JP4005320B2 - Medical control system - Google Patents

Description

予め入力されている情報と一致していると判定された場合（表１の判定レベルが曖昧以上の類似度８０％以上の場合）には、ＣＰＵ７３は、ステップＳ６において類似度算出回路１０４により類似度レベルの判定が行われ、「曖昧」レベル以下の場合（類似度７９％以下の場合）ではＣＰＵ７３は音声指示回路１０８を制御し再発声要求指示メッセージを表示装置１０６に表示させ再度発声が促しステッブＳ２に戻る。
【００３０】
また、ステップＳ６で「一致」と判断すると、ＣＰＵ７３は、ステップＳ７で話者照合を開始する。ステップＳ８、９では上記ステップＳ４，５と同じ作用が行われ、続いてステップ１０で入力された音声がメインドクターの音声であるかが表２に基づいて判断される。
【００３１】
【表２】

このとき「メインドクター本人」のレベルと判断すると、ＣＰＵ７３は、ステップ１１で所定時間経過したか判断し、経過していなければ次のステップＳ１２に移行し、所定時間経過してしまったらステップＳ２に戻り再度音声入力する。
【００３２】
なお、上記ステップＳ２〜Ｓ６とステップＳ７〜Ｓ１０のステップは実際には並行して行った方が効率的である。
【００３３】
続いてＣＰＵ７３は、不特定話者認識エンジン７１を作動させ、ステップ１２でメインドクターより表３に示すような機能を操作するキーワードを順次メインドクター用マイク６１で音声登録を行う。
【００３４】
【表３】

ステッブＳ１２で、例えば、「電気メス」とメインドクターがメインドクター用マイク６１を使い発声すると、音声入力回路１００を介してその音声情報は次段の特微量抽出回路１０２にて音声の特徴抽出が行われ、記憶回路１０３のデータエリア内のアドレス（１ａ、１ｂ）に「電気メス」の特徴データが記録・更新される。
【００３５】
この処理は、必要なキーワード登録が完了するまで繰り返される。表３は９個のキーワード（単語）を登録した状態を示している。
【００３６】
このようにして、メインドクターによるキーワード登録が終了し、例えば、術中にメインドクターが電気メスの設定を変更する場合には、ＣＰＵ７３は、ステップＳ１３からステップＳ１４に処理を移行する。
【００３７】
すなわち、設定変更を行う本人がメインドクター用マイク６１に向かい「電気メス」と発声を行うと、ＣＰＵ７３は、不特定話者認識エンジン７１を作動させ、ステップＳ１３で特微量抽出回路１０２にて音声の特徴抽出を行うと共に、ステップＳ１４で類似度算出回路１０４にて既に登録がされている記憶回路１０３内より入力音声である「電気メス」に近い特徴データを探し出し、類似度算出を行う。
【００３８】
そして、ＣＰＵ７３は、ステップＳ１５で類似度を判定し、音声が登録されていない「他人」のものであるかどうか判断する。ここで、類似度は、例えば、表２に示すような、レベル付けが予め行われている。また、類似度は任意に設定が可能である。類似度を高くすればより「他人」としての排他率を高めることができ、逆に本人であったとしても排他される可能性が高まることになる。
【００３９】
ステップＳ１５でＣＰＵ７３の判断により本人ではない（キーワード（単語）登録を行った本人ではない状態：「他人」）と判定された場合で、表２の「他人」レベルの場合にはエラーメッセージが表示装置１０６に表示され、再度発声を促されることになり、ステップＳ １ １に戻る。
【００４０】
また、ＣＰＵ７３は、ステップＳ１ ５で表２の「本人」レベルと判断すると、ステップＳ１６において本人であることの同定を行い、その後の音声情報によりコントロール回路１０７を介して周辺機器１０５の設定、変更が行われ、表示装置１０６の表示の変更も行われる。
【００４１】
一方、ステップＳ１０でメインドクターでないと判断されると、ＣＰＵ７３は、ステップ１７で所定時間経過したか判断し、経過していなければ次のステップＳ１８に移行し、一定時間経過してしまったらステップＳ２に戻り再度音声入力する。
【００４２】
一定時間経過していない場合は、サブドクター又はナースはステップＳ１８で表３に示す機能を操作するキーワードを順次サブドクター用マイク６２またはナース用マイク６３で音声登録する。
【００４３】
ステップＳ１８で「電気メス」とサブドクター用マイク６２またはナース用マイク６３を使い発声すると、ＣＰＵ７３は、不特定話者認識エンジン７１を作動させ、ステップＳ１９では特微量抽出回路１０２にて音声入力回路１００を介した音声情報の特徴抽出を行い、記憶回路１０３のデータエリア内のアドレス（１ａ、ｌｂ）に「電気メス」の特徴データを記録・更新する。この処理は、必要なキーワード登録が完了するまで繰り返される。表３は、９個のキーワード（単語）を登録した状態を示している。
【００４４】
このようにして、サブドクター又はナースによるキーワード登録は終了し、例えば、術中にサブドクターまたはナースが電気メスの設定を変更する場合には、ステップＳ１９からステップＳ２０に処理が移行する。
【００４５】
