The present invention relates to an endoscope system that detects a detection target, such as a bleeding point, and a method of operating the same.
In a medical field, an endoscope system including a light source device, an endoscope, and a processor device has been widely used. In endoscopic diagnosis, there is a case where a detection target such as a bleeding point is detected during endoscopic treatment. The detection of the detection target has been performed not only by detection by visual observation but also estimation by comparison with a past image. WO2019/202827A1, JP2015-529489A (corresponding to US2014/031659A1), and JP2011-036371A disclose that a bleeding point or region is detected from an image.
However, it is difficult to specify a position of the detection target by visual observation or from an image in a case where a factor that reduces visibility of the detection target, such as accumulated blood due to bleeding, occurs over time.
An object of the present invention is to provide an endoscope system and a method of operating the same with which a position of a detection target can be specified even in a case where the visibility of the detection target is reduced.
An endoscope system according to an aspect of the present invention comprises: a processor, in which the processor is configured to: acquire an endoscope image; acquire detection target actual position information of a detection target by performing a first detection process on the endoscope image; acquire position information of a landmark by performing a second detection process on the endoscope image in a case where the detection target actual position information is detected; perform a landmark setting process of associating the detection target actual position information with the position information of the landmark; after the landmark setting process, in a case where the detection target actual position information is not acquired and the landmark setting process has been performed, perform a position information estimation process, and calculate detection target estimated position information; and display the detection target estimated position information on a display.
It is preferable that the processor is configured to display the detection target actual position information, the detection target estimated position information, and the position information of the landmark on the display in different modes.
It is preferable that the processor is configured to: perform notification using either or both of a notification sound and notification on the display; and perform the notification in at least one case of a case where the detection target actual position information is detected during the first detection process or a case where the position information of the landmark is detected during the second detection process.
It is preferable that the processor is configured to: perform notification using either or both of a notification sound and notification on the display; and perform the notification in at least one case of a case where the landmark setting process is completed, a case where the detection target is not detected, a case where the position information estimation process is started, or a case where the detection target estimated position information is calculated during the position information estimation process, as a result of the landmark setting process.
It is preferable that the processor is configured to: perform notification using either or both of a notification sound and notification on the display; and perform the notification in a case where a required number of pieces of the position information of the landmarks are not capable of being acquired during the second detection process and the landmark setting process is not capable of being executed, and in a case where the detection target actual position information disappears before the landmark setting process is completed and the landmark setting process fails.
It is preferable that the processor is configured to display, on the display, acquisition number information of position information by character display including at least a quantity of the detection target estimated position information.
It is preferable that the processor is configured to display, on the display, acquisition number information of the detection target actual position information and the position information of the landmark.
It is preferable that the processor is configured to limit the number of landmarks used in the position information estimation process.
It is preferable that the processor is configured to select the landmark to be displayed on the display among the landmarks, and limit the landmarks to be displayed on the display.
It is preferable that the processor is configured to receive a user operation for designating whether or not the landmark is usable in the position information estimation process.
It is preferable that the endoscope image includes a first endoscope image based on first illumination light and a second endoscope image based on second illumination light having a spectrum different from a spectrum of the first illumination light, and the processor is configured to perform the first detection process and the second detection process from the second endoscope image, and display the detection target actual position information, the detection target estimated position information, and the position information of the landmark on the display from the first endoscope image.
It is preferable that a start timing of the first detection process is any of a time at which water supply is detected, a time at which incision is detected, a time at which a treatment tool is detected, or a time at which a user operation is performed, an end timing of the first detection process is any of a time at which a predetermined time has elapsed with non-detection of the detection target, a time at which a predetermined time has elapsed since disappearance of the detection target, a time at which a bleeding region is in a non-enlargement state, or a time at which a user operation is performed, a start timing of the second detection process is a time at which the detection target is detected, an end timing of the second detection process is any of a time at which the detection target disappears, a time at which a required number of the landmarks are not detected within a predetermined time, or a time at which a user operation is performed, a start timing of the position information estimation process is a time at which the landmark setting process is completed and the detection target disappears, and an end timing of the position information estimation process is any of a time at which the landmark disappears, a time at which a hemostasis point is detected, a time at which bleeding around an estimated position of the detection target is not detected, or a time at which a user operation is performed.
