DE102018106388A1 - Device and method for controlling and / or regulating medical devices - Google Patents

Abstract

The present invention relates to a device (10) for controlling and / or regulating medical devices (12) in an operating room (14), comprising a voice control module (16) on which a speech recognition software based on a neural network (18) can be executed offline at least one microphone array (34) which can be connected or connected to the voice control module (16) can receive voice commands (C) emitted by at least one person (22) in the operating room (14) and voice signals (S.) that can be processed by the voice control module (16) ), wherein the speech control module (16) is arranged such that the speech signals (S) using the speech recognition software in device - related control signals (SD) and personal control signals (SP) are convertible, at least one medical device (12), which with the Voice control module (16) is connectable or connected and which with the device-related NEN control signals (SD) can be controlled and / or regulated, and at least one speaker (26), with which the personal control signals (SP) can be output acoustically. Furthermore, the invention relates to a method for operating such a device (10).

Description

The present invention relates to a device and a method for controlling and / or regulating medical devices.

For hygienic reasons, medical operations are performed on the human body in a largely sterile operating room. During surgery, the patient is typically stored on an operating table so that the site to be operated is readily accessible to a physician performing the surgery. In surgery, a variety of medical devices are used, such as pumps, with which the patient a certain amount of infusion fluid is supplied. Medical devices should be understood as devices that serve a medical purpose in the broadest sense. The medical devices are usually not operated directly by the operating physician, but by an assistant who operates the medical device according to the instructions of the operating physician. It is not excluded that the instructions of the operating doctor by the assistant are not understood correctly and therefore not implemented as desired. This can lead to delays in the course of the operation. But it can also lead to serious complications and irreversible health consequences for the patient.

In addition to problems in the communication between the operating doctor and the assistant, the cost of the operation due to the high staffing requirements are very high.

One way to reduce the need for personnel is the use of a voice control, with which the medical devices can be controlled and / or regulated by means of voice commands issued by the operating physician or the assistant. For example, the delivery rate of the above-mentioned pump can be increased or decreased. The basic structure of devices for speech recognition and voice control is for example from the US 2015/0364135 A1 known. The EP 1 868 184 A1 describes the use of speech recognition and voice control devices in medical applications.

However, such speech recognition and voice control devices have not been established to support operations because such devices are cumbersome in operation and suffer from insufficient recognition rates of the voice commands emitted by the persons in the operating room. Inadequate recognition rates are caused, among other things, by several persons who communicate with one another in the operating theater, by different voice positions of the persons speaking or by background noises.

It is therefore an object of one embodiment of the present invention to provide a device for controlling or regulating medical devices, which has an increased recognition rate compared to known devices and is easy and reliable to operate. Furthermore, an embodiment of the present invention has the object to provide a method with which such a device can be operated.

This object is achieved with the features specified in claims 1 and 7. Advantageous embodiments are the subject of the dependent claims.

One embodiment of the invention relates to a device for controlling and / or regulating medical devices in an operating room, comprising a voice control module, on which a neuronal network-based speech recognition software is executable offline, at least one connectable to the voice control module or connected microphone array with at least two microphones for providing a far-field characteristic with which speech commands emitted by at least one person in the operating room are recordable and convertible into voice signals that can be processed by the speech control module, wherein the speech control module is configured such that the speech signals are converted into device-related drive signals and personalized drive signals using the speech recognition software are convertible, at least one medical device which is connectable or connected to the voice control module and which with the gerätebez ogenen control signals can be controlled and / or regulated, and at least one speaker, with which the personalized control signals can be output acoustically.

An essential aspect of the proposed device is that the speech recognition software is based on a neural network. An essential feature of neural networks is that they are self-learning and can recognize certain patterns. The more frequently the device is used, the better the recognition rate becomes. In addition, the voice recognition software is able to make a suggestion when the voice command issued can not be unambiguously associated with any of the voice recognition software known voice signals corresponding to a number of voice commands. It can the Speech recognition software based on methods of artificial intelligence. Moreover, the neural network-based speech recognition software is capable of recognizing noise emitted by medical devices, for example, and ignoring it when processing voice commands. In operating theaters there is typically a noise floor of 40db. In many cases, the operating rooms are completely tiled, which can result in reverberation times of up to three seconds. The neural network-based speech recognition software ensures the necessary recognition rate even under these difficult conditions.