すなわち、設定変更を行う本人がサブドクター用マイク６２またはナース用マイク６３に向かい「電気メス」と発声を行うと、ＣＰＵ７３は、不特定話者認識エンジン７１を作動させ、ステップＳ１９で特微量抽出回路１０２にて音声の特徴抽出を行うと共に、ステップＳ２Ｏで類似度算出回路１０４にて既に登録がされている記憶回路１０３内より入力音声である「電気メス」に近い特徴データを探し出し、類似度算出を行い、ステップＳ２１でＣＰＵ２９が類似度を判定し、音声が登録されていない「他人」のものであるかどうか判断する。
【００４６】
ここで、類似度は、例えば、表２に示すようなレベル付けが予め行われている。また、類似度は任意に設定が可能であり、類似度を高くすればより「他人」としての排他率を高めることができ、逆に本人であったとしても排他される可能性が高まることになる。
【００４７】
ステップＳ２１でＣＰＵ７３の判断により、本人ではないと判定された場合で、表２の「他人」レベルの場合には、エラーメッセージが表示装置１０６に表示され、再度発声を促されることになり、ステップＳ１７に戻る。
【００４８】
類似度レベルが規定値以上であると、次のステッブＳ２２に進み、メインドクターの指示を待つことになる。
【００４９】
ここでメインドクターがサブドクター又はナースに音声操作を許可するには、ステップＳ２３で、メインドクターがステップ２で登録した「ＯＫ」を音声入力する。この音声入力をすることにより、ＣＰＵ７３は、ステップＳ２４で特定話者照合部７０を作動させ、ステップＳ２５で特徴量抽出回路１０２によって音声の特徴抽出が行われると共に、ステップＳ２６で類似度算出回路１０４にすでに登録がされている記憶回路１０３内より入力音声である「ＯＫ」に近い特徴データーを探し出し類似度算出を行う。
【００５０】
そして、ＣＰＵ７３は、ステップＳ２７で入力音声が承認するキーワードであるか判断し、異なる場合はステップＳ２２に戻り、再度音声入力を待ち、一致している場合はステップＳ２８でＣＰＵ７３は話者照合を行う。
【００５１】
ステップＳ２９、３０は、上記と同じ作用が行われる。すなわち、ＣＰＵ７３は、ステップＳ２９で特定話者照合部７０の特微量抽出回路１０２にて音声の特徴抽出を行うと共に、ステップＳ３０で類似度算出回路１０４にて既に登録がされている記憶回路１０３内より入力音声である「ＯＫ」に近い特徴データを探し出し、類似度算出を行う。
【００５２】
そして、ＣＰＵ７３は、ステップＳ３１で類似度を判定し、音声が登録されていない「他人」のものであるかどうか判断する。ここで、類似度は、例えば、表２に示すような、レベル付けが予め行われている。また、類似度は任意に設定が可能である。類似度を高くすればより「他人」としての排他率を高めることができ、逆に本人であったとしても排他される可能性が高まることになる。
【００５３】
そして、ＣＰＵ７３は、ステップＳ３１で「ＯＫ」の音声がメインドクターであるか判断し、一致していないときはステップＳ２２に戻り再度入力し、ステップＳ１６に進み一致していればサブドクター又はナースが音声で医療機器を操作できるようになる。
【００５４】
医療機器をサブドクター又はナースが操作している途中でメインドクターが操作を中止させたいときは、ステップＳ２３で「ＮＯ」を入力することにより、上記のステップを踏みサブドクター及びナースの医療機器の操作を中止させることができる。
【００５５】
このように本実施の形態では、不特定話者認識エンジン７１により特定話者の音声（メインドクター）と不特定話者（サブドクター又はナース）の音声を確実に識別し、また不特定話者認識エンジン７１及び特定話者照合部７０により特定の話者（メインドクター）が意図しない操作を受け付けないようにすることができる。
【００５６】
図８は本発明の第２の実施の形態に係るシステムコントローラの構成を示すブロック図である。
【００５７】
第２の実施の形態は、第１の実施の形態とほとんど同じであるので、異なる点のみ説明し、同一の構成には同じ符号をつけ説明は省略する。
【００５８】
本実施の形態は、図８に示すように、第１の実施の形態の特定話者照合部７０に代わり特定話者認識エンジン１２０を設けたものである。
【００５９】
多くの医療器具を使用する内視鏡手術に於いて、スムーズな手術進行のためには術者が滅菌域から機器操作を行えることが重要である。
【００６０】
内視鏡外科手術ではＴＶカメラや光源のように患者に直接関係が無いローリスク機器と、電気メスや超音波処置装置のように患者に直接関係するハイリスク機器が混在して使用されている。不特定話者エンジンを用いたシステムは術者以外の音声で操作できるため術者が意図しない操作が行われることがあり、術者は術者以外の音声操作を常に監視しなければならない。
【００６１】
そこで、本実施の形態は、特定話者エンジンと不特定話者エンジンの両方を備え、内視鏡外科手術特有の機器のリスクに応じて認識エンジンを使い分けることで、術者の負担を軽減できるようにしたものである。
【００６２】
具体的には、特定話者認識エンジン１２０にメインドクターがハイリスクの機器、例えば電気メス３２、気腹器３３を操作する機能操作単語を認識させ、不特定話者認識エンジン７１にはメインドクター、サブドクター及びナースがローリスクの機器、例えば第１ＴＶカメラ２３ａ，第２ＴＶカメラ２３ｂ，第１光源２４ａ，第２光源２４ｂを操作する機能操作単語を認識させることにより、上記目的を達成させることができる。
