The present invention relates to a surgery assistance apparatus and a surgery assistance method for assisting endoscopic surgery, and further relates to a computer readable recording medium that includes a program for realizing the surgery assistance apparatus and the surgery assistance method recorded thereon.
A surgery assistance apparatus that extracts a target part image corresponding to a target part from an image obtained by capturing an image of the inside of a human body using an endoscope, and that uses the extracted target part image to assist an observer during surgery is known. Such an apparatus improves the accuracy of surgery by capturing an image of the inside of a colon, etc., using an endoscope, detecting a tumor that is likely to become cancerous from the captured image, and providing a notification to the observer, for example.
As a related technique, Patent Document 1 discloses an apparatus that extracts an affected-part image (target part image) corresponding to an affected part (target part) from an image obtained by capturing an image of the inside of a human body using an endoscope, and identifies the pathological type of the affected part on the basis of a result of feature amount matching processing between the extracted affected-part image and learning images.
Patent Document 1: International Publication No. 2015/118850
However, the target part image cannot be tracked even if the above-described surgery assistance apparatus or the apparatus disclosed in Patent Document 1 is used. Furthermore, in the case where surgery is performed using an endoscope, it is difficult to track the target part image even if a conventional tracking technique (such as optical flow, for example) is used.
In the case where surgery is performed using an endoscope, movement of the target part image out of the frame, changes in the target part image brought about by the image-capturing distance, angle, etc., between the image-capturing unit of the endoscope and the target part changing as the endoscope is inserted and extracted, etc., make it difficult to track the target part image, for example.
It is also difficult to track the movement of the target part image because the target part is frequently concealed by parts of the human body other than the target part (for example, concealment of the target by intestinal folds, intestinal mucus, etc.) or by water ejected from the endoscope, etc.
Furthermore, it is difficult to track the target part image on the basis of shape or color because the internal parts of the human body and the target part are non-rigid objects (objects without definite shapes).
An example object of the present invention is to provide a surgery assistance apparatus, a surgery assistance method, and a computer readable recording medium that includes a surgery assistance program recorded thereon that improve the accuracy of endoscopic surgery by identifying a target part image.
In order to achieve the above-described object, a surgery assistance apparatus according to an example aspect of the present invention includes:
a feature amount calculation unit configured to calculate, from a human-body internal image captured using an endoscope, a feature amount of a target part image corresponding to a target part;
a similarity degree calculation unit configured to calculate a similarity degree of the feature amount between different ones of the human-body internal images; and
an identification unit configured to identify the target part image in each of the different human-body internal images if the similarity degree is greater than or equal to a predetermined value.
In addition, in order to achieve the above-described object, a surgery assistance method according to an example aspect of the present invention includes:
Furthermore, in order to achieve the above-described object, a computer readable recording medium that includes a surgery assistance program recorded thereon according to an example aspect of the present invention causes the following steps to be carried out:
As described above, according to the present invention, the accuracy of endoscopic surgery can be improved by identifying a target part image.
In the following, an example embodiment of the present invention will be described with reference to
First, a configuration of a surgery assistance apparatus in the present example embodiment will be described with reference to
A surgery assistance apparatus 1 in the present example embodiment, which is illustrated in
Of these units, the feature amount calculation unit 2 calculates, from a human-body internal image captured using an endoscope, a feature amount of a target part image corresponding to a target part. The similarity degree calculation unit 3 calculates a similarity degree of the feature amount between different ones of the human-body internal images. The identification unit 4 identifies the target part image in each of the different human-body internal images if the similarity degree is greater than or equal to a predetermined value.
In such a manner, in the present example embodiment, a target part image in different human-body internal images captured using an endoscope is identified if a similarity degree of a feature amount between the different human-body internal images is greater than or equal to a predetermined value. Thus, the accuracy of endoscopic surgery can be improved because a target part image can be tracked.
Specifically, an observer conventionally visually tracks a target part image, and thus there are cases where the observer cannot refind a target part image if the observer loses track of the target part image. In such cases, there is a risk of a target part which is likely to become cancerous and need surgery, etc., of being overlooked. However, because a target part image can be identified using the surgery assistance apparatus in the present example embodiment, cases where a target part is overlooked can be reduced, regardless of the observer's skill and the like, and thus the accuracy of surgery can be improved.
