Inspection support device, method of operating the inspection support device, and inspection support program

Abstract

To provide an inspection support device for providing comfortable inspection support in response to various requests of doctors although the support device is the one connected to an endoscope system for providing inspection support.SOLUTION: An inspection support device receives selection of a first processing mode and a second processing mode on the processing of a diagnosis part beforehand from a user. In the case where the first processing mode is selected, a captured image is input at least one of a discrimination model and a detection model when captured images continuously acquired by an acquisition part are static images that are static over a preset period, and at least one of an estimation probability and a lesion area is output as a diagnosis result. In the case where the second processing mode is selected, the captured images acquired by the acquisition part are continuously input to the detection model, and the lesion area is sequentially output as a diagnosis result.SELECTED DRAWING: Figure 5

Claims

1. An acquisition unit that sequentially acquires images captured by a camera unit inserted into the subject's body, A diagnostic unit that performs a diagnosis of the body by inputting the acquired image to at least one of the following: a differential model, which is a trained model that outputs at least an estimated probability indicating the likelihood of a lesion being included in a lesion region extracted from the input image; and a detection model, which is a trained model adjusted to determine that a lesion region is present at a lower standard than the differential model, and outputs at least the lesion region extracted from the input image; A receiving unit that receives from the user a selection of processing modes, including at least a first processing mode and a second processing mode, relating to the processing of the diagnostic unit. Equipped with, The diagnostic unit, when the reception unit accepts the selection of the first processing mode, inputs the captured images to at least one of the differentiation model and the detection model if it determines that the captured images acquired continuously by the acquisition unit are still images that remain stationary for a predetermined period of time, and outputs at least one of the estimated probability and the lesion area as a diagnostic result; and when the reception unit accepts the selection of the second processing mode, the acquisition unit continuously inputs the captured images acquired by the acquisition unit to the detection model, and sequentially outputs the lesion area as a diagnostic result; this is an examination support device.

2. The examination support device according to claim 1, wherein the diagnostic unit, when the reception unit accepts the selection of the first processing mode, inputs the captured image to the detection model after the first set period of the period has elapsed and outputs the lesion area, and further, when it is determined that the still image has continued for a second set period which is longer than the first set period, inputs the captured image to the differentiation model and outputs the estimated probability.

3. The examination support device according to claim 1, wherein the diagnostic unit, upon receiving the selection of the first processing mode from the reception unit, determines whether to input to the detection model to output the lesion region or to input to the differentiation model to output the estimated probability, according to the duration for which the acquisition unit determined that the continuously acquired captured images were static images.

4. The reception unit receives from the user the selection of either the first derived mode or the second derived mode as the first processing mode. The diagnostic unit, when the reception unit accepts the selection of the first derived mode as the first processing mode, inputs the captured image to the detection model and outputs the lesion region after the elapsed of the first set period within the period, and when it determines that the still image has continued for a second set period which is longer than the first set period, inputs the captured image to the differentiation model and outputs the estimated probability, and when the reception unit accepts the selection of the second derived mode as the first processing mode, the acquisition unit determines, according to the duration for which it has determined that the continuously acquired captured image has been a still image, whether to input the captured image to the detection model and output the lesion region, or to input the captured image to the differentiation model and output the estimated probability.

5. The diagnostic unit, when the reception unit accepts the selection of the second derived mode as the first processing mode, inputs the captured image to the detection model and outputs the lesion region if the duration is shorter than a preset division period, and inputs the captured image to the differentiation model and outputs the estimated probability if the duration is equal to or greater than the division period, according to claim 4.

6. The examination support device according to any one of claims 1 to 5, wherein the diagnostic unit, when the reception unit accepts the selection of the second processing mode, determines that the captured images acquired by the acquisition unit are still images that remain stationary for a predetermined period of time, inputs the captured images to the discrimination model and outputs the estimated probability as a diagnostic result.

7. The system includes a display control unit that displays the diagnostic results output by the diagnostic unit to a display unit, The display control unit, when the diagnostic unit outputs a plurality of lesion regions using the detection model, displays an index superimposed on the captured image so that each of the lesion regions is visible, and when the diagnostic unit outputs the estimated probability using the differentiation model, even if a plurality of lesion regions are output, superimposes an index on the captured image so that only the single lesion region to be displayed is visible, and displays the estimated probability, as described in claim 1.

8. The system includes a display control unit that displays the diagnostic results output by the diagnostic unit to a display unit, The inspection support device according to claim 1, wherein the display control unit makes the display time for displaying the diagnostic result output by the diagnostic unit using the identification model longer than the display time for displaying the diagnostic result output by the diagnostic unit using the detection model.

9. A method for operating an inspection support device, A reception step in which the reception unit receives a selection from the user from a processing mode that includes at least a first processing mode and a second processing mode related to the diagnostic process, The acquisition step involves the acquisition unit sequentially acquiring images captured by a camera unit inserted into the subject's body, A diagnostic step in which the diagnostic unit performs a diagnosis of the internal organs by inputting the acquired image in the acquisition step into at least one of the following: a differential model which is a trained model that outputs at least an estimated probability indicating the likelihood of a lesion being included in the lesion region extracted from the input acquired image; and a detection model which is a trained model adjusted to determine that it is the lesion region with a lower standard than the differential model, and outputs at least the lesion region extracted from the input acquired image. It has, The diagnostic step is a method for operating an examination support device, wherein, if the selection of the first processing mode is accepted in the acceptance step, the captured images acquired continuously in the acquisition step are determined to be still images that remain stationary for a predetermined period of time, and the captured images are input to at least one of the differentiation model and the detection model, and at least one of the estimated probability and the lesion area is output as a diagnostic result; and if the selection of the second processing mode is accepted in the acceptance step, the captured images acquired in the acquisition step are continuously input to the detection model, and the lesion area is sequentially output as a diagnostic result.

10. A reception step that accepts from the user a selection from processing modes, including at least a first processing mode and a second processing mode, regarding the diagnostic processing, The acquisition step involves sequentially acquiring images captured by a camera unit inserted into the subject's body, A diagnostic step in which the acquired image is input to at least one of the following: a differential model which is a trained model that outputs at least an estimated probability indicating the likelihood of a lesion being included in a lesion region extracted from the input acquired image; and a detection model which is a trained model adjusted to determine that a lesion region is present at a lower standard than the differential model, and outputs at least the lesion region extracted from the input acquired image; and a diagnostic step in which the acquired image is input to at least one of these models to perform the diagnosis of the internal organs. Have the computer run it, The diagnostic step is an examination support program that, when the selection of the first processing mode is accepted in the acceptance step, determines that the captured images acquired continuously in the acquisition step are still images that remain stationary for a predetermined period of time, inputs the captured images into at least one of the differentiation model and the detection model, and outputs at least one of the estimated probability and the lesion area as a diagnostic result; and when the selection of the second processing mode is accepted in the acceptance step, continuously inputs the captured images acquired in the acquisition step into the detection model, and sequentially outputs the lesion area as a diagnostic result.

Images