【００６３】
処理の流れについては第１の実施の形態と同じ流れ（図４ないし図７参照）で操作をすればメインドクター、サブドクターとも機器の操作ができるので説明を省略する。
【００６４】
［付記］
（付記項１） 前記医療機器は、内視鏡外科手術特有のリスクが高い機器及びリスクが低い機器である
ことを特徴とする請求項１に記載の医療用制御システム。
【００６５】
（付記項２） 前記リスクが高い機器は、電気メス等のエネルギーを使用する外科手術装置である
ことを特徴とする付記項１に記載の医療用制御システム。
【００６６】
（付記項３） 前記リスクが低い機器は、ＴＶカメラ、光源装置等の観察装置である
ことを特徴とする付記項１に記載の医療用制御システム。
【００６７】
（付記項４） 前記特定の話者は、内視鏡外科手術のメインドクターである
ことを特徴とする請求項１に記載の医療用制御システム。
【００６８】
（付記項５） 前記不特定の話者は、内視鏡外科手術のサブドクター又はナースである
ことを特徴とする請求項１に記載の医療用制御システム。
【００６９】
（付記項６） 所定の制御信号を入力し、咳制御信号に基づき所定の医療機器の機能を制御する医療機器制御手段と、
特定の話者が発する特定の音声を認識する特定話者認識手段と、
不特定の話者が発する特定の音声を認識する不特定話者認識手段と、
前記特定話者と前記不特定話者の切り替えを行うフットスイッチと
を備えたことを特徴とする医療用制御システム。
【００７０】
（付記項７） 一つあるいは複数の医療装置と、
特定話者認識エンジンと不特定話者認識エンジンとを有する前記医療装置の複数の機能を音声により操作可能な音声操作手段と、
前記医療装置の複数の機能に応じて前記音声操作手段が使用する前記特定話者認識エンジンと不特定話者認識エンジンとを切り替る切り替え手段と
を備えたことを特徴とする医療用制御システム。
【００７１】
（付記項８） 所定の制御信号を入力し、該制御信号に基づき所定の医療機器の機能を制御する医療機器制御手段と、
特定の第１の話者が発する特定の音声を認識する第１の話者認識手段と、
前記第１の話者とは異なる第２の話者が発する特定の音声を認識する第２の話者認識手段と、
前記第２の話者認識手段で前記第２の話者の音声が認識されたときに、前記第１の話者認識手段で認識された前記第１の話者の音声に基づく制御信号を前記医療機器制御手段に送出する制御信号送出手段と
を備えたことを特徴とする医療用制御システム。
【００７２】
【発明の効果】
以上説明したように本発明によれば、特定話者の音声と不特定話者の音声を確実に識別し、音声により複数の機器を容易に操作すると共に、特定の話者の意図しない操作を確実に防止することができるという効果がある。
【図面の簡単な説明】
【図１】本発明の第１の実施の形態に係る外科手術システムの構成を示す構成図
【図２】図１のシステムコントローラの構成を示すブロック図
【図３】図２の特定話者照合部及び不特定話者認識エンジンの内部構造示すブロック図
【図４】図２のシステムコントローラの作用を説明する第１のフローチャート
【図５】図２のシステムコントローラの作用を説明する第２のフローチャート
【図６】図２のシステムコントローラの作用を説明する第３のフローチャート
【図７】図２のシステムコントローラの作用を説明する第４のフローチャート
【図８】本発明の第２の実施の形態に係るシステムコントローラの構成を示すブロック図
【符号の説明】
１０…外科手術システム
２５…システムコントローラ
６１…メインドクター用マイク
６２…サブドクター用マイク
６３…ナース用マイク
７０…特定話者照合回路
７１…不特定話者認識エンジン
７２…通信Ｉ／Ｆ
７３…ＣＰＵ
７４…フットスイッチ
１００…音声入力回路
１０１…音声入力部
１０２…特微量抽出回路
１０３…記憶回路
１０４…類似度算出回路
１０５…周辺装置
１０６…表示装置
１０７…コントロール回路
１０８…音声指示回路[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a medical control system that controls a plurality of devices by voice.
[0002]
[Prior art]
2. Description of the Related Art In recent years, endoscopes that can observe various organs in a body cavity by inserting an elongated insertion portion into the body cavity or perform various therapeutic treatments using a treatment tool as necessary have been widely used.
