Method, device and device device for classifying brain pressure and corresponding computer program and computer-readable medium

Abstract

A computer-implemented method of classifying brain pressure includes receiving data including time-of-flight values based on time-of-flight values of one or more ultrasound signals measured at consecutive points in time through a cross-section of a skull; classifying a characteristic of the brain pressure by means of one or more characteristics of a representative curve of the temporal process of the time-of-flight values to obtain a classified characteristic of the brain pressure; and outputting the classification characteristics of the brain pressure.

Description

Technical Field

[0001] The present invention relates to a computer-implemented method for classifying intracerebral pressure, a device for classifying intracerebral pressure, a device arrangement for classifying intracerebral pressure, and a corresponding computer program and computer-readable medium. Background Art

[0002] The brain is continuously supplied with oxygen and nutrients via blood. If blood flow through the brain decreases, damage to the brain, particularly neuronal cell death, can occur. Brain pressure, the pressure prevailing inside the skull, including the cerebrospinal ventricles (also known as intracranial pressure, ICP), is important in this context because it affects blood flow through the brain and, therefore, the supply of oxygen and nutrients. Under "normal" conditions, brain pressure is permanently below approximately 10 mmHg, which does not result in any restriction of blood flow through the brain because the brain's tissues have a natural elasticity to absorb blood. If the brain is damaged by injury or bleeding, it typically swells. However, since the brain is surrounded by a hard bony shell that protects it from damage, but due to its rigidity, it can only yield slightly during swelling, so during swelling, brain pressure can increase, for example, permanently above 15 mmHg. If brain pressure increases, the elasticity of the brain tissue decreases, meaning that the tissues in the brain can expand less during systole and diastole, and less oxygen-rich blood reaches the brain. As a result, blood flow through the brain decreases. If, in extreme cases, brain pressure exceeds a limit, i.e., becomes abnormal, blood flow to the brain is poor due to the lack of elasticity of the tissue, and nerve cells in the brain can die.

[0003] Therefore, it is important to be able to determine whether the brain pressure is in the normal range or in the abnormal range, that is, it is elastic enough to absorb oxygen-rich blood so that it can respond accordingly.

[0004] According to current prevailing standards, intracranial pressure is determined invasively by a pressure measuring probe in the cranial cavity (the so-called gold standard for intracranial pressure measurement). In this case, the pressure measuring probe is usually inserted into the cranial cavity through the external bony shell of the skull via a drainage catheter during surgery, preferably into the lateral ventricles of the brain. By inserting the pressure measuring probe directly into the brain, the current prevailing pressure in the skull can be measured directly by the pressure measuring probe. Intracranial pressure and therefore the elasticity of the brain, that is, the brain's ability to absorb oxygen-rich blood, can be determined from the changes in pressure measured in the skull during one or more diastolic and systolic phases. In addition, a possible cerebrospinal fluid drain can be placed in the skull while the intracranial pressure measurement is being performed, which can be used as an effective intracranial pressure reduction therapy. This invasive method is associated with considerable stress on the patient during the operation, can lead to complications, and requires appropriate aftercare. In addition, this invasive method places high demands on the equipment and personnel in the medical facility.

[0005] Previous approaches to improving the gold standard for intracranial pressure measurement have primarily aimed to improve the sensitivity and size of pressure measuring probes, or to improve their access. Specifically, by reducing the size of pressure measuring probes and inserting cerebrospinal fluid drains into the skull during surgical interventions, the surgical effort required to detect and treat elevated intracranial pressure has been reduced. However, as a result, the severity of the intervention in patients has been only marginally reduced.

[0006] WO 2020 / 219773 A1 also discloses a non-invasive method in which a transmitter and receiver are attached to opposite sides of the skull. The distance and travel time of a sound signal between the transmitter and receiver are determined. ICP is estimated based on the correlation between distance and travel time. However, this method cannot reliably determine ICP. Summary of the Invention

[0007] Against this background, the object of the present invention is to provide a particularly simple method with which elevated intracerebral pressure can be determined.

[0008] This object is achieved by a computer-implemented method for classifying intracranial pressure according to the main claim, wherein the method comprises: receiving data including time-of-flight values based on time-of-flight values of one or more ultrasound signals through a cross section of the skull measured at consecutive time points; classifying a characteristic of the intracranial pressure by means of one or more features of a representative curve of the time course of the time-of-flight values to obtain a classification characteristic of the intracranial pressure; and outputting the classification characteristic of the intracranial pressure. All method steps of the computer-implemented method described herein are performed by one or more computers.

[0009] The time-of-flight values of one or more ultrasound signals can be measured by means of an ultrasound probe before receiving the data, so that the patient's presence is not required for the method steps performed by the computer. The data can be supplied to the computer, for example, by means of a data carrier (e.g., a USB stick) or can be sent via a wireless or wired line. The data is processed independently without physical contact with the patient.

[0010] The present invention is based on the inventors' discovery that a non-invasive assessment of intracranial pressure can be achieved using the method according to the present invention. Based on this method, intracranial pressure can be determined non-invasively, in particular, whether it is normal or abnormal. This means, in particular, that surgical intervention is not required to determine whether intracranial pressure is normal or abnormal. This reduces stress on the patient, significantly reduces medical personnel and logistical effort, and ultimately reduces costs.

[0011] In this context, the present invention is based, inter alia, on the discovery by the inventors that changes in brain pressure conditions within the skull are directly related to changes in the time of flight of one or more ultrasound signals through a cross section of the skull (this will also be combined with Figure 2 explained in more detail).

[0012] The heartbeat induces a pulse wave, which pushes blood into the brain's blood vessels. The pulse wave induced by the heartbeat causes a volume change or dilation of the brain's blood vessels, which receive the blood being pushed into the brain. This volume change causes a pressure change in the skull, which has been measured directly in the brain using the gold standard, measuring waves.

[0013] The inventors have now realised that the measured time-of-flight values of one or more ultrasound signals through a cross section of the skull depend on, and are therefore related to, the superposition of all tissue layers and fluids located within the sound field.

[0014] Thus, the time-of-flight values of one or more ultrasound signals through a cross-section of the skull measured at consecutive time points detect time-of-flight differences caused by, for example, pressure-induced volume changes and pressure-induced compression of brain components (e.g., blood, cerebrospinal fluid, and tissue).

[0015] Thus, a curve of the time course of time-of-flight values measured at consecutive time points maps changes in volume and density of brain components, and thus changes in brain pressure, during one or more pulse waves.

[0016] Based on the characteristics of the curve reflecting the time course of the time-of-flight values of the pressure-induced volume changes and the pressure-induced compression of brain components during one or more heartbeats, the brain pressure can thus be determined or classified, in particular properties, such as whether it is normal or abnormal, can be classified non-invasively.

[0017] In particular, the method according to the invention allows the conditions within the skull, in particular the brain pressure, to be determined without distortions caused by the heartbeat. Therefore, no distance measurement between the probes is required.