Another aspect is that the device is operated offline during the operation. This does not exclude that the device can be operated online outside of operations, for example for remote maintenance or for updating the speech recognition software. During operation, off-line operation of the device prevents external manipulation, which can lead to malfunction. It also prevents sensitive patient data from being stolen.

An additional essential aspect of the present invention is the use of a microphone array. In known generic devices headsets are used, which have several disadvantages. Not a few doctors or assistants feel disturbed by headsets. In addition, they may slip during surgery or even come off the head, causing delays in the operation. Furthermore, they are disadvantageous from a hygienic point of view and can endanger the necessary sterility. In contrast, microphone arrays can be installed so that they can be well sterilized. Furthermore, they do not disturb the doctor or assistant while working in the operating room. In addition, there is no danger that the microphone array can slip or fall uncontrollably.

The far-field characteristic can be realized by arranging at least two microphones in a microphone array. Only in the geometrical arrangement of the microphones in an array does the far-field characteristic (microphone array with far-field characteristic) develop, with which one is able to implement an adaptive beamforming algorithm, which leads to a significant improvement of the audio signal and consequently leads to an increase in the recognition rate.

Due to the spacing of the microphones of a microphone array relative to one another, a specific region can be scanned over a phase offset in the field of view of a microphone array, which can amount to 180 °, for example, to generate the far-field characteristic. As a result, the orientation and the position of the person who issues a voice command can be detected very accurately. The acoustic signal, which can be assigned to the recognized orientation and the detected position, can be selectively amplified and the acoustic signals which do not correspond to the recognized orientation and the detected position are attenuated.

Near field is the immediate area around the sound source, which is characterized by an uneven change between locations with constructive and destructive interference. In contrast, the far field refers to an area that is far away from the sound source.

Microphone arrays can capture voice commands with sufficient accuracy even if the person issuing the voice commands is at some distance, which may be up to 25 m, from the microphone array. The freedom of movement of the voice commands issuing person is not restricted.

The speech recognition software is designed to convert the speech signals generated by the microphone array as a result of the delivery of voice commands, firstly into device-related drive signals and secondly into personalization drive signals. Device-related activation signals should be understood as activation signals with which the relevant medical devices can be controlled and / or regulated, that is to say used by the medical devices. Person-processed control signals should be understood to mean such control signals which correspond in terms of content to the device-related control signals but can be output acoustically or optically and are accessible and understandable by a person and are not used to control the medical devices. For example, when the operating physician issues the voice command "turn pump on", the device-related drive signals sent to the pump cause it to turn on. The personal control signals control the loudspeaker in such a way that it outputs the acoustic indication "turn on the pump". Thus, the proposed device provides audible feedback to the operating physician or assistant so that the person who has issued the voice command will receive an audible feedback as to whether the device has correctly recognized the delivered voice command. Control of the device and the assurance that the device is issuing the desired command are thereby increased.

In a further developed embodiment, the device may comprise a radio or other device generating audio signals which is connectable or connected to the voice control module or to the microphone array, the device comprising a filter device for filtering the audio signals. To improve the working environment during an operation, in many cases a radio is running in the background. However, the audio signals produced by the radio can interfere with the processing of the voice commands and therefore reduce the recognition rate. In addition, the audio signals generated by the radio may also coincidentally contain commands that would lead to a driving of the medical devices. In order to eliminate such effects, the radio is connected via an audio output to the filter device, which filters out the audio signals during the processing of the voice commands, in particular before being supplied to the speech recognition software, so that the audio signals do not interfere with the recognition rate.

It is also possible to use the radio as a speaker for acoustically outputting the personalization control signals. In this case, the volume at which the audio signals are output may be lowered prior to the output of the personalization drive signals, so that the personalized drive signals are well understood and are not overlaid by the audio signals. Upon completion of the audible output of the personalization drive signals, the volume at which the audio signals are output may be restored to the original level. Other audio-generating devices such as MP3 players can be used alternatively or cumulatively to the radio.