[0003]
As a surgical operation apparatus for performing a surgical operation under such an endoscope, there are, for example, Japanese Patent Application No. 2000-319815 and Japanese Patent Application Laid-Open No. 6-178756 filed earlier by the present applicant.
[0004]
The surgical operation apparatus of Japanese Patent Application No. 2000-319815 is a personal identification device for specifying a person who has made a predetermined voice, a memory means for holding a plurality of voices uttered by a specific person as stored voices, and a predetermined input Identification means for identifying the user's voice input from the means, and the user's voice identified by the identification means is collated with the stored voice held in the memory means, and the user's voice is identified Determining means for determining whether or not a person has uttered, and at least one of the plurality of stored sounds stored in the memory means when the determining means determines that the sound is not emitted by the specific person Selection means for selecting the stored voice and notification means for notifying the user to generate the same voice as the stored voice selected by the selection means.
[0005]
In the surgical operation apparatus disclosed in Japanese Patent Laid-Open No. 6-178756, a voice input start button is provided in the operation unit of the electronic endoscope, and a microphone is provided in the operation unit. The operation term is identified by the voice signal inputted in the above manner, and the control corresponding to the identified operation term is executed. For example, the operation term such as freeze, VTR recording, VTR stop, video print, etc. Once identified, these operations are performed. Therefore, the operation can be performed by voice without using many operation buttons.
[0006]
[Problems to be solved by the invention]
However, surgical procedures are generally performed by multiple people, such as a main doctor, a sub-doctor, and a nurse. In the surgical device of the above Japanese Patent Application No. 2000-319815, the voice of a specific speaker is recognized and the medical device used for the surgery is operated. Since the medical device can be operated even with the voice of a specific speaker, there is a possibility that an operation unintended by the specific speaker may be performed.
[0007]
Further, in the surgical operation apparatus disclosed in Japanese Patent Laid-Open No. 6-178756, only a technique for voice operation of a device having a low risk in the procedure such as freeze, VTR recording, and video printer is disclosed, and the risk in the procedure is high. When devices (for example, electric scalpels) and low-risk devices are mixed, it is possible for everyone to perform voice operations on high-risk devices, and the surgeon performs unintended voice operations on high-risk devices. There is a problem that the operator's burden is increased because it is necessary to constantly monitor voice operations other than the surgeon.
[0008]
The present invention has been made in view of the above circumstances, and can reliably identify the voice of a specific speaker and the voice of an unspecified speaker, easily operate a plurality of devices using the voice, An object of the present invention is to provide a medical control system that can reliably prevent an unintended operation.
[0009]
[Means for Solving the Problems]
A medical control system according to an aspect of the present invention includes a medical device control unit that inputs a predetermined control signal and controls a function of the predetermined medical device based on the predetermined control signal;
voice Approve the operation For voice input of the main doctor The medical device can be operated with permission speaker The first keyword for controlling the predetermined medical device issued from the first keyword is determined by determining the similarity with the registered first keyword. Whether or not Unspecified speaker recognition means,
The unspecified speaker recognition means 1 keyword When it is determined that the voice of the first keyword By the prescribed medical device of Whether to allow control Was issued as Second keyword different from the first keyword Determines whether or not is due to the voice input of the main doctor approving the voice operation Specific speaker recognition means;
It has.
[0010]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0011]
1 to 7 relate to the first embodiment of the present invention, FIG. 1 is a block diagram showing the configuration of the surgical system, FIG. 2 is a block diagram showing the configuration of the system controller of FIG. 1, and FIG. 2 is a block diagram showing the internal structure of the specific speaker verification unit 2 and the non-specific speaker recognition engine, FIG. 4 is a first flowchart illustrating the operation of the system controller of FIG. 2, and FIG. 5 is the operation of the system controller of FIG. FIG. 6 is a third flowchart for explaining the operation of the system controller of FIG. 2, and FIG. 7 is a fourth flowchart for explaining the operation of the system controller of FIG.
[0012]
As shown in FIG. 1, in the surgical operation system 10 for performing a surgical operation using the endoscope 1, there are many types of peripheral medical devices to be used, and various medical devices cannot be stored in one cart. These multiple medical devices are divided into several carts 20 and 30 and mounted, while these carts are collected in almost one place to improve operability.
[0013]
For example, the first cart 20 includes a monitor 21, a central display panel 22, a first TV camera 23a, a first light source 24a, a second TV camera 23b, a second light source 24b, a system controller 25, a video mixer 26, a VTR 27, and a distributor. 28, a communication connector 29 and the like are mounted, and the second cart 30 is equipped with a monitor 31, an electric knife 32, an insufflator 33, a CO2 cylinder 34, a suction bottle 35, a distributor 36, a communication connector 37 and the like. Has been.