[0018] In the learning phase, the underlying algorithm for classifying the properties of brain pressure is trained so that individual features of the representative curve can be assigned to properties of brain pressure.

[0019] During the learning phase, the time-of-flight values are measured by the skull in a time-dependent manner, and in parallel thereto, direct intracranial pressure measurements are determined for a plurality of patients with normal and abnormal intracranial pressures according to a proven gold standard. The following physiological inclusion criteria are considered when selecting patients: Pathology: Severe traumatic brain injury (Glasgow Coma Scale < 10); Heart rate range: 50 min⁻¹ < HF < 120 min⁻¹; Pulse pressure (difference between systolic and diastolic blood pressures) < 90 mmHg; Systolic blood pressure < 190 mmHg; Diastolic blood pressure > 30 mmHg.

[0020] Compared with the direct intracranial pressure measurements recorded in parallel with the time-of-flight values on the same patient, based on the characteristics of a plurality of representative curves from the time course of the time-of-flight values, it is thus possible to determine whether the characteristics in the curves are characteristic of a specific intracranial pressure or whether the characteristics reliably separate the data from abnormal and normal intracranial pressures. In the training data set, the characteristics that show the greatest importance for the separation between elevated ICP (e.g., ≈15 mmHg) and normal ICP (e.g., ≈10 mmHg) are identified.

[0021] During the learning phase, various characteristics on multiple curves are considered. The characteristics considered can be divided into four groups: Statistical characteristics (e.g., mean, standard deviation, skewness of the distribution); Characteristics from data aggregation (e.g., minimum time-of-flight value, maximum time-of-flight value, average amplitude); Characteristics from curve discussion (e.g., peak width, spline, extreme values, inflection points); Frequency domain characteristics from a common expert library of time-related signals (such as tsfresh).

[0022] During the learning phase, it is thus determined which of the inspection characteristics in the representative curves from the corresponding time course are characteristic of a specific intracranial pressure or normal or abnormal intracranial pressure, which is determined by the gold standard.

[0023] By determining these characteristics of the curve of the time course of the time-of-flight values measured through the cross-section of the skull, the characteristics characteristic of the intracranial pressure can be assigned. As a result, the intracranial pressure can be classified non-invasively from the curve of the time course of the time-of-flight values of one or more ultrasonic signals measured through the cross-section of the skull.

[0024] During the time course, at least one time-of-flight value is assigned to one of the consecutive time points. In other words, the time course can be described as a curve in a two-dimensional coordinate system, where the time-of-flight values are on one axis and the corresponding consecutive time points are on the complementary axis.

[0025] In one aspect, the time course of the time-of-flight values is related to the time course of the intracranial pressure within the skull.

[0026] Since the inventors have recognized that the time course of the time-of-flight values measured at consecutive time points is correlated with changes in the volume and density of brain components and therefore directly maps the time course of pressure in the skull, the characteristics of the brain pressure can be non-invasively classified based on the curve of the time course of the time-of-flight values, and in particular, it can be derived whether the brain pressure is normal or abnormal.

[0027] In one aspect, the time course of the time-of-flight values includes a plurality of peaks, each peak being associated with an increase and a decrease in brain pressure during systole and diastole.

[0028] Peaks are formed by increasing or extending the time-of-flight values of one or more ultrasound signals through a cross-section of the skull, measured at successive time points. Consequently, peaks are associated with increases in brain pressure during a heartbeat. Longer time-of-flight values are formed by increases in brain pressure during systole, i.e., increases in pressure in the skull. Shorter time-of-flight values are formed by decreases in brain pressure during diastole, i.e., decreases in pressure in the skull. Thus, the rising region of the peak corresponds to the systole of the blood vessels, in which blood is being pushed into the brain, and the falling region of the peak corresponds to the diastole, in which blood is flowing out of the brain.

[0029] The time course of the time-of-flight values thus maps multiple systoles and diastole periods, i.e., multiple consecutive pulse waves, during which blood is pumped into the brain's blood vessels. Therefore, the multiple pulse waves can be distinguished by the time course of the time-of-flight values. As a result, intracranial pressure can be reliably classified based on the time course curve.

[0030] In one aspect, the change between the time-of-flight values at at least two consecutive time points during the time course is directly related to the change in brain pressure at at least two consecutive time points.

[0031] Since the inventors have realized that changes between time-of-flight values of one or more ultrasound signals are directly related to changes in brain pressure at at least two consecutive time points, brain pressure can be non-invasively classified by the method according to the invention.

[0032] In one aspect, the classification characteristics of intracranial pressure indicate the presence of normal intracranial pressure, in particular an intracranial pressure of S1 or lower, or the presence of abnormal intracranial pressure, in particular an intracranial pressure of S2 or higher, where S2 ≥ S1. The method according to the present invention can distinguish normal intracranial pressure from abnormal intracranial pressure with the help of different thresholds S1, S2 and is therefore not limited to specific values of S1 and S2. The thresholds S1, S2 are determined with the help of a patient population, where S1 ≤ S2 is applicable. These thresholds S1, S2 are each typically between 5 mmHg and 20 mmHg, where thresholds S1 = 10 mmHg and S2 = 15 mmHg are particularly preferred to avoid classifying actual abnormal intracranial pressure as normal.

[0033] Thus, intracranial pressure can be classified as normal or abnormal. Thus, the method can output a corresponding finding of normal or abnormal intracranial pressure; for example, the classification can be displayed on a display or transmitted to another entity in the form of data. Based on the finding, a diagnosis of a disease present in the patient (or absence of the patient) can be performed by medical personnel.

[0034] At an intracerebral pressure preferably below about 10 mmHg, there is sufficient elasticity in the brain to absorb oxygen. At an intracerebral pressure preferably of 15 mmHg or higher, the intracerebral pressure is assumed to be in the elevated range. In case of doubt (e.g., due to measurement inaccuracies), if, for example, the intracerebral pressure is between 10 mmHg and 15 mmHg, the intracerebral pressure is inferred to be elevated.

[0035] In one aspect, the representative curve includes at least one characteristic peak.

[0036] Therefore, a characteristic peak is a characteristic of the time-of-flight variations of multiple pulse waves in the brain. For example, a characteristic peak can be formed by taking the median of the time-of-flight values that form multiple peaks over the time course of the time-of-flight values. In other words, the characteristic peak in the representative curve maps the characteristic increases and decreases in intracranial pressure over multiple systoles and diastole periods. Thus, at least one characteristic peak better maps the characteristic behavior of the brain. Consequently, intracranial pressure can be more reliably classified based on the time course of the time-of-flight values.

[0037] In one aspect, the characteristic of brain pressure is determined by a computer by extracting one or more features of the curve of at least one characteristic peak.

[0038] The characteristic peak reflects the flight time variation across multiple pulse waves in the brain, thereby more meaningfully mapping brain behavior. More reliable classification is ensured by extracting one or more features from the curve of at least one characteristic peak.