In a further developed embodiment, the device may comprise at least one screen for displaying operation-related data, with which the personalized control signals can be displayed optically. The optical output is made in addition to the acoustic output. With the optical output, the voice command can be specified in clear text on the screen. Alternatively or cumulatively, icons may be displayed which correspond to the voice commands, which is beneficial for international operational teams to keep language barriers low. The visual output of the personalization drive signals in addition to the audible output provides reassurance to the operating physician or assistant. In addition, the visual output may continue until the next voice command is issued. Thus, it is always clear which voice command has been issued previously. Also, a history of the delivered voice commands can be displayed in chronological order, so that the operation process can be understood at a glance.

Since screens are used today in almost every operating room to display operation-related data such as blood pressure and heart rate of the patient to be operated, this embodiment of the device requires no additional equipment. In addition, each additional device is a source of contamination, so that in this embodiment, the number of contamination sources is not increased.

In a further embodiment, the at least one microphone array can be arranged on the screen. As already mentioned, screens are used in almost every operating room. The arrangement of the microphones on the screens is therefore advisable, since no additional equipment is required. In addition, each additional device is a source of contamination, so that in this embodiment, the number of contamination sources is not increased. Since the surgeon or assistant often directs his gaze to the screens during the operation, the arrangement of the microphones also has the effect that the voice commands are in many cases spoken directly into the microphones, which improves the quality of the voice signals compared to voice commands, which are delivered in a different direction, increased. The recognition rate can be kept high in this embodiment.

A further embodiment is characterized in that the medical devices include lasers, pumps, endoscopes, lamps, operating tables, recording devices, an operating room tempering air conditioning system or the like. Such medical devices are often used during operations, so that their control and / or regulation with the proposed device is particularly useful for the operation of the operation. Also, an operating room tempering air conditioning can be controlled with the proposed device. Even though air conditioning is generally not considered to be primarily a medical device, an air conditioning system can still produce a medical effect when it tempers the operating room. For example, hypothermia of the patient with a corresponding temperature of the operating room can be avoided. In this respect, an operating room tempering air conditioning should be understood in the context of this description as a medical device.

• - Offline Ausführen einer auf einem neuronalen Netz basierenden Spracherkennungssoftware mittels eines Sprachsteuerungsmoduls,
• - Aufnehmen zumindest eines von zumindest einer im Operationssaal befindlichen Person abgegebenen Sprachbefehls mittels eines mit dem Sprachsteuerungsmodul verbundenen Mikrofon-Arrays,
• - Umwandeln des aufgenommenen Sprachbefehls in vom Sprachsteuerungsmodul verarbeitbare Sprachsignale mittels des Mikrofon-Arrays,
• - Umwandeln der Sprachsignale in gerätebezogene Ansteuerungssignale und personenaufbereitete Ansteuerungssignale unter Verwendung der Spracherkennungssoftware,
• - Steuern und/oder Regeln zumindest eines medizinischen Geräts mit den Ansteuerungssignalen, und
• - akustisches Ausgegeben der personenaufbereiteten Ansteuerungssignale mittels eines Lautsprechers.

An embodiment of the invention relates to a method for controlling and / or regulating medical devices in an operating room with a device according to one of the preceding claims, comprising the following steps:

• Offline execution of a speech recognition software based on a neural network by means of a speech control module,
• Receiving at least one voice command issued by at least one person in the operating room by means of a microphone array connected to the voice control module,
• Converting the recorded voice command into voice signals that can be processed by the voice control module by means of the microphone array,
• Converting the speech signals into device-related drive signals and personalized drive signals using the speech recognition software,
• Controlling and / or regulating at least one medical device with the drive signals, and
• - Acoustic output of the personal control signals by means of a loudspeaker.

The technical effects and advantages that can be achieved with the proposed method correspond to those discussed for the present device. In summary, it should be noted that a high recognition rate of the delivered voice commands can be provided at the same time due to the offline operation of high data security. In addition, the procedure is user-friendly.

• - akustisches Ausgegeben von einigen oder allen der personenaufbereiteten Ansteuerungssignale mittels eines Lautsprechers,
• - Empfangen eines Bestätigungsbefehls oder eines Ablehnungsbefehls, die von zumindest einer im Operationssaal befindlichen Person abgegeben werden, und
• - für den Fall, dass die im Operationssaal befindliche Person ein Bestätigungsbefehl abgibt, Steuern und/oder Regeln zumindest eines medizinischen Geräts mit den gerätebezogenen Ansteuerungssignalen.