[0014]
Various medical devices are electrically connected to the distributors 28 and 36 disposed in the carts 20 and 30 via communication cables (not shown) in the first cart 20 and the second cart 30. Has been. The first cart 20 and the second cart 30 are electrically connected via a universal cable 51 provided with a communication cable. Further, the first cart 20 and the second cart 30 and the peripheral device controller 40 are electrically connected via a universal cord 52 provided with a communication cable.
[0015]
In the peripheral device controller 40, setting switches that are frequently used among the medical devices mounted on the first cart 20 and the second cart 30 are collected in a central control operation unit 41 that is a central control input unit. It has come to be. In the central control operation unit 41, setting switches are displayed on a monitor provided with a touch panel, and the setting switches can be selectively pressed by a touch panel provided on the monitor.
[0016]
That is, the central control operation unit 41 aggregates only frequently used switches related to the settings of each medical device, and outputs control to avoid erroneous operation of the output unit that controls the irradiation of energy to the human body. It is set as the structure which provides a part separately.
[0017]
A main doctor microphone 61, a sub doctor microphone 62, and a nurse microphone 63 are connected to the system controller 25 so that each can control a medical device.
[0018]
FIG. 2 is a block diagram showing a schematic configuration of the system controller 25. The system controller 25 performs a function operation from the specific speaker verification circuit 70 in which the main doctor registers a word for starting voice operation from the main doctor microphone 61, the main doctor microphone 61, the sub-doctor microphone 62, and the nurse microphone 63. The unspecified speaker recognition engine 71 for recognizing the word, the communication I / F 72 that manages communication with an external device, and the CPU 73 that controls these circuits, the main doctor microphone 61, the sub doctor microphone 62, and For example, the first TV camera 23 a, the second TV camera 23 b, the first light source 24 a, the second light source 24 b, the electric knife 32, and the insufflator 33 can be controlled by the sound from the nurse microphone 63.
[0019]
Further, a foot switch 74 is connected to the CPU 73, and when the main doctor steps on the foot switch 74, the CPU 73 controls so that the sub-doctor and the nurse cannot control the device. In other words, by operating the foot switch 74, the sub-doctor and nurse can be set as a specific speaker or non-specific speaker. When the sub-doctor and nurse are set as non-specific speakers, the system controller 25 Recognition as a specific speaker by voice, which will be described later, is rejected.
[0020]
Specifically, as shown in FIG. 3, the system controller 25 includes an audio input unit 101 including a main doctor microphone 61, a sub doctor microphone 62, a nurse microphone 63, and an audio input circuit 100 for inputting audio. A feature quantity extraction circuit 102 that extracts the feature quantity of the voice input from the voice input unit 101; a storage circuit 103 that records the feature quantity of the keyword extracted by the feature quantity extraction circuit 102 as feature quantity data; The similarity calculation circuit 104 that calculates the similarity of speech from the feature amount of the voice extracted by the micro extraction circuit 102 and the keyword feature data stored in the storage circuit 37, and the periphery of the electric knife 32, the abdominal stomach 33, etc. The device 105 is controlled, and a desired message or the like is displayed on the display device 106 such as the central display panel 22 and the central control operation unit 41. Based on the calculation results of the control circuit 107 and the similarity calculation circuit 104, voice identification identification is performed and the control circuit 107 is controlled based on the voice feature extracted by the feature extraction circuit 102 when the identity is identified. The CCU 73 and a voice instruction circuit 108 that instructs the control circuit 107 to display a message prompting voice input on the display device 106 under the control of the CCU 73 are configured.
[0021]
The feature extraction circuit 102, the storage circuit 103, and the similarity calculation circuit 104 show the internal structure of the specific speaker verification unit 70 and the unspecified speaker recognition engine 71 shown in FIG. A circuit is necessary (not shown in FIG. 3 to enhance the understanding of the explanation).
[0022]
A circuit corresponding to the communication I / F 72 in FIG. 2 includes a control circuit 107 and a voice instruction circuit 108.
[0023]
When the registration switch (not shown) is turned on, the CCU 73 instructs the voice instruction circuit 108 to prompt a keyword registration, and records the input amount of the keyword in the storage circuit 103 as the amount data. In this way, the feature quantity extraction circuit 102 is controlled. Further, the CCU 73 outputs to the control circuit 107 a control signal corresponding to the sound information for controlling the peripheral device 105 input based on the feature amount from the feature amount extraction circuit 102 at the time of identification. Thereby, the control circuit 107 performs various settings for the peripheral device 105 in accordance with the input voice.
[0024]
Next, the operation of the system controller 25 of the present embodiment configured as described above will be described.
[0025]
As shown in FIGS. 4 to 7, when a keyword registration switch (not shown) is turned on in step S1 and voice recognition by the main doctor is started, the CPU 73 controls the voice instruction circuit 108 in step S2 to input a keyword. A prompt message is displayed on the display device 106.
[0026]
This message needs to be different from the keyword for operating the peripheral device 105.
[0027]
Therefore, for example, “OK” is displayed as a keyword for approving that the main doctor operates the sub-doctor with voice, “NO” is displayed as a keyword when not approved, and “START” is displayed as a keyword for starting voice operation. The main doctor speaks all the displayed keywords.