[0039] In one aspect, a first characteristic of the one or more characteristics is a peak width of at least one characteristic peak at a percentage of a maximum amplitude of the at least one characteristic peak.

[0040] During the learning phase, the inventors discovered that a characteristic that can be assigned to normal or abnormal intracranial pressure is the peak width in the time course of the time-of-flight values. Thus, normal or abnormal intracranial pressure can be determined based on the peak width of at least one characteristic peak at a certain percentage of the maximum amplitude.

[0041] In one aspect, a second feature of the plurality of features is a similarity of a curve shape of at least one characteristic peak to a synthetic function, in particular a continuous wavelet transform.

[0042] Furthermore, during the learning phase, the inventors have discovered that another feature reflecting normal or abnormal brain pressure is the similarity of the curve shape of at least one characteristic peak to the synthetic function.

[0043] During the learning phase, a plurality of characteristic peaks are convolved or superimposed with a plurality of defined synthesis functions, wherein various synthesis functions are tried. The similarity of the curve shapes can be determined based on the mathematical convolution of the curve shapes of the synthesis functions. Various wavelet functions are used as synthesis functions, wherein the properties of the wavelet functions are varied by scaling factors (e.g. their coefficients), and the corresponding functions are compressed and stretched in this way. Examples of wavelet functions used are: db4, db16, haar, coif, sym4, sym8, bior1.3 or bior3.1 functions. A description of the wavelet functions can be retrieved, for example, at https: / / de.mathworks.com / help / wavelet / ref / cweighthtml or https: / / en.wikipedia.org / wiki / Ricker_wavelet

[0044] In this case, the inventors have discovered, by means of a comparison with corresponding brain pressure measurements according to a gold standard, that certain scaling coefficients of the wavelet function reflect normal or abnormal brain pressure.

[0045] The characteristics of the brain pressure can be reliably classified by means of the second feature extracted from the representative peak. A particular reliability of the classification can be achieved by combining the first feature and the second feature determined based on at least one characteristic peak.

[0046] In one aspect, the at least one characteristic peak is for a plurality of peaks in at least one time period of the time course of the time-of-flight values.

[0047] According to this aspect, a time period based on the time course of the time-of-flight values, ie a time segment, forms at least one characteristic peak. In other words, a segment can be understood as a time segment containing time-of-flight values assigned to a subset of time points of consecutive time points.

[0048] For example, this method can be used to select segments with the best time-of-flight data quality. Alternatively, segments with poor time-of-flight data quality can be excluded. This results in a more reliable classification of brain pressure characteristics. Furthermore, the required computing power can be reduced by reducing the amount of data required for evaluation.

[0049] In one aspect, at least one characteristic peak is determined by the following steps: extracting multiple individual peaks from at least one time period of the time course of flight time values; normalizing the time points of the flight time values in the individual peaks; grouping the individual peaks by applying a similarity criterion to the individual peaks; identifying the group with the largest number of peaks, and forming at least one characteristic peak from the peaks in the identified peak group.

[0050] Thus, the at least one characteristic peak is formed by the time-of-flight values of peaks from the group of most peaks that meet the similarity criterion.

[0051] Similarity criteria may include, for example, forming silhouette coefficients for individual peaks.

[0052] As a result, at least one characteristic peak maps a plurality of peaks with a relatively meaningful data quality. As a result, the classification of the characteristics of the brain pressure with the aid of the measured time-of-flight values becomes more reliable.

[0053] In one aspect, forming at least one characteristic peak comprises forming, for each time point, a median of the time-of-flight values of the peaks from the identified group.

[0054] Thus, the time-of-flight value of at least one characteristic peak is formed from the median of the time-of-flight values of the plurality of peaks. Thus, the at least one characteristic peak maps the plurality of peaks. As a result, the classification of the characteristics of the brain pressure using the measured time-of-flight values becomes more reliable.

[0055] In one aspect, at least one segment is determined by extracting one or more time segments from the time course of the flight time values, wherein the segments include multiple peaks in the flight time values; and determining at least one time segment by selecting one or more time segments wherein the peaks meet a quality criterion.

[0056] According to this aspect, segments are selected in which the time-of-flight value curve meets a certain quality standard. Consequently, time periods in which the measured time-of-flight values contain interference, such as due to heart failure, can be excluded and not used for further data processing. By selecting individual segments from the time course of the time-of-flight values for classification based on quality criteria, the classification of intracranial pressure characteristics becomes more reliable. Furthermore, the computational workload during data evaluation can be reduced.

[0057] In one aspect, the quality criterion is formed by an autocorrelation function which assesses the self-similarity of the time course of the time-of-flight values in a segment.

[0058] The autocorrelation function is a measure of the self-similarity of a curve. Since the time course of the time-of-flight values is characterized by rhythmic, repetitive fluctuations with high similarity, the autocorrelation function represents a criterion for the presence of a brain impulse curve.

[0059] According to this aspect, segments in which peaks in the time course of the time-of-flight values curve include interference, such as those corresponding to non-existent heartbeats, can be excluded. As a result, the reliability of the classification of the intracranial pressure characteristics is improved. In one aspect, the quality criterion is formed by comparing the maximum amplitude of the most dominant frequency component in the time course of the time-of-flight values in the segment with the amplitudes of the other frequency components.

[0060] In other words, the quality criterion in this respect can be described as the Peak2mean criterion, which describes the ratio between the amplitude of the impulse and its harmonics in the time course of the time-of-flight value and the remaining frequency components. This is therefore an operation in the frequency domain. The Peak2mean is a measure of the degree to which the brain impulse curve represents the dominant component of the time course and is associated with correspondingly high amplitudes. If the signal-to-noise ratio is too low, the corresponding segment is not used for further analysis.

[0061] According to this aspect, a segment in which a peak in the time course curve of the flight time value includes interference, through which a peak caused by the heartbeat appears as a non-most important component, can be excluded. As a result, the reliability of the classification of the characteristics of the brain pressure is improved.

[0062] In one aspect, the quality criterion is formed by evaluating the frequency in the time course of the time-of-flight values in a segment.

[0063] The quality criterion for this aspect can be specified from the upper band noise criterion describing the higher frequency components (≥15 Hz) that are not generated by the harmonics of the heartbeat. The high component is a sign of increased noise; the respective segments are not used for further analysis. The noise amplitude in this frequency range cannot be attributed to a physiological origin.

[0064] By this aspect, segments in which peaks in the curve of the time course of the time-of-flight values comprise interference caused, for example, by non-physiological noise can be excluded. Furthermore, background frequencies not caused by heartbeats can be excluded.

[0065] These aspects of the quality criteria can be applied individually or in combination for selecting segments. By combining a plurality of quality criteria, segments with particularly meaningful data quality can be selected.

[0066] Thus, for example, segments can be selected which satisfy a quality criterion formed by the autocorrelation function and additionally a quality criterion formed by a comparison of the maximum amplitudes of the most dominant frequency components or a quality criterion formed by an evaluation of the frequency over time.