Cases where the observer loses track of a target part image that is captured using an endoscope can be reduced because a target part image can be identified even if the target part image moves out of the frame or the target part image undergoes changes (a change in the size of the target part, a change in the image-capturing range, rotation of the image, etc.) as the endoscope is inserted and extracted, for example. Accordingly, the accuracy of endoscopic surgery can be improved.
Also, cases where the observer loses track of a target part image can be reduced because a target part image can be identified even if the target part is frequently concealed by parts of the human body other than the target part (for example, concealment of the target by intestinal folds, intestinal mucus, etc.) or by water ejected from the endoscope, etc. Accordingly, the accuracy of endoscopic surgery can be improved.
Furthermore, cases where the observer loses track of a target part image can be reduced because a target part image can be identified even in the case of non-rigid objects (objects without definite shapes) such as internal parts of the human body and the target part. Accordingly, the accuracy of endoscopic surgery can be improved.
Next, the configuration of the surgery assistance apparatus 1 in the present example embodiment will be specifically described with reference to
As illustrated in
The endoscope 20 transmits, to the surgery assistance apparatus 1 connected to the endoscope 20, a human-body internal image in which the inside of a human body is captured. For example, the endoscope 20 includes an insertion unit that is inserted into the human body, an image-capturing unit that is provided on the distal end-side of the insertion unit, an operation unit for controlling bending of the insertion unit, the capturing of images by the image-capturing unit, etc., and a connection unit that connects the endoscope 20 and the surgery assistance apparatus 1. In addition to the image-capturing unit, the endoscope 20 also includes an illumination unit, a nozzle (nozzles) used for feeding air and water and for suction, a forceps port, and the like on the distal end-side of the insertion unit.
The output device 21 acquires, from the output information generation unit 22, output information converted into formats that can be output, and outputs images, sound, etc., generated on the basis of the output information. The output device 21, for example, includes an image display device utilizing liquid crystals, organic electroluminescence (EL), or a cathode ray tube (CRT), and further includes a sound output device such as a speaker, and the like. Note that the output device 21 may also be a printing device such as a printer.
The feature amount calculation unit 2 acquires a plurality of human-body internal images captured in time series by the endoscope 20, and extracts a feature amount of a target part image corresponding to a target part from the human-body internal images. Furthermore, the feature amount calculation unit 2 includes a feature extraction unit 23, a detection result calculation unit 24, and a heatmap calculation unit 25 that extract feature amounts.
The feature extraction unit 23 extracts, from a human-body internal image, feature extraction information (feature amount f; feature vector) indicating features of a target part image. Specifically, local binary patterns (LBP) is one method for extracting local features of an image, and the extraction method is disclosed for example in the following document: “T. Ojala, M. Pietikainen, and D. Harwood, ‘Performance evaluation of texture measures with classification based on Lullback discrimination of distributions,’ in the Proceedings of IEEE International Conference on Pattern Recognition, 1994.”
In
Next, for each of the acquired human-body internal images 31, the feature extraction unit 23 performs extraction of feature extraction information of a target part image 33 corresponding to the target part. In
The detection result calculation unit 24 calculates detection result information (feature amounts r; feature vectors) using the feature extraction information (feature amount f; feature vector) extracted from each of the human-body internal images 31. Specifically, the detection result calculation unit 24 applies processing, such as sliding window, for detecting a target part image 33 to each of the images 32 corresponding to the feature extraction information, and calculates the detection result information for each of the human-body internal images 31.
The detection result information includes, for example, region information indicating the position and size of a detection target part and confidence information indicating the probability of the region of the detection target part corresponding to the target part. The region information and the confidence information are calculated using features inside the windows 41 and 41′ (in the following, the window 41′ is also referred to as the window 41). The region information, for example, includes position information indicating the position of a rectangle circumscribing the target part, and size information indicating the size of the circumscribing rectangle.
