frequency of healthy elderly subjects (HE) and healthy young subjects (HY). It can be seen that the median frequency increases during flexion in both healthy elderly subjects and healthy young subjects, that is, the spectrum shifts to higher frequencies (p<0.001). Furthermore, no significant difference was found between elderly subjects and young subjects (p=0.457).

[0059] Next, the results of an experiment conducted on another task will be explained. The second task involved maintaining the posture of the hand and forearm for one minute under three conditions: rest, outer weight, and outer finger up. Figures 17(a) to (c) are diagrams explaining the conditions of rest, outer weight, and outer finger up. Outer weight refers to applying weight to the back of the hand with the palm facing down.

[0060] The measurements were performed using a commercially available sensor and a high-sensitivity sensor. Figures 18(a) to (c) show the spectra obtained for a PD patient and a healthy elderly person under three conditions: rest, lateral weight bearing, and lateral finger lifting.

[0061] 19(a) to (c) are diagrams showing spectra obtained by a high-sensitivity sensor for a PD patient and a healthy elderly person under three conditions: rest, lateral weight application, and lateral finger lifting.

[0062] 20(a) and (b) show the spectra obtained by a high-sensitivity sensor for a PD patient and a healthy elderly person.

[0063] The second problem also revealed significant differences in the high-frequency spectrum between PD patients and elderly people.

[0064] III Experiment 3 The task in Experiment 3 involved participants using their dominant hand to hold two conditions, a pinch condition and a rest condition, for 40 seconds each, with three trials per condition.

[0065] Figures 21(a) and (b) show the pinch condition and the rest condition. In the pinch condition in Figure 21(a), a clothespin is pinched between the middle finger, index finger, and thumb. A rectangular parallelepiped with a width of 20 mm is clamped on the handle side of the clothespin to restrict the range of motion. In the rest condition in Figure 21(b), the palm is kept naturally open.

[0066] Figure 22 shows a summary of the participants in Experiment 3. PD patients were measured in the "On state" when anti-Parkinson's drugs were effective. Measurements were taken on the hand on the side where tremor was not present (symptoms subdominant side). Figure 23 shows the attachment position of the sensor. The sensor was attached to the abdominal muscle of the superficial head of the flexor pollicis brevis.

[0067] Next, the analysis method in Experiment 3 will be explained. Figure 24 is a diagram explaining the analysis method. Welch's power spectral density (PSD) was calculated from the acceleration data. Then, the frequency band where tremors were observed was divided into a low frequency band (5-15 Hz) and a high frequency band (15-35 Hz), and the muscle sound spectrum energy (area) of each band was calculated.

[0068] Figures 25(a) and (b) show the average spectra of healthy elderly people and PD patients. It can be seen that faint muscle sounds in the high frequency band that were hidden under the noise floor of the commercially available sensor were observed. In addition, faint muscle sounds were observed in both the PD patient and healthy elderly groups. It can also be seen that the muscle sound spectrum in all frequency bands is larger in the pinch condition than in the rest condition for both the PD patients and the healthy elderly people.

[0069] 26(a) and (b) show the energy of the low and high frequency bands. In the figures, ** indicates that the p-value is smaller than 0.01, and *** indicates that the p-value is smaller than 0.001. It can be seen that in both the low and high frequency bands, muscle sounds increase under flexion conditions when exerting force compared to under resting conditions in both PD patients and healthy elderly people.

[0070] Figure 27 shows the spectra of PD patients and healthy elderly subjects measured under resting conditions. Under resting conditions, the spectra of the PD patient group are smaller than those of the healthy elderly subjects in both the low and high frequency bands.

[0071] 28(a) and (b) show the energy in the low and high frequency bands. When comparing the resting conditions, no significant difference was found between healthy elderly subjects and PD patients.

[0072] Figure 29 shows the spectra of PD patients and healthy elderly subjects measured under pinch conditions. Under pinch conditions, the spectra of both groups are comparable in the low frequency band, but the muscle sounds of PD patients are lower than those of healthy elderly subjects in the high frequency band.

[0073] 30(a) and (b) show the energy in the low and high frequency bands. When comparing pinch conditions, no significant difference was found between healthy elderly subjects and PD patients in the low frequency band, but the energy in the PD patients was significantly smaller than that in the healthy elderly subjects in the high frequency band (p<0.05).

[0074] The experimental results show that by measuring the high-frequency components of muscle sound signals under pinch conditions, it is possible to significantly distinguish between PD patients and healthy elderly people.

[0075] The present invention has been described above based on an embodiment. This embodiment is merely an example, and it will be understood by those skilled in the art that various modifications are possible in the combination of each component and each processing step, and that such modifications are also within the scope of the present invention. Such modifications will be described below.

[0076] Although the experiments described in this specification are mainly conducted on PD patients, the scope of application of the diagnostic device is not limited thereto, and the device can be used to diagnose various diseases. The diagnostic device may simply output a value that has a correlation with a high-frequency muscle sound signal. For example, by comparing this value before and after administering a certain drug to a PD patient, it is possible to evaluate the efficacy of the drug.