[0067] In other respects, for example, segments may be selected that meet a quality criterion formed by evaluation of the frequency over time and additionally meet a quality criterion formed by comparison of the maximum amplitudes of the most dominant frequency components.

[0068] In a particularly preferred aspect, segments can be selected that meet a quality criterion formed by the autocorrelation function, a quality criterion formed by the comparison of the maximum amplitudes of the most dominant frequency components, and a quality criterion formed by the evaluation of the frequency over time. This aspect is particularly preferred because segments that meet all three quality criteria can be selected, thereby providing a particularly meaningful data quality for one or more selected segments.

[0069] In one aspect, the method further comprises measuring, at consecutive time points, time-of-flight values of one or more ultrasound signals through a cross section of the skull by means of at least two probes positioned on opposite sides of the skull.

[0070] According to this aspect, the flight time of one or more ultrasound signals may be measured. The time-of-flight value corresponds to the flight time required for the one or more ultrasound signals to pass through the skull between the probes along the cross section.

[0071] To measure the time of flight, at least two probes are positioned on opposite sides of the skull. The at least two probes may include: at least one transmitter configured to output one or more ultrasound signals; and at least one detector configured to receive the one or more ultrasound signals output by the at least one transmitter. The at least one transmitter and the at least one detector may be positioned on different, opposite sides of the skull such that the acoustic field between the probes is substantially parallel to a frontal cross-section of the skull. With the at least one transmitter and the at least one detector, the time of flight required for the one or more ultrasound signals to travel from the at least one transmitter through the cross-section of the skull to the at least one detector can be measured.

[0072] As described above, by measuring the time of flight of one or more ultrasound signals, characteristics of brain pressure can be non-invasively determined and classified because the time of flight of the one or more ultrasound signals images the state of the brain along a cross section in the skull.

[0073] Furthermore, one or more of the ultrasound signals may be longitudinal waves. The inventors have found that this type of wave is particularly suitable for determining and classifying brain pressure.

[0074] The probe includes at least one transmitter for outputting one or more ultrasound signals and at least one corresponding detector for detecting the outputted one or more ultrasound signals, wherein the at least one transmitter and the at least one detector are positioned on opposite sides of the skull.

[0075] With this positioning, the time of flight through a cross section along the skull can be measured.

[0076] Both the emitter and detector are located in the area above the external auditory canal.

[0077] The inventors have found that positioning the emitter and detector on opposite areas of the skull above the ear canal is particularly advantageously suitable for imaging internal states of the brain, with the result that properties of the brain can be classified particularly reliably.

[0078] The flight time values of one or more ultrasonic signals through the skull can be measured in the following manner: output multiple wave packets, each wave packet having an n-periodic signal with a constant frequency at a corresponding time point in consecutive time points; and measure the flight time value through the cross section of the skull for each of the n periods of the periodic signal for each wave packet in the multiple wave packets.

[0079] As a result, a particularly advantageous measurement method is disclosed for providing a sufficient number of time-of-flight values for processing, although the present invention is not limited to this particular measurement method. A wave packet is output by a probe, in particular a transmitter, at a certain point in time. The wave packet comprises a periodic signal having n periods, where n is an integer. The wave packet passes through a cross section of the skull and is detected at the probe on opposite sides of the skull. A time-of-flight value is determined for each of the n periods of the wave packet. This means that n time-of-flight values can be assigned to each point in time.

[0080] As a result, the number of available measurement values for subsequent processing of the measured values increases. By increasing the number of measurement values, the signal better describes the temporal course of the internal state of intracranial pressure in the brain. This results in a more reliable assessment of the measured time-of-flight values. Ultimately, the classification of intracranial pressure characteristics is therefore more reliable.

[0081] In one aspect, receiving data having flight time values includes: receiving n flight time curves, wherein in each of the n flight time curves, a flight time value is assigned to one time point in a series of time points; for each of the n flight time curves, subtracting an offset from the flight time value in the flight time curve; and forming an average flight time curve by averaging the respective flight time values of the n flight time curves at one time point in the series of time points.

[0082] On the other hand, in one of the n flight time curves, the flight time value at a time point is formed by the period of the n-period wave packet of one or more ultrasonic signals output or received at the time point.

[0083] Based on the previous measurements, n flight time values are determined at each time point (where n corresponds to the number of periods in the wave packet). As a result, n time distributions of flight time values can be formed, wherein in each of the n time distributions, the flight time values of the same period in the wave packet are assigned to the time points of the wave packet. Each time distribution thus forms a flight time curve. Thus, n flight time curves are formed. An offset is subtracted from each of the n flight time curves. The offset can map multiple factors that are not directly related to cardiac activity. For example, frequency-related interference signals from the measuring device or the patient's breathing can be subtracted from the offset. An average flight time curve is formed by averaging the flight time values at the corresponding time points of the n flight time curves.

[0084] This average time-of-flight curve can be used as a basis for further processing of the data, for example, by forming one or more segments and / or forming average peaks.

[0085] This aspect modifies the measured data so that it is freed from background information not directly related to cardiac activity and, through averaging, a meaningful picture of the brain's changing state along a cross-sectional plane is reproduced. As a result, the morphology of the measured time-of-flight values over time can be better calculated. Classification of the time course using the time-of-flight values thus becomes more reliable.

[0086] In one aspect, each of the n time-of-flight curves has a periodicity given by the heart rate.

[0087] The measurements received at the computer therefore reflect changes in brain pressure induced by the heartbeat.

[0088] In one aspect, one or more time periods are selected from a mean flight time curve for determining at least one segment.

[0089] The mean time-of-flight curve is thus used as the basis for determining the time period of the at least one characteristic peak. As a result, a better and more reliable data quality is achieved for determining the at least one characteristic peak, and as a result, the characterization of the properties of the brain pressure becomes more reliable.

[0090] In one aspect, a bandpass filter is applied to the time-of-flight values in the time-of-flight curve to determine the offset.

[0091] The bandpass filter can cut off low-frequency components (such as trends or respiratory effects). As a result, frequency components not directly related to cardiac activity can be cut off, making it possible to more reliably classify the characteristics of brain pressure.

[0092] Furthermore, the aforementioned object is achieved by a device, in particular a computer, configured to execute the method for classifying intracerebral pressure according to the present invention. The computer may include one or more processors configured to execute the method according to the present invention. Furthermore, the computer may include a storage medium storing instructions which, when executed by the one or more processors, cause the one or more processors to execute the method according to the present invention.

[0093] In addition, the purpose mentioned at the beginning is achieved by an apparatus device, which includes the above-mentioned apparatus and at least two probes, the at least two probes being used to determine the flight time values of one or more ultrasound signals through a cross-section of the skull, wherein the at least two probes include at least one transmitter for outputting one or more ultrasound signals and at least one corresponding detector for detecting the output one or more ultrasound signals, wherein in particular, the at least one transmitter and the at least one detector can be positioned on opposite sides of the skull, in particular in a plane of a frontal cross-section through the skull.