If the images 32 are regarded as two-dimensional coordinate systems, position information indicating a position of the window 41 can be indicated by coordinates inside the window 41. Center coordinates (X, Y) as illustrated in
Similarly to the region information of the window 41, the region information of the detection result information can be indicated by the center coordinates (Xr, Yr) of the rectangle 42 circumscribing the target part, and size information indicating the size of the rectangle 42 (the width (Wr) and height (Hr) of the rectangle), as illustrated in
Note that the detection result information may for example, be expressed in a form such as: feature vector r=(Xr′, Yr′, Wr, Hr, conf). Furthermore, the region information of the detection result information need not have a rectangular shape. The shape may be circular, elliptical, or the like, for example.
The heatmap calculation unit 25 calculates heatmap information (feature amount h; feature vector) using the feature extraction information (feature amount f; feature vector) extracted from each of the human-body internal images 31. Specifically, the heatmap calculation unit 25 calculates the heatmap information by applying a semantic segmentation technique, for example, to the images 32 corresponding to the feature extraction information.
Furthermore, the feature extraction unit 23 may calculate feature extraction information (feature amount f′: feature vector) and detection result information (r) for each window 41. In addition, the heatmap calculation unit 25 may calculate heatmap information (feature amount h′; feature vector) for each window 41.
Note that the feature amount calculation unit 2 stores the human-body internal images 31 and the above-described feature amounts f, r, and h in an associated state.
The similarity degree calculation unit 3 calculates a similarity degree using the feature extraction information (feature amount f), the detection result information (feature information r), and the heatmap information (feature amount h) in different human-body internal images 31. Specifically, in the calculation of a similarity degree between human-body internal images 31, a distance between feature vectors (similarity degree) is calculated using feature vectors of the feature extraction information (feature amount f), feature vectors of the detection result information (feature information r), and/or feature vectors of the heatmap information (feature amount h). Alternatively, the similarity degree may be expressed using linear combinations. The similarity degree calculation unit 3 calculates the similarity degree according to the methods described in (1) to (5).
(1) The similarity degree calculation unit 3 calculates a similarity degree of the detection result information (r), a similarity degree of the feature extraction information (f), or a similarity degree of the heatmap information (h).
(2) The similarity degree calculation unit 3 calculates a similarity degree of the position information (X, Y) included in the region information of the detection result information in each window, a similarity degree of the size information (W, H) included in the region information, a similarity degree of the confidence information (conf), a similarity degree of the feature extraction information (f′), or a similarity degree of the heatmap information (h′).
(3) The similarity degree calculation unit 3 combines two or more out of the degrees of similarity indicated in (1) and (2), and sets the total sum of the degrees of similarity as a similarity degree. That is, the total sum of two or more out of the similarity degree of the detection result information (r), the similarity degree of the feature extraction information (f), the similarity degree of the heatmap information (h), the similarity degree of the position information (X, Y) included in the region information, the similarity degree of the size information (W, H) included in the region information, the similarity degree of the confidence information (conf), the similarity degree of the feature extraction information (f′), and the similarity degree of the heatmap information (h′) is set as a similarity degree.
(4) The similarity degree calculation unit 3 calculates the similarity degree of the confidence information (conf), and if the calculated similarity degree is greater than or equal to a predetermined confidence value that is set in advance, calculates a similarity degree of feature amounts between the different human-body internal images 31 (one or more of the degrees of similarity indicated in (1) and (2) described above). The predetermined confidence value is a determination value calculated through experimentation, simulation, machine learning, etc., and is stored in a storage unit provided in the surgery assistance apparatus or outside the surgery assistance apparatus. Alternatively, the maximum value of confidence information calculated from a plurality of human-body internal images 31 acquired in the past may be set as the predetermined confidence value. By adopting such a configuration, the calculation of a similarity degree becomes unnecessary in a case where the same target part image 33 is continuously captured in human-body internal images 31.
(5) The similarity degree calculation unit 3 calculates the similarity degree of the confidence information (conf), and if the calculated similarity degree is greater than or equal to the predetermined confidence value, calculates the similarity degree of the region information (X, Y, W, H). Next, the similarity degree calculation unit 3 calculates the similarity degree of the feature extraction information (f) if the similarity degree of the region information is greater than or equal to a predetermined region value. Then, the similarity degree calculation unit 3 calculates the similarity degree of the heatmap information (h) if the similarity degree of the feature extraction information is greater than or equal to a predetermined feature extraction value.