In a further embodiment, the proposed method comprises the following steps:

• audible output of some or all of the personalization driving signals by means of a loudspeaker,
• Receiving a confirmation command or a refusal command issued by at least one person in the operating room, and
• - in the event that the person in the operating room gives an acknowledgment command, controlling and / or regulating at least one medical device with the device-related control signals.

In this embodiment, the personalized control signal generated by the speech module on the basis of the voice command issued is first emitted acoustically. The device to be controlled is only activated according to the voice command when an acknowledgment command is issued. If a refusal command is given, no action is taken. In this embodiment, the assistant or the surgeon has the possibility of not having misrecognized or mispelled voice commands executed, thereby avoiding errors. It is possible to select those personalization control signals to be issued acoustically. For example, if you want to increase the flow rate of the pump, the voice command "increase flow rate" may be used. In this case, the flow rate per voice command is increased by a certain amount. If, however, the delivery rate is to be increased by five times the amount, for example, this can be achieved by issuing the voice command "increase delivery rate" five times in succession. If the device were to activate the pump each time only after confirmation of the voice command, this would be very cumbersome and time-consuming for the person who issues the voice command. Characterized in that the personal processing control signals are selected, which are to be output acoustically, the situation described can be avoided.

• - Hinterlegen einer Anzahl von ersten Sprachsignalen in einer mit der Spracherkennungssoftware zusammenwirkenden Datenbank, wobei die ersten Sprachsignale einer Anzahl von Ansteuerungssignalen entsprechen,
• - Vergleichen eines zweiten Sprachsignals, welches einem von zumindest einer im Operationssaal befindlichen Person abgegebenen Sprachbefehl entspricht, mit den ersten Sprachsignalen mittels der Spracherkennungssoftware, und
• - für den Fall, dass eine hinreichende Übereinstimmung zwischen einem der ersten Sprachsignalen und dem zweiten Sprachsignal von der Spracherkennungssoftware erkannt wird, Umwandeln des zweiten Sprachsignals in entsprechende gerätebezogene Ansteuerungssignale und entsprechende personenaufbereitete Ansteuerungssignale unter Verwendung der Spracherkennungssoftware.

A further development of the method comprises the following steps:

• Storing a number of first speech signals in a database cooperating with the speech recognition software, the first speech signals corresponding to a number of triggering signals,
• Comparing a second speech signal, which corresponds to a speech command emitted by at least one person in the operating theater, with the first speech signals by means of the speech recognition software, and
• in the event that a sufficient correspondence between one of the first speech signals and the second speech signal is recognized by the speech recognition software, converting the second speech signal into corresponding device-related driving signals and corresponding personalized driving signals using the speech recognition software.

In this embodiment, a number of first speech signals are stored in a database, which correspond to certain voice commands. The person who issues a voice command must give it as well as deposited in the database. For example, if a stored voice command is "turn pump on", the person must formulate the voice command in the same way. In this training, the voice commands can be defined so that they are clearly different, so the risk of confusion can be kept low. In addition, the data volume to be processed can be kept low, whereby a fast processing of the voice commands can be ensured.

• - Hinterlegen einer Anzahl von ersten Sprachsignalen in einer mit der Spracherkennungssoftware zusammenwirkenden Datenbank, wobei die ersten Sprachsignale einer Anzahl von Ansteuerungssignalen entsprechen,
• - Vergleichen eines dritten Sprachsignals, welches einem von zumindest einer im Operationssaal befindlichen Person abgegebenen Wortsequenz entspricht, mit den ersten Sprachsignalen mittels der Spracherkennungssoftware, und
• - für den Fall, dass eine hinreichende Übereinstimmung zwischen einem der ersten Sprachsignalen und dem dritten Sprachsignal von der Spracherkennungssoftware erkannt wird, Umwandeln des zweiten Sprachsignals in entsprechende gerätebezogene Ansteuerungssignale und entsprechende personenaufbereitete Ansteuerungssignale unter Verwendung der Spracherkennungssoftware.

Following the further development of the proposed procedure, the following steps are envisaged:

• Storing a number of first speech signals in a database cooperating with the speech recognition software, the first speech signals corresponding to a number of triggering signals,
• Comparing a third speech signal, which corresponds to a word sequence delivered by at least one person located in the operating theater, with the first speech signals by means of the speech recognition software, and
• in the event that a sufficient coincidence between one of the first speech signals and the third speech signal is recognized by the speech recognition software, converting the second speech signal into corresponding device-related driving signals and corresponding personalized driving signals using the speech recognition software.