[0094] In one aspect, the at least one emitter and the at least one detector can each be positioned in a region above the external auditory canal.

[0095] In one aspect, at least two probes are configured to output multiple wave packets, each wave packet having an n-periodic signal of a constant frequency at a corresponding time point in consecutive time points, and to detect a time-of-flight value through a cross section of the skull for each of the n periods of the periodic signal at each wave packet in the multiple wave packets.

[0096] Furthermore, the object mentioned at the outset is achieved by a computer program comprising instructions which, when the method according to the invention is executed by a computer, cause the computer to carry out the method according to the invention, and a computer-readable medium on which the computer program is stored. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Other characteristics, features and advantages of the present invention will become apparent from the following description of preferred embodiments of the present invention with reference to the accompanying exemplary drawings, in which:

[0098] Figure 1 By way of example, a measuring device for determining time-of-flight values of one or more ultrasound signals through a skull is shown;

[0099] Figure 2 Validation of the temporal distribution of the measured time-of-flight values is shown;

[0100] Figure 3 Direct measurements and their pre-processing are shown by way of example;

[0101] Figure 4The classifier is shown by way of example;

[0102] Figure 5 A detailed view of a classifier is shown by way of example;

[0103] Figure 6 The determination of the minimum value of a data set is shown by way of example;

[0104] Figure 7 The grouping of multiple individual peaks on a data set is shown by way of example;

[0105] Figure 8 Characteristic peaks are shown by way of example;

[0106] Figure 9 A time sequence of method steps is shown by way of example;

[0107] Figure 10 By way of example, a device for classifying brain pressure is shown.

[0108] The features disclosed in the above description, the drawings and the claims may be essential for realizing the invention in its various configurations, both individually and in any combination.

[0109] Like reference numerals in the drawings denote like elements. DETAILED DESCRIPTION

[0110] Figure 1 By way of example, a measuring device for determining time-of-flight values of one or more ultrasound signals through a cross section of the skull is shown.

[0111] Two probes are shown, positioned on opposite sides of a skull 1 with a brain 2. The two probes 3 include at least one transmitter configured to output one or more ultrasound signals and at least one detector configured to receive the one or more ultrasound signals output by the at least one transmitter. The one or more ultrasound signals may be longitudinal waves. Furthermore, the one or more ultrasound signals may include one or more wave packets having a constant frequency. The at least one transmitter and the at least one detector are positioned on different, opposite sides of the skull 1.

[0112] Between emission and detection, one or more ultrasound signals traverse the cross-section of the skull 1. Thus, an acoustic field is formed between the emitter and the detector that covers a specific area of the skull cross-section and traverses the one or more ultrasound signals. Preferably, the acoustic field can be aligned parallel to the frontal plane of the skull 1.

[0113] As described above, the propagation time of one or more ultrasound signals measured at different time points depends on changes in the internal state of intracranial pressure in the brain 2. Therefore, changes in the time-of-flight value between two consecutive time points map changes in the state of components in the skull 1.

[0114] The validation of the measured time-of-flight values by comparing them with the time course of the brain pressure measured according to a gold standard, measured simultaneously on the same patient, is explained below.

[0115] Figure 2 Parts (a) and (c) show intracranial pressure measurements, which were performed on patients with normal intracranial pressure (part (a)) and abnormal intracranial pressure (part (c)) using a measurement method according to the gold standard of intracranial pressure measurement. Figure 2 In parts (b) and (d) time-of-flight measurements are shown, which were performed simultaneously with intracranial pressure measurements on the same patient with normal intracranial pressure (part (b)) and abnormal intracranial pressure (part (d)). Figure 3 The time-of-flight values shown have been measured and pre-processed by the method described in more detail in .

[0116] The intracranial pressure measurements shown in parts (a) and (c) and performed according to the gold standard serve as a reference. As mentioned above, in the gold standard for intracranial pressure measurement, a pressure measuring probe is inserted into the brain 2, thereby directly and relatively reliably measuring intracranial pressure. This measurement method is widely recognized and therefore serves as a reference.

[0117] exist Figure 2 Part (a) shows the time course of intracranial pressure (ICP) measured directly in a patient with normal intracranial pressure. The time course of intracranial pressure includes a characteristic curve with multiple peaks, each of which reflects a pressure increase during systole and a pressure decrease during diastole. These peaks clearly have a continuous, specific shape that can be attributed to the elasticity of the patient's brain 2 within the normal range, resulting in a reliable inference of intracranial pressure within the normal range.

[0118] exist Figure 2 In part (c) of FIG. 1 , the time course of the intracranial pressure (ICP) measured directly on a patient with abnormal intracranial pressure is shown. The time course of the intracranial pressure also includes a characteristic curve including a plurality of peaks associated with the systolic and diastolic phases. Figure 2 As can be seen from part (c), the curve has a significantly different shape compared to the curve of the intracranial pressure time course of a patient with normal intracranial pressure. Therefore, abnormal intracranial pressure can be reliably inferred from the curve shape.

[0119] Therefore, normal or abnormal ICP can be reliably inferred from the curves of the time course of ICP measurements according to the gold standard.

[0120] In contrast, Figure 2 In part (b), the time course of the time-of-flight values of one or more ultrasound signals measured by means of the device through a cross-section of the skull is shown, which is determined simultaneously on the same patient together with the normal intracranial pressure as in part (a). The time course of the time-of-flight values also includes a characteristic curve with multiple peaks, which can therefore also be assigned to the systolic and diastolic periods. It can be seen that the curve of the time-of-flight values has a shape that is very similar to the curve of the time course of the intracranial pressure shown in part (a). Therefore, it can be reliably inferred that the curve of the time course of the time-of-flight values is directly related to the changes in the intracranial pressure in the skull 1, and the normal intracranial pressure of the patient can be inferred with the help of the curve of the time course of the time-of-flight values.

[0121] In part (d), the time course of the time-of-flight values of one or more ultrasound signals measured by means of the device through a cross-section of the skull is shown, which is determined simultaneously on the same patient together with the abnormal intracranial pressure as in part (c). The time course of the time-of-flight values also includes a characteristic curve with multiple peaks, which can therefore also be assigned to the systolic and diastolic periods in each case. It can be seen that the curve of the time-of-flight values has a shape that is very similar to the curve of the time course of the intracranial pressure shown in part (c). Therefore, it can be reliably inferred that the curve of the time course of the time-of-flight values is directly related to the change in the intracranial pressure in the skull 1. Therefore, the abnormal intracranial pressure of the patient can be reliably inferred with the help of the time course curve shown in part (d).

[0122] These simultaneous measurements were repeated on a number of patients with normal and abnormal intracranial pressure, and in principle, a similar correlation was observed between the time course of intracranial pressure measured according to the gold standard and the time course of the time-of-flight values. The above-mentioned physiological inclusion criteria were used as a basis for patient selection.