Note that each of the predetermined region value and the predetermined feature extraction value is a determination value calculated through experimentation, simulation, machine learning, etc., and is stored in the storage unit provided in the surgery assistance apparatus or outside the surgery assistance apparatus. By adopting such a configuration, the calculation of a similarity degree becomes unnecessary in a case where the same target part image 33 is continuously captured in human-body internal images 31, and the accuracy of degrees of similarity between human-body internal images 31 can be improved.
Furthermore, in the calculation of the similarity degree of the region information in (5), the similarity degree may be calculated using either the position information (X, Y) or the size information (W, H) in the region information.
The identification unit 4 identifies a target part image in each of the different human-body internal images 31 if the similarity degree is greater than or equal to a predetermined value. Specifically, if the similarity degree calculated according to one of (1) to (5) is greater than or equal to the predetermined value, the identification unit 4 associates the target part images 33 in the human-body internal images 31 with one another and stores the target part images 33 in the storage unit. The identification unit 4 performs the identification according to the methods indicated in (1′) to (5′).
(1′) The identification unit 4 identifies the target part images 33 if the similarity degree of the detection result information (r) is greater than or equal to a predetermined detection result value, the similarity degree of the heatmap information (h) is greater than or equal to a predetermined heatmap value, or the similarity degree of the feature extraction information (f) is greater than or equal to the predetermined feature extraction value. The predetermined heatmap value is a determination value calculated through experimentation, simulation, machine learning, etc., and is stored in the storage unit provided in the surgery assistance apparatus 1 or outside the surgery assistance apparatus 1.
(2′) The identification unit 4 identifies the target part images 33 if the similarity degree of the position information (X, Y) included in the region information, the similarity degree of the size information (W, H) included in the region information, the similarity degree of the confidence information (conf), the similarity degree of the feature extraction information (f′), and the similarity degree of the heatmap information (h′) are all greater than or equal to a predetermined value set for each similarity degree. Alternatively, the identification unit 4 may identify the target part images 33 if any of the degrees of similarity is greater than or equal to the predetermined value.
(3′) The identification unit 4 sets the total sum of two or more out of the degrees of similarity indicated in (1) and (2) as the similarity degree, and identifies the target part images 33 if the similarity degree is greater than or equal to a predetermined value that is set for each combination.
(4′) The identification unit 4 carries out the identification indicated in (1′), (2′), or (3′) if the similarity degree of the confidence information (conf) is greater than or equal to the predetermined confidence value.
(5′) The identification unit 4 identifies that the target part images 33 in the different human-body internal images 31 are the same if the similarity degree of the heatmap information (h), which is indicated in (5), is greater than or equal to the predetermined heatmap value.
By adopting such configurations, the target part image 33c captured in the latest human-body internal image 31c and the target part image 33a in the human-body internal image 31a captured in the past can be associated with one another even if the human-body internal image 31b, in which a target part image is not captured, is present between the human-body internal image 31c and the human-body internal image 31a, that is, even if the observer loses track of the target part image 33, for example.
The output information generation unit 22 generates output information indicating that target part images 33 have been identified if target part images 33 are identified during surgery, and transmits the generated output information to the output device 21. The output device 21 acquires the output information, and then outputs, on the basis of the output information, at least one of a screen and sound indicating that target part images 33 have been identified.