In this embodiment, a word sequence emitted by a person in the operating room is compared with the stored first speech signals. It is not necessary to issue the voice command as stored in the database. Instead of "Turn on pump", the word sequence may also read: "The pump should now be switched on". In this embodiment, the device can be operated very intuitively and naturally. The operating physician or assistant does not have to strictly adhere to the stored voice commands. Situations in which a particular medical device can not be accessed because the surgeon or assistant has forgotten the exact wording of the voice command are avoided.

• - Hinterlegen einer Anzahl von personenbezogenen Sprachprofilen in einer mit der Spracherkennungssoftware zusammenwirkenden Datenbank,
• - Vergleichen der von zumindest einer im Operationssaal befindlichen Person abgegebenen Wortsequenz oder des von zumindest einer im Operationssaal befindlichen Person abgegebenen Sprachbefehls mit den hinterlegten, personenbezogenen Sprachprofilen, und
• - für den Fall, dass eine hinreichende Übereinstimmung zwischen dem Sprachbefehl oder der Wortsequenz und einem der hinterlegten, personenbezogenen Sprachprofilen von der Spracherkennungssoftware erkannt wird, Umwandeln des Sprachbefehls oder der Wortsequenz in entsprechende gerätebezogene Ansteuerungssignale und entsprechende personenaufbereitete Ansteuerungssignale unter Verwendung der Spracherkennungssoftware.

A further embodiment of the proposed method comprises the following steps:

• - depositing a number of personal language profiles in a database interacting with the speech recognition software,
• Comparing the word sequence output by at least one person located in the operating room or the voice command issued by at least one person in the operating room with the stored personal voice profiles, and
• in the event that a sufficient correspondence between the voice command or the word sequence and one of the stored personal voice profiles is recognized by the voice recognition software, converting the voice command or word sequence into corresponding device related drive signals and corresponding hand held drive signals using the voice recognition software.

In this embodiment, a voice command is only used to control and / or rules when a sufficiently large match has been recognized with the stored personal voice profile. Consequently, the medical devices are only activated according to the voice commands if the voice command has been issued by a person who can associate the device with a voice profile. For example, a device can only be controlled by the operating doctor or a specific assistant. If the same voice command is issued by another person, to which no voice profile fits, then the control of the medical device is omitted. In this embodiment, it can be prevented that other persons, such as famulants or nurses in training, control the medical devices. The control over the control of the medical devices remains with the specified persons.

• 1 ein Ausführungsbeispiel einer erfindungsgemäßen Vorrichtung anhand einer prinzipiellen Draufsicht,
• 2 einen Operationssaal, welcher mit einer erfindungsgemäßen Vorrichtung betrieben werden kann,
• 3 ein Ablaufdiagramm zum Darstellen eines Teils eines ersten Ausführungsbeispiels eines erfindungsgemäßen Verfahrens, und
• 4A und 4B ein Ablaufdiagramm zum Darstellen eine Teils eines zweiten Ausführungsbeispiels eines erfindungsgemäßen Verfahrens.

Exemplary embodiments of the invention are explained in more detail below with reference to the accompanying drawings. Show it

• 1 An embodiment of a device according to the invention with reference to a basic plan view,
• 2 an operating room, which can be operated with a device according to the invention,
• 3 a flowchart for illustrating a part of a first embodiment of a method according to the invention, and
• 4A and 4B a flow diagram for illustrating a part of a second embodiment of a method according to the invention.

In 1 is an embodiment of a device according to the invention 10 for controlling and / or regulating medical devices 12 in an operating room 14 (please refer 2 ) illustrated by a basic plan view. The device according to the invention 10 includes a voice control module 16 on which speech recognition software executable on a neural network is executable 18 based. The device 10 has a microphone array 34 with at least two microphones 20 on, with the number of microphones 20 can be arbitrarily far over two. The microphone array 34 takes a voice command C or a word sequence W , the one from the one in the operating room 14 person 22 , especially an operating doctor 48 or an assistant who can be delivered up. The microphones 20 or the microphone array 34 convert the recorded voice command C or the word sequence W using a far-field characteristic in a speech signal S um, which via a communication line 24 to the voice control module 16 is transmitted. The communication line 24 Can be wired or wireless. That from the microphone array 34 generated speech signal S is from the voice control module 16 in a device-related control signal SD and in a personalized control signal SP transformed. The device-related control signal SD and the personalization control signal SP are the same in terms of content, but the device-related control signal SD designed so that at least one of the medical devices 12 can be controlled and / or regulated while the personal processing control signal SP is designed so that it can be displayed in acoustic and / or optical form. The device 10 includes a speaker 26 , with which the personalization control signal SP can be output acoustically. Furthermore, the device comprises 10 a screen 28 , with which the personalization control signal SP can be visually displayed.