It is preferable that a restart timing of the first detection process is any of a time at which the landmark setting process fails, a time at which the position information estimation process fails, or a time at which a user operation is performed.
It is preferable that the landmark is at least any of a mucous membrane pattern, a shape of an organ, or marking by a user operation.
An endoscope system according to the aspect of the present invention comprises: a processor, in which the processor is configured to: acquire an endoscope image; acquire detection target actual position information of a detection target by performing a first detection process on the endoscope image; acquire position information of a landmark by performing a second detection process on the endoscope image; perform a landmark setting process of setting a relative relationship by associating any of the detection target actual position information or detection target estimated position information obtained from a position information estimation process based on the position information of the landmark, with the position information of the landmark each time the endoscope image is updated and the detection target actual position information or the detection target estimated position information is acquired; and display the detection target actual position information or the detection target estimated position information on a display.
It is preferable that the processor is configured to: in a case where a new landmark is detected by acquiring the endoscope image of a new frame in a state where the position information estimation process is continued, perform a new landmark setting process of setting a new relative relationship by associating the detection target estimated position information with the new landmark as the landmark setting process; after the new landmark setting process, in a case where the landmark necessary for the position information estimation process is not recognized, perform a position information estimation process based on the new relative relationship, and calculate a new detection target estimated position information; and display the new detection target estimated position information on the display.
It is preferable that the new landmark is at least any of a mucous membrane pattern, a shape of an organ, or marking by a user operation.
A method of operating an endoscope system according to the aspect of the present invention comprises: via a processor, a step of acquiring an endoscope image; a step of acquiring detection target actual position information of a detection target by performing a first detection process on the endoscope image; a step of acquiring position information of a landmark by performing a second detection process on the endoscope image in a case where the detection target actual position information is detected; a step of performing a landmark setting process of associating the detection target actual position information with the position information of the landmark; a step of performing a position information estimation process and calculating detection target estimated position information in a case where the detection target actual position information is not acquired and the landmark setting process has been performed after the landmark setting process; and a step of displaying the detection target estimated position information on a display.
It is preferable that the method of operating an endoscope system further comprises: via the processor, a step of displaying the detection target actual position information, the detection target estimated position information, and the position information of the landmark on the display in different modes.
According to the present invention, it is possible to specify the position of the detection target even in a case where the visibility of the detection target is reduced.
As shown in
The endoscope 12 is used for illuminating an observation target with illumination light and imaging the observation target to acquire an endoscope image. The endoscope 12 includes an insertion part 12a to be inserted into a body of the observation target, an operating part 12b provided at a base end portion of the insertion part 12a, and a bendable part 12c and a distal end part 12d provided on a distal end side of the insertion part 12a. The bendable part 12c performs a bending operation by operating the operating part 12b. The distal end part 12d irradiates the observation target with illumination light and receives reflected light from the observation target to image the observation target. The distal end part 12d is directed in a desired direction by the bending operation of the bendable part 12c. The operating part 12b includes a mode selector switch 12f used for a mode switching operation, a still image acquisition instruction switch 12g used for providing an instruction of acquisition of a still image of the observation target, a zoom operation part 12h used for an operation of a zoom lens 21b, and a water supply switch 12i used for a water supply operation.
The processor device 14 is electrically connected to the display 15 and the user interface 16. The processor device 14 receives the endoscope image from the endoscope 12. The display 15 outputs and displays an image, information, or the like of the observation target processed by the processor device 14. The user interface 16 includes a keyboard, a mouse, a touch pad, a microphone, and the like, and has a function of receiving an input operation such as function setting. The extended processor device 17 is electrically connected to the processor device 14. The extended processor device 17 receives the image or various kinds of information from the processor device 14. The display 18 outputs and displays an image, information, or the like processed by the extended processor device 17.
