Recording medium, method for generating learning model, image processing device, and surgical operation assisting system
A non-transitory recording medium recoding a program that causes a computer to execute processing includes: acquiring an operation field image obtained by imaging an operation field of an endoscopic surgery; inputting the acquired operation field image to a learning model trained to output information on a connective tissue between a preservation organ and a resection organ in a case where the operation field image is input, and acquiring information on the connective tissue included in the operation field image; and outputting navigation information when treating the connective tissue between the preservation organ and the resection organ on the basis of the information acquired from the learning model.
1 . A non-transitory recording medium recoding a program that causes a computer to execute processing comprising:
acquiring an operation field image obtained by imaging an operation field of an endoscopic surgery;
inputting the acquired operation field image to a learning model trained to output a recognition result of a connective tissue between a preservation organ and a resection organ where there is a space behind the connective tissue, in a case where the operation field image is input, and recognizing the connective tissue included in the operation field image;
calculating an exposed area of the connective tissue that is exposed as a result of the resection organ being pulled away from the preservation organ, and that is recognized by the learning model; and
outputting an indicator representing a magnitude of the calculated exposed area as navigation information when treating the connective tissue between the preservation organ and the resection organ.
2 . The non-transitory recording medium according to claim 1 ,
wherein the computer is caused to execute processes of,
determining if the connective tissue can be cut on the basis of the calculated exposed area, and
outputting information whether the connective tissue can be cut as the navigation information.
3 . The non-transitory recording medium according to claim 1 ,
wherein the computer is caused to execute processes of,
determining an operating direction of an operation tool when developing or cutting the connective tissue on the basis of the recognition result that is output from the learning model, and
outputting information on the determined operating direction as the navigation information.
4 . The non-transitory recording medium according to claim 1 ,
wherein the resection organ being pulled away from the preservation organ causes tension on the connective tissue thereby putting the connective tissue in a tense state,
wherein the learning model is trained to recognize a connective tissue that is in a tense state in a case where the operation field image is input, and
the computer is caused to execute processes of,
outputting information indicating that cutting the connective tissue is appropriate as the navigation information, in a case where the connective tissue that is in the tense state is recognized by the learning model.
5 . The non-transitory recording medium according to claim 1 ,
wherein the learning model is trained to output information on if the connective tissue between the preservation organ and the resection organ can be cut in a case where the operation field image is input, and
the computer is caused to execute a process of outputting whether the connective tissue can be cut that is obtained from the learning model as the navigation information.
6 . The non-transitory recording medium according to claim 1 ,
wherein the learning model is trained to output information on a cutting site in the connective tissue between the preservation organ and the resection organ in a case where the operation field image is input, and
the computer is caused to execute a process of outputting the information on the cutting site which is obtained from the learning model as the navigation information.
7 . The non-transitory recording medium according to claim 1 ,
wherein the learning model is trained to output a score relating to an anatomical state in a case where the operation field image is input.
8 . The non-transitory recording medium according to claim 7 ,
wherein the learning model is trained so that the score fluctuates in accordance with at least any one among an exposed area of the connective tissue, a tense state of the connective tissue, the number of structures existing at the periphery of the connective tissue, and the degree of a damage of the preservation organ, wherein a tense state of the connective tissue is caused by the resection organ being pulled away from the preservation organ causing tension on the connective tissue.
9 . The non-transitory recording medium according to claim 7 ,
wherein the computer is caused to execute a process of storing the operation field image that is input to the learning model, and the score obtained by inputting the operation field image to the learning model in a storage device in association with each other.
10 . The non-transitory recording medium according to claim 7 ,
wherein the learning model is trained so that the score fluctuates in accordance with a state of a processing target region, and
the computer is caused to execute a process of outputting an instruction to an assistor who assists the endoscopic surgery on the basis of the score that is output from the learning model.
11 . The non-transitory recording medium according to claim 1 ,
wherein the computer is caused to execute processes of,
inputting the acquired operation field image to an image generation model that is trained to generate a prediction image of an anatomical state in a case where the operation field image is input, and acquiring a prediction image with respect to the operation field image, and
outputting the acquired prediction image as the navigation information.
12 . A non-transitory recording medium recoding a program that causes a computer to execute processing comprising:
acquiring an operation field image obtained by imaging an operation field of an endoscopic surgery;
inputting the acquired operation field image to a learning model trained to recognize a connective tissue between a preservation organ and a resection organ by distinguishing between connective tissue under tension and connective tissue not under tension in a case where the operation field image is input, and acquiring information on the connective tissue included in the operation field image, wherein connective tissue under tension occurs when the resection organ is pulled away from the preservation organ, and wherein connective tissue not under tension occurs when the resection organ is not pulled away from the preservation organ; and
displaying textual information on a display screen indicating that connective tissue can be cut on the basis of a calculated exposed area, when recognizing the connective tissue under tension based on an acquired information from the learning model.