[0123] Normal or abnormal brain pressure can therefore be reliably inferred with the aid of the curve of the time course of the time-of-flight values measured at successive points in time.

[0124] This finding can be explained by changes in the state of brain pressure within the skull 1 , which are directly related to changes in the speed of sound and therefore to the time of flight of one or more ultrasound signals through a cross section of the skull 1 .

[0125] Thus, the time-of-flight values of one or more ultrasound signals through the cross section of the skull 1 measured at consecutive time points detect time-of-flight differences caused by pressure-induced volume changes and pressure-induced compression of components of the brain 2 (e.g., blood, cerebrospinal fluid, and tissue).

[0126] Therefore, the time course of the time-of-flight values measured at consecutive time points reflects the changes in volume and density of components of the brain 2 during the pulse wave, and thus reflects the changes in brain pressure, as a result of which the brain pressure can be reliably classified non-invasively.

[0127] The correlation between the flight time value of one or more ultrasound signals through the brain and the internal state of intracranial pressure in the brain 2 has been scientifically proven.

[0128] Combine Figure 3 , the measurement of time-of-flight values of one or more ultrasound signals through the brain 2 and the subsequent pre-processing of the data of the measured time-of-flight values are explained in more detail.

[0129] It should be noted that the measurement method and the preprocessing of the measured values are shown merely as examples, and that other measurement methods and data preprocessing are also possible in order to achieve sufficient data quality for classifying the intracerebral pressure using the time course of the time-of-flight values. Thus, for example, other measurement methods that provide sufficient data quality can be used, without requiring corresponding data processing to determine an evaluable time course.

[0130] A constant-frequency wave packet of one or more ultrasonic signals is output by the transmitter at one of a series of time points. The wave packet comprises a periodic signal with n cycles, where n is an integer. The wave packet passes through a cross-section of skull 1 and is detected at a detector on the opposite side of skull 1. A time-of-flight value is determined for the time of flight from the transmitter to the detector during each of the n cycles of the wave packet. This means that n time-of-flight values are assigned to each time point.

[0131] Based on these measurements, n time-of-flight values are determined at each time point, wherein a time point is assigned to each of the n time-of-flight values. Then, n time distributions of the time-of-flight values are formed, wherein in each of the n time distributions, a time-of-flight value for the same period in the wave packet is assigned to the time point of the wave packet. Thus, each of the n time distributions forms a time-of-flight curve. Thus, n time-of-flight curves are formed depending on the number of n periods in the wave packet.

[0132] Figure 3 (a) shows one of n time-of-flight curves with directly measured time-of-flight values, wherein the time points are specified in seconds on the X-axis and the time-of-flight values are specified in microseconds on the Y-axis. In the time-of-flight curve, the time-of-flight value of the period of the periodic signal of the wave packet is assigned to each consecutive time point. The time-of-flight value fluctuates within the approximately 5 seconds shown due to the heartbeat. In addition, there are n-1 additional time-of-flight curves (not shown), wherein in each of the n-1 time-of-flight curves, the time points assigned by the same period are assigned to the corresponding time points.

[0133] An offset is subtracted from each of the n time-of-flight curves. The offset is subtracted by applying a bandpass filter to the time-of-flight values in the time-of-flight curves. Low-frequency components (such as trends or breathing effects) can be cut off by the bandpass filter.

[0134] Figure 3 (b) shows by way of example the output from Figure 3 (a) Time-of-flight curve, where the time points are specified in seconds on the X-axis and the time-of-flight values are specified in nanoseconds on the Y-axis.

[0135] An average flight time curve is then formed by averaging the flight time values of each of the n flight time curves at one of the consecutive time points. In the average flight time curve, the average flight time value from the corresponding n flight time values is assigned to the time point. The corresponding average flight time curve is Figure 3 This is shown by way of example in (c), where the time points are specified in seconds on the X-axis and the time-of-flight values are specified in nanoseconds on the Y-axis.

[0136] The averaged time-of-flight curve can be used as a basis for further processing of the data. However, in the case of sufficient signal quality, time-of-flight values which have not yet undergone such a measurement and processing process can also be used as a basis for further processing.

[0137] Furthermore, the temporal distribution of the time-of-flight values can be subjected to an optional preparatory step for classification. In this preprocessing step for classification, the directly measured temporal distribution of the time-of-flight values can be combined with the corresponding assigned time-of-flight values to form time segments of consecutive time points, where these segments are used as the basis for data preprocessing and to form a plurality of segments with corresponding averaged time-of-flight curves. Alternatively, time segments of consecutive time points with corresponding assigned time-of-flight values can be formed from the averaged curve and used as the basis for further processing.

[0138] One or more segments that meet a quality criterion may be selected from the segments formed in this manner.

[0139] The quality criterion may be formed by an autocorrelation function that evaluates the self-similarity of the time course of the time-of-flight values in a segment. Alternatively or separately, the quality criterion may be formed by comparing the maximum amplitude of the most dominant frequency component in the time course of the time-of-flight values in a segment with the amplitudes of other frequency components. Alternatively or separately, the quality criterion may be formed by evaluating the frequencies in the time course of the time-of-flight values in a segment.

[0140] The selected segments can be used as a basis for further processing and in particular can be loaded into a classifier.

[0141] Figure 4 The operating mode of a classifier for classifying the characteristics of the time course of a time-of-flight curve is shown by way of example.The classifier is implemented by a computer.

[0142] Data comprising time-of-flight values of one or more ultrasound signals through a cross section of the skull 1 measured at successive time points are loaded into the classifier. The loaded data may be data directly from successive measurements of time-of-flight values that have not yet undergone a preprocessing step, data that have already undergone a preprocessing step, in particular an average time-of-flight curve, or segments selected according to quality criteria.

[0143] In the classifier, the properties of the brain pressure are classified with the aid of one or more features of the representative curve of the time course of the time-of-flight values, in order to subsequently output a classification feature of the brain pressure.

[0144] Figure 5 A detailed view of a classifier for classifying the characteristics of brain pressure is shown.

[0145] As described above, data comprising time-of-flight values measured at consecutive time points are loaded into the classifier.

[0146] In one aspect, a plurality of selected segments are loaded into the classifier and combined to form a time course of time-of-flight values. In another aspect, unprocessed time-of-flight values are loaded into the classifier. In another aspect, an average time-of-flight curve is loaded into the classifier.

[0147] Individual peaks were initially identified by detecting the minimum in the time course between adjacent peaks in the corresponding time course of the time-of-flight values. Figure 6 The corresponding data set is shown, where the consecutive time points are plotted on the X-axis at 10 -2 The time of flight is specified in seconds on the Y-axis while the relative time of flight values are specified in nanoseconds on the Y-axis. The minimum value detected is indicated by an unfilled point.

[0148] exist Figure 5 In the next step the peaks determined in this way are extracted and normalized in time to a common starting time point.