[0077] In the embodiment, it has been shown that the high frequency band spectrum energy and the intermediate frequency are correlated with the presence or absence of a neurodegenerative disease, but the information used for diagnosis is not limited to these. For example, diagnosis may be based on the shape of the spectrum, or, if a clear peak exists, diagnosis may be based on the frequency of the peak.

[0078] In the embodiment, PD diagnosis has been described, but the present invention can also be applied to the diagnosis of Parkinson's disease syndromes (normal pressure hydrocephalus, chronic subdural hematoma, progressive supranuclear palsy, multiple system atrophy, corticobasal degeneration, multiple chronic infarctions, etc.) that have symptoms similar to those of PD but are not PD. In addition, the present invention can also be applied to Alzheimer's disease, amyotrophic lateral sclerosis, frailty, etc.

[0079] In the embodiment, the analysis uses the spectral energy and intermediate frequency, but the statistical processing method is not limited to this, and the acceleration signal itself may be analyzed. Alternatively, the amplitude of the acceleration signal and its root mean square (RMS) may be analyzed, or its peak value may be used. Alternatively, other power spectrum features, such as mean power frequency (MPF), may be used. Alternatively, a time factor such as the reaction time of the muscle sound may be used. Furthermore, the muscle sounds of multiple muscles may be analyzed by independent component analysis, principal component analysis, factor analysis, non-negative matrix factorization (NMF), or the like.

[0080] In the embodiment, the force generated by the subject under non-resting conditions is set to 20% or less of its maximum output, but even if it is increased to about 50%, it is possible to measure a significant muscle sound signal for tremors and shaking. The output under non-resting conditions can be determined according to the type of muscle being measured, and the greater the muscle strength, the smaller the ratio to the maximum output should be.

[0081] In the experiment, by comparing a low-noise sensor with a noise density of 100 μG / √Hz with a commercially available sensor with a noise density of 400 μG / √Hz, it was found that the former can detect muscle sound signals in the high frequency band, while the latter cannot detect muscle sound signals in the high frequency band. In other words, the present technology can be said to be capable of using an acceleration sensor with a noise density lower than a predetermined value, and the predetermined value can be said to be between 100 μG / √Hz and 400 μG / √Hz. From this perspective, an aspect of the present disclosure can be understood as follows. A diagnostic device of one embodiment is equipped with an acceleration sensor that can be attached to a subject and has a noise density lower than a predetermined value between 100 μG / √Hz and 400 μG / √Hz, and a processing device that diagnoses the subject based on high-frequency components higher than a predetermined frequency contained in a muscle sound signal obtained from the output of the acceleration sensor.

[0082] The present invention has been described using specific terms based on the embodiments, but the embodiments merely illustrate the principles and applications of the present invention, and many modifications and changes in arrangement are permitted to the embodiments without departing from the spirit of the present invention as defined in the claims. [Industrial Applicability]

[0083] The present invention relates to diagnostic devices. [Explanation of symbols]

[0084] 100 Diagnostic Equipment 102 Acceleration Sensor 104 Processing equipment

Claims

1. An acceleration sensor that can be attached to a subject and has a noise density of less than 100 μG / √Hz; a processing device for diagnosing the subject based on high-frequency components higher than a predetermined frequency included in a muscle sound signal obtained from an output of the acceleration sensor and low-frequency components lower than the predetermined frequency included in the muscle sound signal; A diagnostic device comprising:

2. 2. The diagnostic device according to claim 1, wherein the predetermined frequency is set between 10 and 30 Hz.

3. 3. The diagnostic device according to claim 1, wherein the muscle sound signal is measured under a non-resting condition in which the subject generates force.

4. 4. The diagnostic device according to claim 3, wherein the force generated under the non-resting condition is 50% or less of the maximum output.

5. 5. The diagnostic device according to claim 3, wherein the non-resting condition is a state in which the subject is holding an object with his / her fingers.

6. 6. The diagnostic device according to claim 3, wherein the non-resting condition is a state in which the subject performs lateral finger lift.

7. 7. The diagnostic device according to claim 3, wherein the non-resting condition is a state in which an external weight is applied to the subject.

8. 8. The diagnostic device according to claim 1, wherein the processor calculates the spectral energy of the high frequency components.

9. 9. The diagnostic device according to claim 1, wherein the processor calculates the intermediate frequency of the high frequency components.

10. 10. The diagnostic device according to claim 1, wherein the muscle sound signal is measured under a resting condition in which the subject is not generating force.

11. 11. The diagnostic device according to claim 1, wherein the acceleration sensor is attached to a superficial muscle or a flexible muscle group of the forearm on the dominant hand side of the subject.

12. 11. The diagnostic device according to claim 1, wherein the acceleration sensor is attached to superficial muscles of the palm of the subject's dominant hand.

Citation Information

Patent Citations

  • Human body muscle sound signal analysis system

    CN108903948A

  • Measuring system for noninvasively evaluating damaged state of human muscle

    JP2009297206A

  • Determination auxiliary method for presence or absence of excitation-contraction coupling failure

    JP2015066401A

  • Motor function evaluation method of central nervous system disease patient

    JP2015217282A

  • Device and Method for Suppressing Vibration

    JP2016511651A