[0028]
Subsequently, the CPU 73 operates the specific speaker verification unit 70 in step S3 to start speech recognition. In step S4, the feature extraction circuit 102 performs voice feature extraction, and in step S5, the similarity calculation circuit. Based on Table 1, feature data close to “OK”, “NO”, and “START” as input voices are searched from the storage circuit 103 already registered in 104.
[0029]
[Table 1]

When it is determined that the information matches the information input in advance (when the determination level in Table 1 is 80% or more of a similarity level that is more than ambiguous), the CPU 73 uses the similarity calculation circuit 104 to perform similarity in step S6. When the degree level is determined and the level is less than the “ambiguous” level (when the degree of similarity is 79% or less), the CPU 73 controls the voice instruction circuit 108 to display the recurrent voice request instruction message on the display device 106 and prompts the voice again. Return to step S2.
[0030]
On the other hand, if it is determined as “match” in step S6, the CPU 73 starts speaker verification in step S7. In steps S8 and S9, the same operation as in steps S4 and S5 is performed, and it is determined based on Table 2 whether the voice input in step S10 is the voice of the main doctor.
[0031]
[Table 2]

At this time, if it is determined that the level is “main doctor”, the CPU 73 determines in step 11 whether or not a predetermined time has elapsed. If not, the CPU 73 proceeds to the next step S12, and if the predetermined time has elapsed, proceeds to step S2. Return and input voice again.
[0032]
It should be noted that the steps S2 to S6 and steps S7 to S10 are actually more efficient when performed in parallel.
[0033]
Subsequently, the CPU 73 activates the unspecified speaker recognition engine 71, and in step 12, the main doctor sequentially performs voice registration of keywords for operating functions as shown in Table 3 using the main doctor microphone 61.
[0034]
[Table 3]

In step S12, for example, when the “electric knife” and the main doctor utter using the main doctor microphone 61, the voice information is extracted by the feature extraction circuit 102 in the next stage via the voice input circuit 100. The characteristic data of “electric knife” is recorded / updated at addresses (1a, 1b) in the data area of the storage circuit 103.
[0035]
This process is repeated until necessary keyword registration is completed. Table 3 shows a state in which nine keywords (words) are registered.
[0036]
Thus, the keyword registration by the main doctor is completed. For example, when the main doctor changes the setting of the electric knife during the operation, the CPU 73 shifts the process from step S13 to step S14.
[0037]
That is, when the person who changes the setting speaks to the main doctor microphone 61 and speaks “electric knife”, the CPU 73 activates the unspecified speaker recognition engine 71, and the special trace extraction circuit 102 performs voice in step S 13. In step S14, feature data close to the “electric knife” that is the input voice is searched from the storage circuit 103 already registered in the similarity calculation circuit 104, and the similarity is calculated.
[0038]
Then, the CPU 73 determines the degree of similarity in step S15, and determines whether the voice is of “others” whose voice is not registered. Here, the similarity is leveled in advance as shown in Table 2, for example. Also, the similarity can be arbitrarily set. If the degree of similarity is increased, the exclusion rate as “another person” can be increased, and conversely, even if the person is the person himself, the possibility of exclusion is increased.
[0039]
If it is determined in step S15 that the user is not the person (the person who is not the person who registered the keyword (word): “other person”), and an error message is displayed in the case of the “other person” level in Table 2. It will be displayed on the device 106 and will be prompted to speak again, and the process returns to step S 1 1.
[0040]
If the CPU 73 determines that it is the “person” level in Table 2 in step S 15, it identifies the person in step S 16 and sets or changes the peripheral device 105 via the control circuit 107 based on the subsequent audio information. The display on the display device 106 is also changed.
[0041]
On the other hand, step S 10 If it is determined at step 17 that the predetermined time has not elapsed, the CPU 73 determines whether or not a predetermined time has elapsed. If not, the CPU 73 proceeds to the next step S18. input.
[0042]
If the predetermined time has not elapsed, the sub-doctor or nurse sequentially registers the keywords for operating the functions shown in Table 3 using the sub-doctor microphone 62 or the nurse microphone 63 in step S18.
[0043]
When the “electrical knife” and the sub-doctor microphone 62 or the nurse microphone 63 are used to utter in step S18, the CPU 73 activates the unspecified speaker recognition engine 71. The feature information of the voice information via 100 is extracted, and the feature data of “electric knife” is recorded / updated at the address (1a, lb) in the data area of the storage circuit 103. This process is repeated until necessary keyword registration is completed. Table 3 shows a state in which nine keywords (words) are registered.
[0044]
In this way, keyword registration by the sub-doctor or nurse is completed. For example, when the sub-doctor or nurse changes the setting of the electric knife during the operation, the process proceeds from step S19 to step S20.