There is also a radio 30 via a filter device 32 to the voice control module 16 connected. Alternatively, the radio can also be connected to the microphone array 34 be connected, which can be done wired or wireless. The radio 30 serves to lighten the working atmosphere in the operating room 14 , The radio 30 output audio signal SR is about the filter device 32 the voice control module 16 or the microphone array 34 supplied, so that disturbing influences of the audio signal SR be eliminated to the recognition rate of the speech recognition software. Alternatively or cumulatively to the radio, another device generating audio signals, such as an MP3 player, a CD player or the like, which can also be internet-capable and like the radio, can be used 30 over the filter device 32 to the voice control module 16 is connected.

In 2 is an operating room 14 shown by a principal plan view, with the in 1 illustrated device 10 can be operated. The device 10 can be at least partially outside the operating room 14 or completely or partially on the screen 28 be arranged. Other arrangements of the device 10 For example, in a separate unit, are also conceivable.

It can be seen that a microphone array 34 of four microphones in this case 20 on the screen 28 is arranged. In addition, in the operating room 14 following medical devices 12 arranged, which with the device according to the invention 10 controlled and / or regulated: it is an operating table 36 present on which a patient to be operated on 38 is stored. There is also an endoscope 43 to recognize with which the patient 38 can be operated on. Furthermore, an infusion device 40 present with which the patient 38 a medical fluid can be administered. The infusion device 40 is about a feed 41 with a pump 42 connected, with which the amount of the administered liquid can be adjusted. The endoscope 43 is also with the feed 41 connected. Furthermore, in the operating room 14 a recording device 44 , For example, an X-ray machine, arranged with the recordings of the patient 38 can be created. In addition, a lamp 46 in the operating room 14 attached to the necessary illumination of the operating room 14 during the operation. The list of medical devices 12 is not exhaustive.

In the operating room 14 The following persons are present: The operating physician 48 , a medical assistant 50 , a first surgical nurse 52 and a second surgical force 54 as an assistant.

To control, for example, the pump 42 gives the operating doctor 48 the voice command C "Switch on pump", whereby this command can also be delivered in any selectable language, especially in English. This voice command C is from the microphone array 34 recorded and in corresponding voice signals S transformed. The voice control module 16 converts these speech signals S into the corresponding device-related control signals SD and the corresponding personalization control signals SP around. The personalized control signals SP be on the screen 28 visually displayed. Furthermore, the speakers 26 also on the screen 28 arranged so that there are also the personal control signals SP be issued acoustically. Consequently, on the screen 28 "Switch on pump" either in plain text or visually displayed with a corresponding symbol. In addition, the voice message "turn on pump" via the speakers 26 output. This is what they get in the operating theater 14 the person holding the call, if the voice command issued C from the device 10 has been correctly recognized or whether it needs to be corrected. Depending on Embodiment of the device 10 must be the acoustically issued, personalized control signal SP be confirmed before the relevant medical device 12 , in this case the pump 42 , is controlled accordingly. This can be done by the operating doctor saying "yes" as a confirmation command or "no" as a refusal command.

It should be noted that the device 10 while the operation is being operated offline, indicated by the icon D is shown. This does not exclude that the device 10 can be operated online outside of an operation to perform an update or a remote maintenance.

Furthermore, the device 10 be set up so that the speech recognition software is put into a sleep mode, if within a certain time of, for example, 20 seconds no recognizable voice command C or no recognizable word sequence W is delivered. To a medical device 12 with a voice command C or a word sequence W To be able to control, in this embodiment, the speech recognition software must first be put into active mode, which can be done by means of an activation word or an activation word sequence, also referred to as "trigger". This trigger can be chosen arbitrarily and, for example, called "wake up". As confirmation that the speech recognition software has been put into active mode, a voice message or a sound can be played.