The endoscope system 10 comprises a mono-light emission mode, a multi-light emission mode, and a tracking mode, and the modes are switched by the mode selector switch 12f The mono-light emission mode is a mode in which the observation target is continuously illuminated with illumination light having the same spectrum. The multi-light emission mode is a mode in which the observation target is illuminated while switching a plurality of illumination light beams having different spectra according to a specific pattern. The illumination light includes normal light (broadband light such as white light) used for giving brightness to the entire observation target to observe the entire observation target, or special light used for emphasizing a specific region of the observation target. Further, in the mono-light emission mode, the illumination light may be switched to illumination light having a different spectrum by the operation of the mode selector switch 12f For example, first illumination light and second illumination light having different spectra may be switched. The tracking mode is not exclusive of the mono-light emission mode and the multi-light emission mode, and the tracking can be performed even in the mono-light emission mode and the multi-light emission mode.
The tracking mode is a mode in which actual position information of a detection target is detected, and position information of the detection target and position information of a landmark associated with the position information of the detection target are displayed on the display 18 (or the display 15) in order to allow a user to grasp a position of the detection target such as a bleeding point BS even though a change occurs in an image of a subject or the like.
As shown in
The light source device 13 independently controls the light amounts of the four colors of violet light V, blue light B, green light G, and red light R. As shown in
The processor device 14 may implement a function of the extended processor device 17 and replace the extended processor device 17. The processor device 14 performs various kinds of processing executed by functions implemented by the extended processor device 17 described below, in addition to a function of receiving an endoscope image from the endoscope 12. In this case, an image or information subjected to various kinds of processing may be displayed on the display 15 or may be displayed on the display 18.
As shown in
As shown in
The image acquisition unit 20 acquires the endoscope image transmitted from the processor device 14. The processor device 14 transmits the endoscope image to the extended processor device 17 for each frame. The image acquisition unit 20 acquires the endoscope image for each frame transmitted from the processor device 14.
The detection target detection unit 21 detects a detection target and acquires actual position information of the detection target by performing a first detection process on the endoscope image. As shown in
It is preferable that the detection target detection unit 21 is a trained model that has been trained through machine learning using teacher image data including the detection target. The machine learning includes supervised learning, semi-unsupervised learning, unsupervised learning, reinforcement learning, deep reinforcement learning, learning using a neural network, deep learning, and the like. In a case where the detection target is detected by the first detection process, information on the detected detection target is stored in the detection memory 26.
In a case where the detection target is detected by the detection target detection unit 21, the landmark detection unit 22 detects the landmark LM and acquires position information of the landmark LM by performing a second detection process on the endoscope image. Examples of the landmark LM include various structures such as blood vessels and glandular structures, a mucous membrane pattern, a shape of an organ, marking by a user operation, marking after cauterization, and marking given to a body (marking given by a coloring agent or marking with lame or marker given on a bulging material to be injected in a case of incision). As shown in
The landmark LM detected in the second detection process is subjected to a landmark setting process of calculating estimated position information of the bleeding point BS using position information of the plurality of landmarks LM by the landmark processing unit 23. However, a use of the landmark LM with low reliability may cause the estimated position of the bleeding point BS to be far away from the actual position. In order to prevent this, the number of the landmarks LM used in the landmark setting process may be limited. For example, as shown in
Regarding the limitation of the landmarks LM to be detected, the landmark LM with a high reliability degree may be automatically detected by setting the upper limit number or the threshold value, but the landmark LM with low reliability may be designated by a user operation and excluded from the target of the landmark setting process. For example, the landmark LM having the distinction number “4” indicated as having low reliability in a display mode of the landmark position display indicator 32 as shown in
In a case where the required number or more of the landmarks LM are detected by the second detection process, the landmark processing unit 23 performs the landmark setting process of associating the position information of the landmark LM with the actual position information of the detection target. As shown in