[0045]
That is, when the person who changes the setting speaks “electric knife” toward the sub-doctor microphone 62 or the nurse microphone 63, the CPU 73 activates the unspecified speaker recognition engine 71 and extracts the characteristic amount in step S19. The feature extraction of the voice is performed by the circuit 102, and the feature data close to the “electric knife” which is the input voice is searched from the storage circuit 103 already registered by the similarity calculation circuit 104 in step S2O. In step S21, the CPU 29 determines the degree of similarity, and determines whether the sound belongs to “others” who are not registered.
[0046]
Here, for example, leveling as shown in Table 2 is performed in advance on the similarity. In addition, the degree of similarity can be set arbitrarily, and if the degree of similarity is increased, the exclusion rate as “others” can be increased, and conversely, even if the person is the person himself, the possibility of being excluded increases. Become.
[0047]
If it is determined by the CPU 73 in step S21 that the person is not the person and the “other person” level in Table 2 is displayed, an error message is displayed on the display device 106, and the user is prompted to speak again. Return to S17.
[0048]
If the similarity level is greater than or equal to the specified value, the process proceeds to the next step S22 and waits for an instruction from the main doctor.
[0049]
Here, the main doctor allows the sub-doctor or nurse to operate the voice. S23 Then, the main doctor inputs “OK” registered in step 2 by voice. By inputting this voice, the CPU 73 activates the specific speaker verification unit 70 in step S24, and the voice feature extraction is performed by the feature amount extraction circuit 102 in step S25, and the similarity calculation circuit 104 in step S26. The feature data close to “OK”, which is the input voice, is searched from the storage circuit 103 that has already been registered, and the similarity is calculated.
[0050]
Then, the CPU 73 determines whether or not the input voice is a keyword to be approved in step S27, and if different, returns to step S22, waits for voice input again, and if they match, the CPU 73 performs speaker verification in step S28. .
[0051]
In steps S29 and S30, the same operation as described above is performed. That is, the CPU 73 performs voice feature extraction in the feature extraction circuit 102 of the specific speaker verification unit 70 in step S29, and in the storage circuit 103 already registered in the similarity calculation circuit 104 in step S30. Further, feature data closer to “OK” that is the input voice is searched for, and the similarity is calculated.
[0052]
Then, the CPU 73 determines the degree of similarity in step S31, and determines whether the sound is of “others” whose voice is not registered. Here, the similarity is leveled in advance as shown in Table 2, for example. Also, the similarity can be arbitrarily set. If the degree of similarity is increased, the exclusion rate as “another person” can be increased, and conversely, even if the person is the person himself, the possibility of exclusion is increased.
[0053]
In step S31, the CPU 73 determines whether or not the voice of “OK” is the main doctor. If the voices do not match, the CPU 73 returns to step S22 to input again. If the voices match, the process proceeds to step S16. Medical devices can be operated with voice.
[0054]
If the main doctor wants to stop the operation while the sub-doctor or nurse is operating the medical device, enter “NO” in step S23 to take the above steps and The operation can be stopped.
[0055]
As described above, in this embodiment, the unspecified speaker recognition engine 71 reliably identifies the voice of the specific speaker (main doctor) and the unspecified speaker (sub-doctor or nurse), and also specifies the unspecified speaker. The recognition engine 71 and the specific speaker verification unit 70 can prevent an operation not intended by a specific speaker (main doctor) from being accepted.
[0056]
FIG. 8 is a block diagram showing the configuration of the system controller according to the second embodiment of the present invention.
[0057]
Since the second embodiment is almost the same as the first embodiment, only different points will be described, and the same components are denoted by the same reference numerals and description thereof will be omitted.
[0058]
In this embodiment, as shown in FIG. 8, a specific speaker recognition engine 120 is provided in place of the specific speaker verification unit 70 of the first embodiment.
[0059]
In endoscopic surgery using many medical instruments, it is important for the surgeon to be able to operate the instrument from the sterilized area for smooth progress of surgery.
[0060]
Endoscopic surgery uses a mixture of low-risk devices that are not directly related to the patient, such as TV cameras and light sources, and high-risk devices that are directly related to the patient, such as an electric knife and an ultrasonic treatment device. Since a system using an unspecified speaker engine can be operated with a voice other than the surgeon, an operation unintended by the surgeon may be performed, and the surgeon must always monitor voice operations other than the surgeon.
[0061]
Therefore, the present embodiment includes both a specific speaker engine and an unspecified speaker engine, and can reduce the burden on the operator by using different recognition engines according to the risks of equipment peculiar to endoscopic surgery. It is what I did.
[0062]
Specifically, the specific speaker recognition engine 120 causes the main doctor to recognize a high-risk device, for example, a function operation word for operating the electric scalpel 32 and the insufflator 33, and the unspecified speaker recognition engine 71 The above-mentioned object can be achieved by recognizing functional operation words for operating the low-risk devices such as the first TV camera 23a, the second TV camera 23b, the first light source 24a, and the second light source 24b. .
[0063]
The processing flow is the same as that in the first embodiment (see FIGS. 4 to 7), and the operation of the device can be performed for both the main doctor and the sub-doctor.