Next, the operations of the surgery assistance apparatus in the example embodiment of the present invention will be described with reference to
In step A1, the feature amount calculation unit 2 acquires human-body internal images 31 that have been captured in time series by the endoscope 20. Next, in step A2, the feature amount calculation unit 2 calculates feature amounts of target part images 33 corresponding to a target part from the human-body internal images 31 captured using the endoscope 20. See
Specifically, the feature amount calculation unit 2 (feature extraction unit 23) extracts, from the human-body internal images 31, feature extraction information (feature amounts f; feature vectors) indicating features of the target part images 33. Next, the feature amount calculation unit 2 (detection result calculation unit 24) calculates detection result information (feature amounts r; feature vectors) using the feature extraction information (feature amount f; feature vector) extracted from each of the human-body internal images 31. For example, processing, such as sliding window, for detecting a target part image 33 is applied to images 32 corresponding to the feature extraction information, and the detection result information is calculated for each of the human-body internal images 31. See
Alternatively, the feature amount calculation unit 2 (heatmap calculation unit 25) calculates heatmap information (feature amount h; feature vector) using the feature extraction information (feature amount f; feature vector) extracted from each of the human-body internal images 31. The heatmap information is calculated by applying semantic segmentation to the images 32 corresponding to the feature extraction information, for example. See
Note that the feature amount calculation unit 2 stores the human-body internal images 31 and the above-described feature amounts f, r, and h in an associated state. See
In step A3, the similarity degree calculation unit 3 calculates a similarity degree using the feature extraction information (feature amount f), the detection result information (feature information r), and/or the heatmap information (feature amount h) in the latest human-body internal image 31 and a human-body internal image 31 captured before the latest human-body internal image 31. Specifically, in the calculation of the similarity degree between the human-body internal images 31, a distance between feature vectors (similarity degree) is calculated using feature vectors of the feature extraction information (feature amount f), feature vectors of the detection result information (feature information r), and/or feature vectors of the heatmap information (feature amount h). Alternatively, the similarity degree may be expressed using linear combinations. The similarity degree calculation unit 3 calculates the similarity degree according to the methods described in (1) to (5).
For example, the similarity degree calculation unit 3 calculates a similarity degree between the latest human-body internal image 31c and the human-body internal image 31a or 31b captured before the latest human-body internal image 31c.
In step A4, the identification unit 4 identifies the target part image 33 in the latest human-body internal image 31 and the target part image 33 in the human-body internal image 31 captured before the latest human-body internal image 31 if the similarity degree is greater than or equal to a predetermined value. Specifically, if the similarity degree calculated according to one of (1) to (5) is greater than or equal to the predetermined value, the identification unit 4 associates the target part images 33 in the human-body internal images 31 with one another and stores the target part images 33 in the storage unit. The identification unit 4 performs the identification according to the methods indicated in (1′) to (5′).
In step A5, the output information generation unit 22 generates output information indicating that target part images 33 have been identified if target part images 33 are identified during surgery, and transmits the generated output information to the output device 21.
In step A6, the output device 21 acquires the output information, and then outputs, on the basis of the output information, at least one of a screen and sound indicating that target part images 33 have been identified. See
In step A7, the surgery assistance apparatus 1 terminates the identification processing illustrated in steps A1 to A7 if an instruction to terminate the identification processing is acquired (Yes). The surgery assistance apparatus 1 moves on to the processing in step A1 if the identification processing illustrated in steps A1 to A7 is to continue (No).
As described above, according to the present example embodiment, target part images 33 in different human-body internal images 31 captured using an endoscope 20 are identified if a similarity degree of feature amounts of the different human-body internal images 31 is greater than or equal to a predetermined value. Thus, because target part images 33 can be identified, cases where a target part is overlooked can be reduced, regardless of the observer's skill and the like, and thus the accuracy of surgery can be improved.
Cases where the observer loses track of a target part image that is captured using an endoscope can be reduced because a target part image can be identified even if the target part image moves out of the frame or changes as the endoscope is inserted and extracted, for example. Accordingly, the accuracy of endoscopic surgery can be improved.
Also, cases where the observer loses track of a target part image can be reduced because a target part image can be identified even if the target part is frequently concealed by parts of the human body other than the target part (for example, concealment of the target by intestinal folds, intestinal mucus, etc.) or by water ejected from the endoscope, etc. Accordingly, the accuracy of endoscopic surgery can be improved.
Furthermore, cases where the observer loses track of a target part image can be reduced because a target part image can be identified even in the case of non-rigid objects (objects without definite shapes) such as internal parts of the human body and the target part. Accordingly, the accuracy of endoscopic surgery can be improved.