The display control unit 24 performs any of a first display control process of displaying the actual position information of the detection target and the position information of the landmark LM on the display 18 in a case where the detection target is detected, a second display control process of displaying the position information of the landmark LM on the display 18 in a case where the detection target is not detected and the landmark LM is detected, or a third display control process of displaying the estimated position information of the detection target in a case where the detection target is not detected and the landmark setting process is performed. As shown in
Then, as shown in
In a case where the detection target is not detected and the landmark setting process has been performed, the estimated position information calculation unit 25 calculates the estimated position information of the detection target by the position information estimation process based on the position information of the landmark LM. As shown in
The detected position display indicator 30, the landmark position display indicator 32, and the estimated position display indicator 36 are displayed on the display 18 in different display modes in order to be easily distinguished from each other. As shown in
A processing unit for performing the second detection process in the landmark processing unit 23 is preferably a trained model for landmark detection that has been trained through machine learning using teacher image data including the landmarks LM. In a case where the landmark processing unit 23 can calculate the reliability degree related to the detection of the landmark LM, it is preferable to change a display mode (color, line style, or the like) of the position information of the landmark LM according to the reliability degree. For example, as shown in
In the estimated position display control process, it is preferable to change an estimated position display mode, which is the estimated position display indicator 36 in the present embodiment, according to the reliability degree of the estimated position information. In this case, the estimated position information calculation unit 25 calculates the reliability degree of the estimated position information in accordance with the calculation of the estimated position information. For example, it is preferable to calculate a confidence degree for the estimated position information as the reliability degree for the estimated position information from a model that has been trained through machine learning. The user can select an operation on the observation target according to the reliability degree of the estimated position information. For example, in a case where the reliability degree is high, a hemostasis process for the bleeding point BS is performed, whereas in a case where the reliability degree is low, the hemostasis process is not performed in order to avoid hemostasis at a wrong portion.
Specifically, as shown in
The display control unit 24 may perform notification of a result of each process using a notification sound and notification on the display 18. As shown in
In a case where the landmark setting process is completed, as shown in
In a case where the landmark setting process is not completed, there are, for example, a pattern (
Further, the acquisition number of the position information in the tracking mode may be displayed on the display 18. For example, as shown in
It is preferable that the processing timing setting unit 27 sets a start timing or an end timing of the first detection process, a start timing or an end timing of the second detection process, and a start timing or an end timing of the position information estimation process (see
Specifically, as shown in
It is preferable that the start timing of the second detection process is a timing at which the detection target is detected by the first detection process (at the time of detection target detection). In addition, it is preferable that the end timing of the second detection process is a timing after a predetermined time has elapsed from the detection of the latest landmark LM (after an elapse of a predetermined time after the detection of the landmark) or a timing at which the number of the detected landmarks LM reaches a set upper limit number (after the number of the detected landmarks reaches the upper limit number).
The start timing of the position information estimation process is a timing at which the landmark is set and the detection target disappears. It is preferable that the end timing of the position information estimation process is a timing at which the estimated position information cannot be calculated by the position information estimation process due to disappearance of the landmark (at the time of failure of the position information estimation process), a timing at which a hemostasis point is detected as a result of performing a hemostatic treatment on the bleeding point with a treatment tool or the like (at the time of hemostasis point detection), or a timing at which bleeding cannot be confirmed in the estimated position display indicator 36 and its periphery (at the time of bleeding non-detection at the estimated position). In a case where the tracking mode is continued even though the position information estimation process fails, it is preferable to redo the first detection process.
It is preferable that the re-detection start timing of the first detection process is a timing at which the detection target disappears before the landmark setting process is completed (at the time of failure of the landmark setting process) or a timing at which the landmark disappears before the position information estimation process is completed (at the time of failure of the position information estimation process).