[0064]
[Appendix]
(Additional Item 1) The medical device is a device with a high risk and a device with a low risk peculiar to endoscopic surgery.
The medical control system according to claim 1.
[0065]
(Additional Item 2) The high-risk device is a surgical device that uses energy such as an electric knife.
Item 2. The medical control system according to Item 1, wherein
[0066]
(Additional Item 3) The low-risk device is an observation device such as a TV camera or a light source device.
Item 2. The medical control system according to Item 1, wherein
[0067]
(Additional Item 4) The specific speaker is a main doctor in endoscopic surgery.
The medical control system according to claim 1.
[0068]
(Additional Item 5) The unspecified speaker is a sub-doctor or nurse of endoscopic surgery.
The medical control system according to claim 1.
[0069]
(Additional Item 6) Medical device control means for inputting a predetermined control signal and controlling the function of the predetermined medical device based on the cough control signal;
A specific speaker recognition means for recognizing a specific voice emitted by a specific speaker;
Unspecified speaker recognition means for recognizing a specific voice emitted by an unspecified speaker;
A foot switch for switching between the specific speaker and the non-specific speaker;
A medical control system comprising:
[0070]
(Additional Item 7) One or more medical devices;
A voice operation means capable of operating a plurality of functions of the medical device having a specific speaker recognition engine and an unspecified speaker recognition engine by voice;
Switching means for switching between the specific speaker recognition engine and the unspecified speaker recognition engine used by the voice operation means according to a plurality of functions of the medical device;
A medical control system comprising:
[0071]
(Additional Item 8) Medical device control means for inputting a predetermined control signal and controlling the function of the predetermined medical device based on the control signal;
First speaker recognition means for recognizing a specific voice emitted by a specific first speaker;
Second speaker recognition means for recognizing a specific voice emitted by a second speaker different from the first speaker;
When the voice of the second speaker is recognized by the second speaker recognition means, a control signal based on the voice of the first speaker recognized by the first speaker recognition means is Control signal sending means for sending to the medical device control means;
A medical control system comprising:
[0072]
【The invention's effect】
As described above, according to the present invention, the voice of a specific speaker and the voice of an unspecified speaker are reliably identified, and a plurality of devices can be easily operated by voice, and an operation unintended by a specific speaker can be performed. There is an effect that it can be surely prevented.
[Brief description of the drawings]
FIG. 1 is a configuration diagram showing a configuration of a surgical operation system according to a first embodiment of the present invention.
FIG. 2 is a block diagram showing the configuration of the system controller of FIG.
FIG. 3 is a block diagram showing the internal structure of the specific speaker verification unit and unspecified speaker recognition engine of FIG. 2;
FIG. 4 is a first flowchart for explaining the operation of the system controller of FIG.
FIG. 5 is a second flowchart for explaining the operation of the system controller of FIG. 2;
6 is a third flowchart for explaining the operation of the system controller of FIG.
7 is a fourth flowchart for explaining the operation of the system controller of FIG.
FIG. 8 is a block diagram showing a configuration of a system controller according to a second embodiment of the present invention.
[Explanation of symbols]
10 ... Surgery system
25 ... System controller
61 ... Microphone for main doctor
62 ... Sub-doctor microphone
63 ... Nurse microphone
70 ... Specific speaker verification circuit
71: Unspecified speaker recognition engine
72 ... Communication I / F
73 ... CPU
74 ... Foot switch
100: Voice input circuit
101 ... Voice input unit
102 ... Special trace extraction circuit
103: Memory circuit
104. Similarity calculation circuit
105. Peripheral device
106: Display device
107: Control circuit
108: Voice instruction circuit

Claims (4)

Medical device control means for inputting a predetermined control signal and controlling the function of the predetermined medical device based on the predetermined control signal;
The medical device operating under the permission of the voice input of the main doctor to approve the voice operation speaker emits possible, the first keywords for controlling the predetermined medical device, the registered first a speaker independent recognition means for determining whether a first keyword by determining the keyword similarity,
Wherein when it is determined as the first keyword by unspecified speaker recognition means, the said first keyword wherein said first keyword emitted as whether or not to permit control of the predetermined medical device by voice Specific speaker recognition means for determining whether or not the different second keyword is due to voice input of a main doctor who approves voice operation ;
A medical control system comprising: When the second keyword issued by the main doctor is recognized by the specific speaker recognition unit after being determined as the first keyword issued by the speaker capable of the operation by the unspecified speaker recognition unit And a control signal sending means for sending a control signal based on a first keyword issued by the speaker capable of the operation determined by the unspecified speaker recognition means to the medical device control means. The medical control system according to claim 1.   When the second keyword is for canceling the operation of the predetermined medical device by the second keyword, the control signal sending means operates the predetermined medical device by the first keyword. The medical control system according to claim 1, wherein the medical control system is stopped. 4. An input unit for rejecting recognition of a first keyword issued by a speaker capable of performing the operation by the unspecified speaker recognition unit. The medical control system described.

Images