It suffices for the program in the example embodiment of the present invention to be a program that causes a computer to carry out steps A1 to A7 illustrated in
Also, the program in the present example embodiment may be executed by a computer system formed from a plurality of computers. In this case, the computers may each function as one of the feature amount calculation unit 2, the similarity degree calculation unit 3, the identification unit 4, and the output information generation unit 22 for example.
Here, a computer that realizes the surgery assistance apparatus by executing the program in the example embodiment will be described with reference to
As illustrated in
The CPU 111 loads the program (codes) in the present example embodiment, which is stored in the storage device 113, onto the main memory 112, and performs various computations by executing these codes in a predetermined order. The main memory 112 is typically a volatile storage device such as a dynamic random access memory (DRAM). Furthermore, the program in the present example embodiment is provided in a state such that the program is stored in a computer readable recording medium 120. Note that the program in the present example embodiment may also be a program that is distributed on the Internet, to which the computer 110 is connected via the communication interface 117.
In addition, specific examples of the storage device 113 include semiconductor storage devices such as a flash memory, in addition to hard disk drives. The input interface 114 mediates data transmission between the CPU 111 and input equipment 118 such as a keyboard and a mouse. The display controller 115 is connected to a display device 119, and controls the display performed by the display device 119.
The data reader/writer 116 mediates data transmission between the CPU 111 and the recording medium 120, and executes the reading of the program from the recording medium 120 and the writing of results of processing in the computer 110 to the recording medium 120. The communication interface 117 mediates data transmission between the CPU 111 and other computers.
Furthermore, specific examples of the recording medium 120 include a general-purpose semiconductor storage device such as a CompactFlash (registered trademark, CF) card or a Secure Digital (SD) card, a magnetic recording medium such as a flexible disk, and an optical recording medium such as a compact disk read-only memory (CD-ROM).
In relation to the above example embodiment, the following Supplementary notes are further disclosed. While a part of or the entirety of the above-described example embodiment can be expressed by (Supplementary note 1) to (Supplementary note 18) described in the following, the present invention is not limited to the following description.
A surgery assistance apparatus including:
a feature amount calculation unit configured to calculate, from a human-body internal image captured using an endoscope, a feature amount of a target part image corresponding to a target part;
a similarity degree calculation unit configured to calculate a similarity degree of the feature amount between different ones of the human-body internal images; and
an identification unit configured to identify the target part image in each of the different human-body internal images if the similarity degree is greater than or equal to a predetermined value.
The surgery assistance apparatus according to Supplementary note 1, wherein
the feature amount includes feature extraction information indicating features of the target part image extracted from the human-body internal image, region information indicating the position and size of a window region with which a part or an entirety of an image corresponding to the target part is detected from the feature extraction information, confidence information indicating the probability of the image in the window region being an image corresponding to the target part, heatmap information calculated using the feature extraction information, or information that is a combination of two or more out of the feature extraction information, region information, confidence information, and heatmap information.
The surgery assistance apparatus according to Supplementary note 2, wherein
the similarity degree calculation unit calculates the similarity degree of the feature amount between the different human-body internal images if the confidence information is greater than or equal to a predetermined confidence value.
The surgery assistance apparatus according to Supplementary note 2 or 3, wherein
the similarity degree calculation unit calculates the similarity degree of the region information if the similarity degree of the confidence information is greater than or equal to a predetermined confidence value, calculates the similarity degree of the feature extraction information if the similarity degree of the region information is greater than or equal to a predetermined region value, and calculates the similarity degree of the heatmap information if the similarity degree of the feature extraction information is greater than or equal to a predetermined feature extraction value, and
the identification unit identifies that the target part images in the different human-body internal images are the same if the similarity degree of the heatmap information is greater than or equal to a predetermined heatmap value.
The surgery assistance apparatus according to any one of Supplementary notes 1 to 4, wherein
the target part is a tumor, and the target part image is an image in which the tumor is captured.
The surgery assistance apparatus according to any one of Supplementary notes 1 to 5, further including
an output device configured to, if the target part image is identified during surgery, output information indicating that the target part image has been identified to an observer during the surgery.
