DE102007063007A1 - Method and device for sleep diagnosis and therapy monitoring - Google Patents

Abstract

The invention relates to a method with apparatus for sleep medical pre-diagnosis and therapy support. Claimed are a method and a device for sleep phase detection, in which the sleep phases are broken down into three sleep phases low, light and non-light low or two deep and non-deep sleep phases based on breathing noise analysis. Respiratory noises are detected by means of a microphone system and used for sleep phase detection according to a "neural spectral mean and RMS method".

Description

THE INVENTION CONCERNS A METHOD FOR DIAGNOSIS OF SLEEP DISORDERS IN THE HOME USE, BASED ON THE ANALYSIS SLEEP RELATED Breathing sounds of the patients.

State of the art

Of the Sleep has always aroused people's interest. nowadays A person spends about one third of his life in sleep. Also the remaining two-thirds are strong on sleep quality embossed.

Therefore especially in general and sleep medicine many Illnesses through an everyday observation of sleep quality not only recognized early or better combated, but also avoided.

one the focus of sleep quality assessment is in addition to the sleep-related history (sleep-related questionnaire), the detection and analysis of sleep stages. For many sleep disorders the duration of sleep stages 3 and 4 (deep sleep) is reduced or the deep sleep phases are not present at all.

A complex diagnosis of sleep disorders is currently only by a so-called electrophysiological sleep polygraphy, used exclusively in sleep laboratories of sleep medicine Centers can be carried out. Here is the patient connected to a variety of sensors and possibly existing Sleep disorders examined. To determine the sleep stages are mainly the waveforms electroencephalogram (EEG), electromyogram (EMG) and of the electrooculogram (EOG) and according to the rules evaluated by the righteous and the bald.

There Such a diagnosis in a sleep laboratory very expensive and costly is generally used for preoperiagnosis at home mobile so-called. Screening devices used. These works generally based on a smaller number of sensors.

Out DE4114357C2 a device for the automatic detection of apnea is known. In this device, however, several sensors, including breathing belts in the chest, must be connected to the patient. This is often perceived as disturbing, since the sensors have to be worn during sleep. In addition, this device needs a separate reading and writing device for recording the data, which must be evaluated by a doctor afterwards.

Out DE69930720T2 Another monitoring device for the detection of sleep apnea is known. This device is incorporated in a small plastic unit which is positioned under the patient's nose. The device measures the air flows during inhalation and exhalation and records the respiration or respiratory failure during sleep. There is a much simplified form of diagnosis based on the recorded data. A doctor can not rely on the recorded data to make his own diagnosis, so the method is only suitable for patients with little chance of real pathology.

Already For such a pre-diagnosis at home is a doctor's visit necessary, unless the patient invests himself in the purchase a corresponding device. For a doctor's visit However, it is already necessary that the patient sleep disorders recognized as an illness. This costs experience, some overcoming, in particular, since in the case of sleep disorders the patient can not observe his own illness, but mostly depends on statements of the bed partner.

task

The basic idea of the new method for sleep phase detection developed in the context of this work is based on observations made by the author, which were supported by physicians and researchers of the research group:
If one observes a sleeping person, then one can estimate on the basis of the respiratory sounds or their variability, whether one sleeps quietly or restlessly. Assuming a more even breathing in deep sleep phases (sleep stages 3 and 4), an assessment of whether someone is in such a sleep, based on a breath sound analysis seems possible.

Even though this hypothesis after an intensive literature search neither proved is still refuted, the author decided, on the basis of the above motivating aspects, the context in the context of a To investigate research work.

The technical objective and the object according to the invention therefore consists of developing a practical method that can be easily operated by each patient with the associated device system or corresponding software that analyzes and displays the quality of sleep and thus provides the patient with rapid information possible sleep disturbances or lack of deep sleep phases granted. In particular, attention should be paid to ease of use and low sensor, preferably only in the form of a microphone.

These Task is inventively using the in the claims characterized working method solved with associated software. The centerpiece The invention consists in that the evaluation of the quality of sleep made by the software using artificial neural networks and the evaluation only on the basis of a breath sound recording is carried out.

The Classification is based both on the evaluation of the RMS values (Root Mean Squares) over a certain signal duration, as also on the evaluation of spectral components of the respiratory sound signal.

Either RMS values as well as spectral components become during the recording is calculated and stored online by the software. The classification with the help of a neural network then finds following, with the previously described stored Values are loaded and, as in the claims under point 1, are processed.

Suggested solution and embodiment

Method for measured value acquisition and extraction the characteristic features:

The breathing sounds are recorded with the aid of an airborne microphone or a structure-borne sound microphone. The airborne microphone is attached to a headband and is attached to the patient's head (see 2 ). The airborne microphone can also be mounted on a stand at a distance of 0-50 cm from the patient's head (see 3 ). For the use of the Köperschallmikrofons this is preferably on the upper lip (see 4 ) or attached to the trachea. With a simple sound blaster then breath sounds with a sampling frequency of up to 11.025 kHz and z. B. 16 bit resolution sampled. In a subsequent first processing step, the data is compressed using a bandpass filter to a frequency range between 0 and 5500 Hz, since no higher frequencies are contained in the breath sounds.

To This preprocessing is z. B. a time-dependent and Sectional spectral analysis performed. For For each short-term spectrum, 256 samples were used by way of example (about 10 ms because of short-term stationarity of breath sounds). Around To accelerate the subsequent pattern recognition will be a data reduction by a factor of up to 16 using a filter bank approximation carried out. The spectrum is in z. B. divided 16 areas. The mean values of the respective areas then form the filter bank coefficients, which are then combined to form a new feature vector. By The filter bank analysis produces a compact approximation of the power density spectrum, the describes the signal as a sequence of feature vectors. Consequently one obtains the spectral mean values of the respiratory sounds.