[0149] In an optional subsequent step, peaks that are too short or too long can be sorted out. The sorting out of too short or too long peaks is generated by determining whether the peak width at the detected minimum is below or above a reference value, which is determined by the median or average peak width of the cutoff peaks at their minima.

[0150] In one step, the cutoff and normalized peaks are grouped. Grouping is performed by applying a similarity criterion to the individual peaks. The peak group containing the most peaks is selected for further processing.

[0151] exist Figure 7 The cutoff and normalized peak grouping are shown by way of example, where the x-axis is shown without units at 10 -2 The time points are normalized to the time of flight in seconds, and the time of flight values normalized to the maximum peak amplitude are shown without units on the Y axis. In this example, the peak group with amplitude maxima within about 2 seconds is selected for further processing (shown by the solid line in the curve).

[0152] At least one characteristic peak of a representative curve can be formed by medianing the corresponding time-of-flight values of the peaks of the selected curve at the corresponding time points. Thus, the median is formed from the time-of-flight values of the individual peaks at the time points. Thus, the at least one characteristic peak characterizes a peak among the plurality of peaks over the time course of the time-of-flight.

[0153] The peaks identified in this way are Figure 8 The example is shown in FIG, where the normalized time points are plotted on the X-axis at 10 -2 Units are specified in seconds while the normalized time-of-flight values of characteristic peaks are specified without units on the Y-axis.

[0154] The characteristics of the curve of the at least one characteristic peak are then determined from the at least one characteristic peak.

[0155] As described above, these features are determined during the learning phase. A feature of the one or more features may be, for example, the similarity of the curve shape of at least one characteristic peak to a synthesis function, in particular a continuous wavelet transform. Furthermore, another feature may be the peak width of at least one characteristic peak at a certain percentage of the maximum amplitude of the at least one characteristic peak.

[0156] Based on one or more features of the characteristic peak, the characteristics of the intracranial pressure can be classified, specifically whether the intracranial pressure is normal or abnormal. Based on this classification, a computer can determine and output the classification. A disease diagnosis can be performed based on the output classification.

[0157] Figure 9 A time sequence of the steps of the above method is shown.

[0158] In a first step, the time-of-flight value of one or more ultrasound signals passing through the skull 1 is measured.

[0159] In a second step, the time-of-flight values thus measured can be pre-processed for further processing.

[0160] In a third step, the time-of-flight values may be prepared for classification.

[0161] In the second and third steps, pre-processing of the time-of-flight values and preparing the time-of-flight values for classification is optional. For example, the measured time-of-flight values can be loaded directly into the classifier.

[0162] In a fourth step, the properties of the brain pressure are classified with the aid of one or more features of the representative curve of the time course of the time-of-flight values to obtain a classification feature of the brain pressure.

[0163] In the fifth step, a classification of the characteristics of the brain pressure is output.

[0164] Figure 10 By way of example, a device for classifying brain pressure is shown.

[0165] The device comprises two probes 3. Each probe 3 includes a transmitter for outputting one or more ultrasonic signals and a detector for detecting the one or more ultrasonic signals. Probes 3 can, for example, be part of a measurement device. Probes 3 measure the time of flight of one or more ultrasonic signals from the transmitter through the skull 1 to the detector.

[0166] The measured time-of-flight values for the ultrasound signal are transmitted to the computer 4. The transmission is performed by means of a suitable medium. For example, the data with the time-of-flight values can be transmitted to the computer 4 by means of a data carrier (e.g. a USB stick). In other examples, the data with the time-of-flight values can be transmitted to the computer 4 in a wired manner. In a further example, the data can be transmitted to the computer 4 wirelessly, for example by means of a WLAN or Bluetooth connection. To this end, the probe 3 or the measuring device is in a corresponding communication connection with the computer 4. The computer 4 executes the above-described method for classifying the intracerebral pressure based on the received data with the time-of-flight values. The output can be performed on a display connected to the computer 4. The classification characteristics of the intracerebral pressure can then be displayed to the user via the display.

[0167] Furthermore, a computer readable medium 5, such as a storage medium or a data carrier, which can be connected to a computer is provided. The computer readable medium 5 comprises instructions which, when executed by a computer, cause the computer to perform the steps according to the method.

[0168] The above aspects and embodiments of the technology described herein can be implemented in a variety of ways. For example, the embodiments can be implemented using hardware, software, or a combination thereof.

[0169] If the features are implemented in software, the software code can be executed on any suitable computer or processor or collection of processors, regardless of whether it is provided in a single computer or distributed across multiple computers. Such a processor can be implemented as an integrated circuit, with one or more processors in an integrated circuit assembly, including commercially available integrated circuit assemblies known in the art as existing under names such as CPU chips, GPU chips, microprocessors, microcontrollers, or coprocessors. Alternatively, the processor can be implemented as a configuration of an ASIC or programmable logic device. Alternatively, the processor can be part of a larger circuit or semiconductor device. As a specific example, some commercially available microprocessors have multiple cores, so that one or a subset of these cores can form a processor. However, the processor can be implemented using circuits of any suitable format.

[0170] Furthermore, the computer may be implemented in any of a variety of forms, for example, a rack-mounted computer, a desktop computer, a laptop computer, or a tablet computer. Furthermore, a computer may be embedded in a device not generally considered a computer but having suitable processing capabilities, including a personal digital assistant (PDA), a smart phone, or other suitable portable or permanently installed electronic device.

[0171] Such computers can be interconnected by one or more networks of any suitable form, including as a local area network or a wide area network, such as an enterprise network or the Internet. Such networks can be based on any suitable technology and can operate according to any suitable protocol, and can include wireless networks, wired networks, or fiber optic networks.

[0172] In addition, the various methods or processes outlined herein can be encoded as software that can be executed on one or more processors using any of a plurality of operating systems or platforms. In addition, such software can be written using a variety of appropriate programming languages and / or programming or scripting tools, and can also be compiled into executable machine language code or intermediate code that is executed on a framework or virtual machine. The terms "(computer) program" or "software" used herein refer to any type of computer code or computer executable instruction set in a general sense that can be used to program a computer or other processor to implement various aspects of the present invention as described above. In addition, it should be noted that according to one aspect of this embodiment, one or more computer programs perform the method of the present invention when executed, and they do not need to reside on a single computer or processor, but can be distributed in a modular manner among a plurality of different computers or processors to implement various aspects of the present invention.