In the tracking mode, as shown in
Next, a series of flows in the tracking mode will be described with reference to a flowchart of
Whether or not the detection target is a new detection target is determined depending on whether or not information relating to the detection target is present in the detection memory 26. In a case where a new detection target is detected, the information relating to the detection target that has already been stored in the detection memory 26 is deleted, and information relating to the new detection target is newly stored in the detection memory 26. In this case, it is preferable that information relating to the landmark LM associated with the detection target to be deleted is also deleted.
On the other hand, in a case where the detection target cannot be detected by the first detection process, it is determined whether or not the landmark setting process has already been performed (determination of whether or not the landmark setting process is being performed). In a case where the landmark setting process has already been performed, the estimated position information of the detection target is calculated based on the position information of the landmark LM. Then, the calculated estimated position information of the detection target and the position information of the landmark LM are displayed on the display 18. The series of processes described above is repeatedly performed as long as bleeding is detected or estimated. Then, in a case where the mode selector switch 12f is operated and the tracking mode is turned OFF, the detection of the detection target and the like are ended.
Since the endoscope 12 is manually operated, even though the estimated position of the detection target is continuously captured in the endoscope image, a range of the endoscope image may change, and in a case where the landmark LM surrounding the estimated position of the detection target does not fall within the endoscope image, the organ may be deformed and the relationship between the landmark and the detection target may be changed. As shown in
In a case where the landmark LM used in the position information estimation process is updated, the landmark LM before update is set as the landmark LM and the landmark LM after update is set as the new landmark LM2. (A) of
After the new landmark setting process, in a case where the endoscope 12 acquires an endoscope image of a new frame and the landmark LM necessary for the position information estimation process is not recognized, as shown in (C) of
Since the range imaged by the endoscope 12 moves even from the state of the position information estimation process using the new landmark LM2, the update of the landmark LM used for the position information estimation process continues. With the new landmark LM2 as the landmark LM and the new link line 39 as the link line 34, the new landmark LM2 updates the relative relationship by the landmark setting process, and the landmark LM performs the position information estimation process.
In the above-described embodiment, hardware structures of processing units executing various processes, such as the image acquisition unit 20, the detection target detection unit 21, the landmark processing unit 23, the display control unit 24, the estimated position information calculation unit 25, the detection memory 26, or the processing timing setting unit are various processors as follows. The various processors include a central processing unit (CPU) that is a general-purpose processor that executes software (programs) to function as various processing units, a graphical processing unit (GPU), a programmable logic device (PLD) that is a processor capable of changing a circuit configuration after manufacture, such as a field programmable gate array (FPGA), and an exclusive electric circuit that is a processor having a circuit configuration exclusively designed to execute various kinds of processing.
One processing unit may be configured of one of these various processors, or may be configured of a combination of two or more processors of the same type or different types (for example, a plurality of FPGAs, a combination of a CPU and an FPGA, or a combination of a CPU and a GPU). In addition, a plurality of processing units may be constituted by one processor. As an example in which the plurality of processing units are configured of one processor, first, as typified by computers such as a client or a server, one processor is configured of a combination of one or more CPUs and software, and this processor functions as the plurality of processing units. Second, as typified by a system on chip (SoC) or the like, a processor that implements the functions of the entire system including the plurality of processing units by using one integrated circuit (IC) chip is used. As described above, the various processing units are configured using one or more of the various processors as a hardware structure.
Further, the hardware structure of these various processors is more specifically an electric circuit (circuitry) in a form in which circuit elements such as semiconductor elements are combined. The hardware structure of the storage unit is a storage device such as a hard disc drive (HDD) or a solid state drive (SSD).
| Number | Date | Country | Kind |
|---|---|---|---|
| 2021-037558 | Mar 2021 | JP | national |
This application is a Continuation of PCT International Application No. PCT/JP2022/004837 filed on 8 Feb. 2022, which claims priority under 35 U.S.C § 119(a) to Japanese Patent Application No. 2021-037558 filed on 9 Mar. 2021. The above application is hereby expressly incorporated by reference, in its entirety, into the present application.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/JP2022/004837 | Feb 2022 | US |
| Child | 18463930 | US |