Abstract: An endoscope processor device generates first video signals. A recognition processing processor device generates second video signals by reflecting a result of recognition processing based on the first video signals. A display displays any one of the second video signals or the first video signals switched from the second video signals on the basis of first video switching signals from the recognition processing processor device. The display displays that a result display of the recognition processing is being stopped in a case where the result display of the recognition processing by the second video signals is stopped.

Abstract: An object of the present invention is to provide an image processing method and an image processing apparatus that make it possible to efficiently learn images having different identities. In learning and recognition using a hierarchical network, it is known that, based on experiences, a layer near the input functions as a feature extractor for extracting a feature that is necessary for recognition, and a layer near the output performs recognition by combining extracted features. Thus, performing learning by setting a higher learning rate to a layer near the input side of the hierarchical network than a learning rate in a layer near the output side in second learning processing as in an aspect of the present invention corresponds to mainly relearning (adjusting) a feature extraction portion in data sets having different identities. Accordingly, the difference between data sets can be absorbed, and learning can be performed more efficiently than in the case of simply performing transfer learning.

Abstract: A medical image processing apparatus includes a recognition unit that detects a region of interest from an acquired medical image and classifies the region of interest; a display control unit that causes emphasis display for emphasizing the region of interest and a classification for the region of interest to be displayed in a screen identical to a screen for displaying an observation target included in the medical image; and an observation image switching unit that switches whether to display consecutive images or to display a still image by using a display device. In a case where consecutive image display of an observation image is switched to still image display by the observation image switching unit, the display control unit makes a change for reducing visual recognizability of at least any of the emphasis display or the classification compared with a case of the consecutive image display of the observation image.

Abstract: An object of the present invention is to provide an image processing device, an endoscope system, and an image processing method capable of observing an accurate structure of a subject while preventing a substantial decrease in a frame rate in a case where an image is acquired using a plurality of observation lights.

Abstract: A medical image processing apparatus according to one aspect of the present invention is a medical image processing apparatus including a processor. The processor is configured to execute an image acquisition process of acquiring a time-series medical image; a region-of-interest detection process of detecting a region of interest from the acquired medical image; a display control process of causing a display apparatus to display the medical image; a first reporting process of, in response to the region of interest being detected in the region-of-interest detection process, causing the display apparatus to perform superimposed display of the medical image and information about the detected region of interest; and a second reporting process of, in response to the region of interest being detected in the region-of-interest detection process, outputting a sound from a sound output apparatus. The processor is configured to execute the second reporting process after the first reporting process.

Abstract: A medical image processing apparatus includes a processor which is configured to execute an image acquisition process of acquiring a time-series medical image; a plurality of detection processes of detecting a region of interest from the acquired medical image; a display control process of causing a display apparatus to display at least one detection result among detection results obtained in the plurality of detection processes; a selection process of selecting, from among the plurality of detection processes, a detection process for which a detection result is to be displayed on the display apparatus; and a switching control process of controlling, in accordance with a detection state of the region of interest in the plurality of detection processes, whether or not to permit switching of a detection process for which a detection result is to be displayed to the selected detection process.

Abstract: In a medical image processing apparatus including a processor, the processor sequentially acquires time-series medical images and causes a display unit to sequentially display the acquired medical images. Further, the processor performs a process of acquiring, based on the acquired medical images, information related to a position of a region of interest in the medical images and classifying the region of interest into a class among a plurality of classes, and displays class information indicating the class of the classified region of interest such that the class information is superimposed at a position of the region of interest in a medical image displayed on the display unit. Further, the processor changes a relative position of the superimposed class information with respect to the region of interest, in accordance with an elapsed time from recognition of the region of interest.

Abstract: Provided are a medical image processing apparatus, a medical image processing method, and a medical image processing program by which it is possible to automatically perform time measurement related to an examination state of time-series medical images. In a medical image processing apparatus including a processor, the processor is configured to perform: medical image acquisition processing for sequentially acquiring time-series medical images; examination state recognition processing for recognizing an examination state on the basis of the sequentially acquired medical images; time measurement processing for performing time measurement related to the examination state; time measurement control processing for controlling a behavior of the time measurement in accordance with a recognition result of the examination state; and report processing for causing a report unit to report information on a state of the time measurement to a user.

Abstract: A processor of an image processing device acquires a medical image including a lesion region, and causes a determiner to perform a determination process of determining a type of a lesion on the basis of the medical image. A first determination process is a process of determining whether the type belongs to a first group including a first type classified as a first category and a second type classified as a second category or a second group including a third type classified as the second category. A second determination process is a process of determining whether the type is the first type or the second type. The processor outputs a first signal for specifying whether the type is classified as the first category or the second category on the basis of a determination result of the first determination process and a determination result of the second determination process.

Abstract: A medical image processing apparatus includes: an image acquisition unit that acquires a plurality of medical images; an image classification unit that classifies the medical images into at least one of a plurality of categories; an image display unit that displays, on a screen, at least one of the plurality of medical images based on a result of the classification as an automatically selected image; an input receiving unit that receives an input from a user to select an image requiring reselection as a key image from among the displayed automatically selected images as a selected image; and a display control unit that switches and displays, as a switching image, a non-automatically selected image other than the automatically selected image among the plurality of medical images classified into each category, instead of displaying the selected image.