[0173] The present invention can be implemented as a computer-readable (memory) medium (or multiple computer-readable media) encoded with one or more programs (e.g., computer memory, one or more floppy disks, compact disks (CDs), optical disks, digital video disks (DVDs), magnetic tape, flash memory, circuit configurations in field programmable gate arrays or other semiconductor devices, or other tangible computer storage media) that, when executed on one or more computers or processors, performs methods that implement the various embodiments of the present invention described above. As can be seen from the above examples, computer-readable storage media can store information for a sufficient time to provide computer-executable instructions in a non-transitory form. Such computer-readable storage media can be portable, such that one or more programs stored thereon can be loaded onto one or more different computers or other processors to implement various aspects of the present invention as described above. As used herein, the term "computer-readable (memory) medium" only includes non-transitory computer-readable media, which can be considered to be physical articles of manufacture (i.e., articles of manufacture) or machines. Alternatively or additionally, the present invention can be implemented as a computer-readable medium other than a computer-readable storage medium, such as a propagating signal.

[0174] Computer-executable instructions or commands can take many forms, such as program modules, that are executed by one or more computers or other devices. Typically, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In general, in various embodiments, the functionality of program modules can be combined or distributed as needed.

[0175] The features of various aspects or embodiments have been described above by way of example so that those skilled in the art may better understand the present invention. However, it is clear that aspects and embodiments other than those described in detail may also be the subject matter of the present invention, wherein the present invention is limited by the scope of the appended claims.

[0176] Reference Signs List

[0177] 1 Skull

[0178] 2 Brain

[0179] 3 Probes

[0180] 4 Computer

[0181] 5 Computer-readable media

Claims

1. A computer-implemented method for classifying intracranial pressure, the method comprising: - receiving data comprising time-of-flight values based on time-of-flight values of one or more ultrasound signals through a cross section of the skull (1) measured at successive points in time; - classifying the characteristic of the brain pressure by means of one or more features of the representative curve of the time course of the time-of-flight values to obtain a classified characteristic of the brain pressure; and - outputting the classification characteristic of the brain pressure.

2. The method according to claim 1, wherein The time course of the time-of-flight values is correlated with the time course of the brain pressure within the skull (1).

3. A method according to any one of the preceding claims, wherein The time course of the time-of-flight values includes a plurality of peaks, each of the plurality of peaks being associated with an increase and a decrease in the brain pressure during systole and diastole.

4. A method according to any one of the preceding claims, wherein The change between the time-of-flight values at at least two consecutive time points during the time course is directly related to the change in the brain pressure at the at least two consecutive time points.

5. A method according to any one of the preceding claims, wherein The classification characteristic of the brain pressure indicates normal brain pressure, in particular a brain pressure of 10 mmHg or lower, or indicates abnormal brain pressure, in particular a brain pressure of 15 mmHg or higher.

6. A method according to any one of the preceding claims, wherein The representative curve includes at least one characteristic peak.

7. The method according to claim 6, wherein: The characteristic of the brain pressure is determined by the computer by extracting the one or more features of the curve of the at least one characteristic peak.

8. The method according to claim 7, wherein: A first characteristic feature of the one or more characteristics is a peak width of the at least one characteristic peak at a certain percentage of a maximum amplitude of the at least one characteristic peak.

9. The method according to claim 7 or 8, wherein A second characteristic feature of the one or more characteristics is the similarity of the curve shape of the at least one characteristic peak to a synthesis function, in particular a continuous wavelet transform.

10. The method according to any one of claims 6 to 9, wherein The at least one characteristic peak characterizes a plurality of peaks in at least one time period of the temporal course of the time-of-flight values.

11. The method according to claim 10, wherein: The at least one characteristic peak is determined by the following steps: - extracting a plurality of individual peaks from said at least one time segment of said time course of said time-of-flight values, - normalizing the time points of the time-of-flight values in the plurality of individual peaks, - grouping the plurality of individual peaks by applying a similarity criterion to the plurality of individual peaks, - identify the group with the largest number of peaks, and - forming the at least one characteristic peak from a plurality of peaks in the identified peak group.

12. The method according to claim 11, wherein Forming the at least one characteristic peak comprises forming, for each time point, a median of the time-of-flight values of the plurality of peaks in the identified group.

13. The method according to any one of claims 10 to 12, wherein: The at least one time period is determined by: - extracting one or more time periods from said time course of said time-of-flight values, wherein a time period comprises a plurality of peaks in said time-of-flight values, - determining said at least one time period by selecting one or more time periods in which said plurality of peaks meet a quality criterion.

14. The method according to claim 13, wherein: The quality criterion is formed by an autocorrelation function which assesses the self-similarity of the time course of the time-of-flight values in a segment.

15. The method according to claim 13 or 14, wherein: The quality criterion is formed by comparing the maximum amplitude of the most dominant frequency component in the time course of the time-of-flight values in a segment with the amplitudes of the other frequency components.

16. The method according to any one of claims 13 to 15, wherein: The quality criterion is formed by evaluating the frequency of the time-of-flight values in a segment over the time course.

17. A method according to any one of the preceding claims, wherein Receiving data including a time-of-flight value includes: - receiving n flight time curves, wherein in each of said n flight time curves a flight time value is assigned to one of said consecutive time points, - for each of the n flight time curves, subtracting an offset from the flight time value in the flight time curve, - forming an average flight time curve by averaging the flight time values of each of the n flight time curves at one of the consecutive time points.

18. The method according to claim 17, wherein: In one of the n flight time curves, the flight time value at a time point is formed by the period of the n-period wave packet of the one or more ultrasonic signals output or received at the time point.

19. The method according to claim 17 or 18, wherein Each of the n time-of-flight curves has a periodicity given by the heart rate.

20. The method according to any one of claims 17 to 19, wherein: The one or more time periods are selected from the average flight time curve to determine the at least one time period.

21. The method according to any one of claims 17 to 20, wherein: A bandpass filter is applied to the time-of-flight values in the time-of-flight curve to determine the offset.

22. A device, in particular a computer (4), configured to perform the method according to any of the preceding claims.

23. An apparatus comprising the apparatus according to claim 22 and at least two probes (3) for determining the time-of-flight values of the one or more ultrasound signals through a cross section of a skull (1), wherein: The at least two probes (3) include at least one transmitter for outputting the one or more ultrasound signals, and correspondingly at least one detector for detecting the outputted one or more ultrasound signals, wherein, in particular, the at least one transmitter and the at least one detector can be positioned on opposite sides of the skull (1), in particular in a plane passing through a frontal cross-section of the skull (1).

24. The apparatus according to claim 23, wherein: The at least one emitter and the at least one detector can each be positioned in a region above the external auditory canal.

25. The apparatus according to claim 23 or 24, wherein: The at least two probes (3) are configured as: - outputting a plurality of wave packets, each wave packet having an n-periodic signal of a constant frequency at a corresponding time point in the continuous time points, - for each of n periods of the periodic signal in each of the plurality of wave packets, detecting a time-of-flight value through a cross section of the skull (1).

26. A computer program comprising instructions causing a computer (4) to perform a method according to any one of claims 1 to 21, when said method is executed by the computer (4).

27. A computer-readable medium (5), wherein: The medium comprises instructions which, when the method is executed by a computer (4), cause the computer (4) to perform the method according to any one of claims 1 to 21, or wherein a computer program according to claim 26 is stored on the medium (5).

Images