A surgery assistance method including:
The surgery assistance method according to Supplementary note 7, wherein
the feature amount includes feature extraction information indicating features of the target part image extracted from the human-body internal image, region information indicating the position and size of a window region with which a part or an entirety of an image corresponding to the target part is detected from the feature extraction information, confidence information indicating the probability of the image in the window region being an image corresponding to the target part, heatmap information calculated using the feature extraction information, or information that is a combination of two or more out of the feature extraction information, region information, confidence information, and heatmap information.
The surgery assistance method according to Supplementary note 8, wherein
in the (b) step, the similarity degree of the feature amount between the different human-body internal images is calculated if the confidence information is greater than or equal to a predetermined confidence value.
The surgery assistance method according to Supplementary note 8 or 9, wherein
in the (b) step, the similarity degree of the region information is calculated if the similarity degree of the confidence information is greater than or equal to a predetermined confidence value, the similarity degree of the feature extraction information is calculated if the similarity degree of the region information is greater than or equal to a predetermined region value, and the similarity degree of the heatmap information is calculated if the similarity degree of the feature extraction information is greater than or equal to a predetermined feature extraction value, and
in the (c) step, the target part images in the different human-body internal images are identified as being the same if the similarity degree of the heatmap information is greater than or equal to a predetermined heatmap value.
The surgery assistance method according to any one of Supplementary notes 7 to 10, wherein
the target part is a tumor, and the target part image is an image in which the tumor is captured.
The surgery assistance method according to any one of Supplementary notes 7 to 11, further including
(d) a step of, if the target part image is identified during surgery, outputting information indicating that the target part image has been identified to an observer during the surgery.
A computer readable recording medium that includes recorded thereon a surgery assistance program that causes a computer to carry out:
The computer readable recording medium that includes the surgery assistance program recorded thereon according to Supplementary note 13, wherein
the feature amount includes feature extraction information indicating features of the target part image extracted from the human-body internal image, region information indicating the position and size of a window region with which a part or an entirety of an image corresponding to the target part is detected from the feature extraction information, confidence information indicating the probability of the image in the window region being an image corresponding to the target part, heatmap information calculated using the feature extraction information, or information that is a combination of two or more out of the feature extraction information, region information, confidence information, and heatmap information.
The computer readable recording medium that includes the surgery assistance program recorded thereon according to Supplementary note 14, wherein
in the (b) step, the similarity degree of the feature amount between the different human-body internal images is calculated if the confidence information is greater than or equal to a predetermined confidence value.
The computer readable recording medium that includes the surgery assistance program recorded thereon according to Supplementary note 14 or 15, wherein
in the (b) step, the similarity degree of the region information is calculated if the similarity degree of the confidence information is greater than or equal to a predetermined confidence value, the similarity degree of the feature extraction information is calculated if the similarity degree of the region information is greater than or equal to a predetermined region value, and the similarity degree of the heatmap information is calculated if the similarity degree of the feature extraction information is greater than or equal to a predetermined feature extraction value, and
in the (c) step, the target part images in the different human-body internal images are identified as being the same if the similarity degree of the heatmap information is greater than or equal to a predetermined heatmap value.
The computer readable recording medium that includes the surgery assistance program recorded thereon according to any one of Supplementary notes 13 to 16, wherein
the target part is a tumor, and the target part image is an image in which the tumor is captured.
The computer readable recording medium that includes the surgery assistance program recorded thereon according to any one of Supplementary notes 13 to 17, wherein
the surgery assistance program further causes the computer to carry out
(d) a step of, if the target part image is identified during surgery, outputting information indicating that the target part image has been identified to an observer during the surgery.
In such a manner, according to the present invention, the accuracy of endoscopic surgery can be improved by identifying a target part image. The present invention is useful in fields in which endoscopic surgery is necessary.
This application is a continuation application of U.S. application Ser. No. 17/258,296 filed on Jan. 6, 2021, which is a National Stage Entry of PCT/JP2018/025871 filed on Jul. 9, 2018, the contents of all of which are incorporated herein by reference, in their entirety.
Number | Date | Country | |
---|---|---|---|
Parent | 17258296 | Jan 2021 | US |
Child | 18603473 | US |