Transfer on this case of investigation, is determined by means of the correlation coefficient Determined if there is a relationship between the feature matrix and of the polysomnographic sleep stage vector.

The Analysis of the frequency banks showed that the four lower correlated better. For this reason, preferably the 4 lower frequency banks for sleep phase detection included.

exemplary Filter banks: Bank1 (0-124 Hz), Bank2 (124-264 Hz), Bank3 (264-421 Hz), Bank4 (421-597 Hz).

Parallel for determining the spectral values of the respiratory sounds in the frequency domain the "Root Mean Square" of the signal. The Root Mean Square can be translated into German as RMS value of the signal.

The RMS of a signal x out of n sample points is calculated as follows:

To The present method has preferably been selected from the patient Respiratory sound data of an RMS vector, each with n = 30 × 11025 = 330750 points (N.B. For segment duration of, for example, 30 seconds, Sampling frequency z. 11025 points / second, mono).

Combination of the spectral mean values with the RMS:

The Feature extraction consists of 2 levels. In the first stage will be the data normalized.

In the second stage, a 5 × N feature matrix is preferably calculated from the individual breath sounds. N denotes the single vector of the segment duration of 30 seconds per vector (preferably). This corresponds to z. Eg 2 points per minute, ie 120 points per hour, or 1200 points for a night recording of 10 hours.

at The 5 features per vector (every 30 seconds) identify four the amplitude of the 4 lower frequency ranges (i.e., for example 0-124 Hz, 124-264 Hz, 264-421 Hz, 421-597 Hz) of the spectral mean values, and one of the RMS amplitudes.

Segmentation process:

The investigations carried out in the present work were compared with the polysomnogram derived sleep stages carried out.

Therefore Breathing sounds of several patients during an entire night in an accredited sleep laboratory after Recording method described above by means of a microphone system and a PC recorded.

The For example, recordings are made using a program written in-house carried out. The program is characterized mainly by its high stability and robustness when dealing with large Data volumes (here up to more than 650 MB for more than 10 Hours of recording). The program has over some well-useful analysis and signal processing functions, which for a quick consideration and investigation of Data are very helpful.

The Polysomnogram recorded in the sleep laboratory usually becomes the next day automatically from the system analysis program evaluated. In the evaluation, sleep phases and Apnoeauftritte calculated as a function of time by the system.

Of the competent doctor will later revise the automatic Evaluations. He checks, among other things the automatic determination by the analysis program of the Sleep stages, and corrects them if necessary. A revision by the doctor can usually be between 45 minutes up to about Take 2 hours. Because many patients in the sleep laboratory due to sleep disorders not always have deep sleep phases, it is important in the investigation the sleep phases to involve both ill and healthy patients.

at For example, the developed new method became noise signals processed in blocks of 30 seconds, since those from the, used in the test, polysomnogram calculated sleep stages were always set over a period of 30 seconds.

Design of the neural recognizer:

First becomes a multi-layered preferably forward-looking neural network designed.

After that will be a training process with appropriate algorithm (here preferably backpropagation). In the definition of the Training process should be network initialization function, segmented Test and validation patterns, number of test cycles, termination criteria and training duration are entered. Subsequently, will the training of the neural network started and finally a network is generated. The trained network will then turn on the system is connected modularly.

Description of the system

The System consists of a microphone system, a PC with AD-DA converter card (Sound blaster card) and a monitoring tool.

5 shows an exemplary schematic representation of the processes in the software. First, the recorded breath sound data is digitized. Subsequently, the data is filtered and normalized in the signal preprocessing. Next, feature extraction and handover of the data to the neural network or hidden Markov model network occurs. In this network, the detection of the sleep phases takes place. The network is trained with a teaching database containing recorded data from clinical sleep laboratories. These data are passed to another neural network or hidden Markov model network, where the detection of sleep disorders and sleep phase patterns takes place. In the process, the network is trained with a further instruction database containing typical sleep disorder pictures from clinical sleep laboratories. If sleep disorders are detected, an advance diagnosis with therapeutic suggestions is made using a knowledge-based automaton.

QUOTES INCLUDE IN THE DESCRIPTION

This list The documents listed by the applicant have been automated generated and is solely for better information recorded by the reader. The list is not part of the German Patent or utility model application. The DPMA takes over no liability for any errors or omissions.

Cited patent literature

• - DE 4114357 C2 [0007]
• - DE 69930720 T2 [0008]

Claims (9)

A method for detection and visualization of the sleep profile using the analysis of sleep-related breath sounds, characterized in that for the detection of sleep-related breath sounds only a nocturnal breath sound recording, preferably using a mounted on a headband above the nose airborne microphone, takes place that of the microphone recorded signals are preferably supplied to a preamplifier, that the amplified signals, digitized in a sound card, the program are provided. The method of claim 1, wherein the digitized ones Signals are normalized and filtered. The method of claim 1 and 2, wherein from the so processed signals in segments over a certain Time range, preferably an epoch, the RMS values (Root Mean Squares) are calculated and stored. The method of claim 1-3, wherein from the filtered and normalized signals spectral components are calculated and over a certain period of time, also preferably an epoch, averaged and stored. The method of claim 1-4, wherein the stored RMS values and the stored spectral components of the lower frequency ranges of the entire recording loaded in the program and put together in a matrix. The method of claims 1-5, wherein the combined Values are temporally smoothed. The method of claims 1-6, wherein the values to a mean of zero and a standard deviation of one be standardized. The method of claim 1-7, wherein the values for the classification of the sleep stages of a previously trained, artificial Neural network fed and evaluated by this. Software program for the execution of the project according to claim 1-8.