In 3 some steps of an embodiment of the method according to the invention are illustrated with reference to a flow chart. In a first step V1 becomes the personalization control signal SP for example, acoustically or optically output. It then waits to see if one of the operating room 14 persons 22 a confirmation command Y or a refusal order N emits. A confirmation command Y can be given by the person 22 "Yes" says (second step V2 ). In this case, the medical device in question 12 in a third step V3 controlled or regulated accordingly.

A refusal order N can be given by the person 22 "No" says (fourth step V4 ). In this case, in a fifth step V5 no action performed.

In the 4A and 4B some steps of a further embodiment of the method according to the invention are illustrated with reference to a flow chart. In a database 56 that interacts with the language recognition software (see 1 ), is a number of personal language profiles P , in this case the language profiles P1 to P5 , deposited. One of them in the operating room 14 person 22 delivered word sequence W or one of them in the operating room 14 person 22 delivered voice command C (first step V1 ) with the deposited language profiles P1 to P6 compared. In the event that there is sufficient agreement between the voice command C or the word sequence W and one of the stored personal language profiles P1 to P5 is recognized by the speech recognition software, the voice command C or the word sequence W in corresponding device-related control signals SD and personalization control signals SP converted (second step V2 ). In the example shown, there is a match with the first voice profile P1 in front. If no sufficient match is found (fourth step V4 ), no action takes place (fifth step V5 ).

The voice control can be set up in the following ways: In a first device, it must be in the operating theater 14 person 22 a voice command C as well as the corresponding first speech signals S1 in the database 56 are deposited (step V1 ). Im in 4A shown example are a total of five first speech signals S11 to S15 stored, which in each case a device-related control signal SD and a personal control signal SP correspond. For example, if the third is the first voice signal S13 , which is to turn on the pump 42 performs as "turn on pump" in the database 56 deposited, the person must 22 the voice command C just as well as formulating "turn on pump". The thus generated second signals S2 are then with the deposited first speech signals S11 to S15 compared and with sufficient agreement in corresponding control signals SD . SP transformed ( V2 . V3 ). If a sufficient match is not found, no action is taken ( V4 . V5 ).

In a second facility, the person can 22 the voice command C formulate freely while doing a word sequence W use, for example "the pump should now be switched on" (step V1 ). The third signals thus generated S3 , the word sequence W correspond, are then with the stored first speech signals S11 to S15 compared and with sufficient agreement in corresponding control signals SD . SP transformed ( V2 . V3 ). If a sufficient match is not found, no action is taken ( V4 . V5 ). It makes sense, the personal control signals SP as voice commands C and not the word sequence W to repeat. Even if the person 22 the word sequence W "The pump should now is turned on, "pump on" is optically or acoustically output to give unambiguous feedback as to what voice command C from the word sequence W has been recognized.

As hardware, a "Digital Signal Processor" (DSP) can be used. The signals are converted to DSP hardware that has the following properties: adaptive beamforming (position determination of sources in wave fields (eg sound fields)), "Adaptive Echo Cancellation", "Noise Reduction" , "Automatic Gain Control", far-field characteristics and microphone solutions with a sufficiently high signal-to-noise ratio.

LIST OF REFERENCE NUMBERS

10

contraption

12

medical device

14

operating room

16

Voice control module

18

neural network

20

microphone

22

person

24

communication line

26

speaker

28

screen

30

radio

32

filtering device

34

array

36

operating table

38

patient

40

Infusion device

41

Intake

42

pump

43

endoscope

44

recording apparatus

46

lamp

48

operating doctor

50

intern

52

first surgical nurse

54

second surgical nurse

56

Database

C

voice command

D

symbol

N

rejection command

S

speech signal

S1, S11 to S15

first speech signal

S2

second speech signal

S3

third speech signal

SD

device-related control signal

SP

personalized control signal

SR

audio signal

P, P1 to P5

language profile

V, V1 to V5

steps

W

word sequence

Y

confirmation command

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 2015/0364135 A1 [0004]
• EP 1868184 A1 [0004]

Claims (10)