Abstract: An input receiving unit receives a designation input of a user for a medical image displayed on a first display screen of a display. An image selection unit acquires selected images selected from a plurality of medical images based on a designated image category that is a category assigned to a designated image and a non-designated image category that is a category assigned to a non-designated image. The selected images are displayed in a list on a second display screen of the display.

Abstract: The present invention provides a trained model conversion method, an inference method, a trained model conversion apparatus, a trained model, and an inference apparatus that are capable of reducing the cost of processing by a regularization layer. A trained model conversion method according to an aspect of the present invention includes a convolutional layer generation step of generating, for a trained convolutional neural network including at least one regularization layer, a second convolutional layer on the basis of a trained parameter of the regularization layer and a trained parameter of a first convolutional layer adjacent to the regularization layer; and a converted model generation step of replacing the regularization layer and the first convolutional layer with the second convolutional layer to generate a converted model which is a converted trained model.

Abstract: There is provided a filing device, a filing method, and a program that cause a still image and a motion image to be associated with each other, which are captured without being associated with each other. The problem is solved through a filing device including an acquisition unit that acquires at least one still image and at least one motion image, a still image analysis unit that analyzes a pixel value of the still image and extracts still image information, a motion image analysis unit that analyzes a pixel value of the motion image and extracts motion image information, and an associating unit that associates the still image with the motion image by comparing the still image information with the motion image information.

Abstract: Provided are endoscopic image acquisition system and method that make observation of images easier when images are collected. The endoscopic image acquisition system includes a wavelength pattern changing unit that changes a wavelength pattern of irradiation light with which a part to be observed in a body cavity of a patient is irradiated or returning light from the part to be observed. Images of an observation wavelength pattern are captured at a certain frame rate. In response to acceptance of an acquisition instruction, images of a plurality of wavelength patterns different from one another are sequentially captured. The images of the plurality of wavelength patterns different from one another are stored in a storage unit. An image of a wavelength pattern other than the observation wavelength pattern is set not to be displayed.

Abstract: There are provided a medical image processing device, an endoscope system, a medical image processing method, and a program which detect an optimal lesion region according to an in-vivo position of a captured image. Images at a plurality of in-vivo positions of a subject are acquired from medical equipment that sequentially captures and displays in real time the images; positional information indicating the in-vivo position of the acquired image is acquired; from among a plurality of region-of-interest detection units that detect a region of interest from an input image and correspond to the plurality of in-vivo positions, respectively, a region-of-interest detection unit corresponding to the position indicated by the positional information is selected; and the selected region-of-interest detection unit detects a region of interest from the acquired image.

Abstract: A medical image processing device performs recognition processing on an acquired medical image, performs a control of displaying, on a display, the medical image and a result of the recognition processing of the medical image, receives an evaluation related to the result of the recognition processing from a user based on the displayed medical image and the displayed result of the recognition processing, determines whether or not to store the medical image, which is a target of the evaluation, in a data storage unit based on the evaluation, and performs a control of storing, in the data storage unit, the medical image determined to be stored.

Abstract: A trained first model is generated through first learning using a first learning image group constituted of a normal image which is a medical image having no region of interest. An input image group including at least the medical image different from the first learning image group is input to the trained first model, and abnormality detection is performed. The extracted image used for learning to prevent erroneous recognition is sorted according to a result of the abnormality detection, and second learning using a second learning image group including at least the extracted image is performed. A second model that detects the region of interest is generated through the second learning.

Abstract: An endoscope processor device generates first video signals. A recognition processing processor device generates second video signals by reflecting a result of recognition processing based on the first video signals. A display displays any one of the second video signals or the first video signals switched from the second video signals on the basis of first video switching signals from the recognition processing processor device. The display displays that a result display of the recognition processing is being stopped in a case where the result display of the recognition processing by the second video signals is stopped.

Abstract: An image processing apparatus according to a first aspect of the present invention acquires a first image and a second image in a first image acquisition mode, or a second image acquisition mode in which a second image acquisition ratio is higher than in the first image acquisition mode, on the basis of a detection result of a specific target (whether or not a specific target has been detected, what type of specific target). For example, in accordance with whether or not a specific target has been detected and the type of specific target, the first image and the second image can be acquired in the first image acquisition mode in a case where the necessity for acquiring the second image is low, whereas the first image and the second image can be acquired in the second image acquisition mode, in which the second image acquisition ratio is high, in a case where the necessity for acquiring the second image is high.

Abstract: There are provided a medical image processing device, an endoscope system, a medical image processing method, and a program which detect an optimal lesion region according to an in-vivo position of a captured image. Images at a plurality of in-vivo positions of a subject are acquired from medical equipment that sequentially captures and displays in real time the images; positional information indicating the in-vivo position of the acquired image is acquired; from among a plurality of region-of-interest detection units that detect a region of interest from an input image and correspond to the plurality of in-vivo positions, respectively, a region-of-interest detection unit corresponding to the position indicated by the positional information is selected; and the selected region-of-interest detection unit detects a region of interest from the acquired image.