Device for controlling and / or regulating medical devices (12) in an operating room (14), comprising a voice control module (16) on which a speech recognition software based on a neural network (18) can be executed offline, at least one microphone array (34) which can be connected or connected to the voice control module (16) and has at least two microphones (20) for providing a far-field characteristic, with which voice commands issued by at least one person (22) in the operating room (14) C) and can be converted into speech signals (S) that can be processed by the speech control module (16), wherein the speech control module (16) is set up in such a way that the speech signals (S) are converted into device-related control signals (SD) and personalized control signals (SP) using the speech recognition software. are convertible, - At least one medical device (12) which is connectable or connected to the voice control module (16) and which is controllable and / or controllable with the device-related control signals (SD), and - At least one speaker (26), with which the personalized control signals (SP) can be output acoustically. Device after Claim 1 , characterized in that the device comprises a radio (30) or other device generating audio signals (SR) which is connectable or connected to the voice control module (16) or to the microphone array (34), the device (10 ) comprises filter means (32) for filtering the audio signals (SR). Device according to one of Claims 1 or 2 , characterized in that the device comprises at least one screen (28) for displaying operation-related data, with which the personalized control signals (SP) can be displayed optically. Device after Claim 3 , characterized in that the microphone array (34) with two or more microphones (20) on the screen (28) is arranged. Device according to one of the preceding claims, characterized in that the medical devices (12) pumps (42), endoscopes (43), lamps (46), operating tables (36), recording devices (44) or the like. Method for controlling and / or regulating medical devices (12) in an operating room (14) with a device (10) according to one of the preceding claims, comprising the following steps: Offline execution of a speech recognition software based on a neural network (18) during an operation by means of a speech control module (16), Recording at least one voice command (C) emitted by at least one person (22) located in the operating room (14) by means of a microphone array (34) connected to the voice control module (16) providing a far-field characteristic, Converting the recorded voice command (C) into speech signals (S) that can be processed by the voice control module (16) by means of the microphone array (34), Converting the speech signals (S) into device-related drive signals (SD) and personal-processed drive signals (SP) using the speech recognition software, Controlling and / or regulating at least one medical device (12) with the device-related drive signals (SD), and - Acoustic output of the personal processing control signals (SP) by means of a loudspeaker (26). Method according to Claim 6 comprising the steps of: - acoustically outputting some or all of the personalization control signals (SP) by means of a loudspeaker (26), - receiving a confirmation command (Y) or a refusal order (N) from at least one person in the operating room (14) (22) and, in the event that the person (22) in the operating room (14) issues a confirmation command (Y), controlling and / or regulating at least one medical device with the device-related control signals (SD). Method according to one of Claims 6 or 7 comprising the following steps: - storing a number of first speech signals (S1) in a database (56) interacting with the speech recognition software, wherein the first speech signals (S1) correspond to a number of drive signals (SD, SP), - comparing a second speech signal (S1) S2) corresponding to a voice command (C) issued by at least one person (22) in the operating room (14), the first voice signals (S1) by means of the speech recognition software, and, in the event that there is a sufficient match between one of the first Speech signals (S1) and the second speech signal (S2) is recognized by the speech recognition software, converting the second speech signal (S2) into device-related drive signals (SD) and personal processing control signals (SP) using the speech recognition software. Method according to one of Claims 6 to 8th comprising the following steps: - storing a number of first speech signals (S1) in a database (56) interacting with the speech recognition software, wherein the first speech signals (S1) correspond to a number of drive signals (SD, SP), - comparing a third speech signal (S1) S3) corresponding to a word sequence (W) delivered by at least one person (22) in the operating room (14), with the first speech signals (S1) by means of the speech recognition software, and - in the event that there is a sufficient match between one of the first Speech signals (S1) and the third speech signal (S3) is recognized by the speech recognition software, converting the second speech signal (S2) into device-related drive signals (SD) and personalization drive signals (SP) using the speech recognition software. Method according to one of Claims 6 to 9 comprising the following steps: - depositing a number of personal language profiles (P) in a database (56) interacting with the speech recognition software, - comparing the word sequence (W) or speech generated by at least one person (22) located in the operating room (14) at least one voice command (C) delivered with the person (22) in the operating room (14) with the stored personal voice profiles (P), and - in the event that a sufficient match between the voice command (C) or the word sequence (W) and one of the deposited personal voice profiles (P) is recognized by the voice recognition software, converting the voice command (C) or the word sequence (W) into corresponding device related drive signals (SD) and corresponding personalized drive signals (SP) using the voice recognition software.

Images