Fujifilm Corporation

Toshihiro Usuda

Masaaki Oosake

There are provided a machine learning system, a learning method, and a program that can facilitate learning of a large number of combined medical images. A machine learning system () includes: an image database () that stores a plurality of medical images; a geometric-shape database () that stores geometric shapes to be superimposed on the medical images; a processor (); and a learning model (), in which the processor () is configured to perform a selection accepting process of accepting selection of a medical image from among the medical images stored in the image database () and selection of a geometric shape from among the geometric shapes stored in the geometric-shape database (), a geometric-shape combining process of combining the selected medical image and the selected geometric shape and generating a composite image, and a training process of making the learning model perform learning by using the composite image.

Machine learning system, recognizer, learning method, and program

An object is to provide an endoscopic image learning device, an endoscopic image learning method, an endoscopic image learning program, and an endoscopic image recognition device that appropriately learn a learning model for image recognition for recognizing an endoscopic image in which a treatment tool for an endoscope appears. The object is achieved by an endoscopic image learning device including an image generation unit and a machine learning unit. The image generation unit generates a superimposed image where a foreground image in which a treatment tool for an endoscope is extracted is superimposed on a background-endoscopic image serving as a background of the foreground image, and the machine learning unit performs the learning of a learning model for image recognition using the superimposed image.

Endoscopic image learning device, endoscopic image learning method, endoscopic image learning program, and endoscopic image recognition device

An information processing apparatus detects, by a first detector, a region-of-interest reflected in medical image data, performs a control of displaying, on a display, the medical image data which is a detection result indicating that the region-of-interest is not detected, receives information for specifying the medical image data evaluated that the detection result is false negative by a user, re-detects, by a second detector, the region-of-interest based on the false negative medical image data, and trains the first detector using a first medical image data set including the medical image data and a re-detection result.

Information processing apparatus, training data generation device, and diagnosis support system

An endoscope system includes an image obtaining unit that obtains an image obtained by image capturing of a photographic subject, a recognition unit that performs a recognition process of recognizing the photographic subject by using the image, an identification unit that identifies an operation performed on the photographic subject, a setting unit that enables or disables the recognition unit in accordance with the result of identification by the identification unit, and a notification unit that gives a notification of whether the recognition unit is enabled or disabled.

Endoscope system

The present invention is directed to providing an image processing apparatus and an image processing method that assist an observer of a time-series image in performing smooth observation while attracting attention of the observer. An image processing apparatus according to a first aspect of the present invention includes an image input unit that inputs a time-series image; an information acquiring unit that acquires information on a region of interest in the time-series image; a figure calculating unit that calculates a polygon or an ellipse surrounding the region of interest and having an area smaller than or equal to an area of a circumscribed rectangle, the circumscribed rectangle being a rectangle circumscribing the region of interest and formed of two sides parallel to a horizontal axis of the time-series image and other two sides parallel to a vertical axis of the time-series image; and a display control unit that causes a display apparatus to display a figure such that the figure is superimposed on the time-series image, the figure being based on the polygon or the ellipse and being disposed at sides or vertexes of the polygon or disposed on a circumference of the ellipse or at a plurality of positions on the circumference.

Image processing apparatus and image processing method

Provided are a learning device, a learning method, and a learned model capable of allowing an operator to ascertain what part inside a body of a subject is observed in the middle of examination by medical equipment. In a learning device, a learning method and, a learned model, an image acquisition unit acquires a medical image of an observation target part inside the body of the subject in each of a plurality of observation steps in which an observation order of the observation target part is determined, a step prediction unit outputs a prediction step predicting an observation step corresponding to the medical image from the medical image based on a learning result in each observation step, a step comparison unit compares the prediction step with a ground truth step corresponding to the prediction step in each observation step, and a learning controller makes the step prediction unit learn a relationship between the medical image and the prediction step based on a result of comparison by the step comparison unit in each observation step such that the prediction step becomes close to the ground truth step.

Learning device, learning method, and learned model

A processor device includes an image signal acquiring unit, an image processing unit, and a display control unit. The image signal acquiring unit acquires a digital image signal corresponding to an observation mode from an endoscope. The image processing unit includes a region-of-interest-detection-mode image processing unit. The region-of-interest-detection-mode image processing unit detects a region of interest from an endoscopic image and outputs a detection result of the region of interest and a cumulative detection time period of the region of interest.

Medical image processing apparatus, endoscope system, and method for operating medical image processing apparatus

An endoscope image processing apparatus includes: an image acquisition unit that acquires endoscopic images; a detection unit that detects lesion images representing lesions in the endoscopic images acquired by the image acquisition unit; a clustering unit that groups the endoscopic images on the basis of a degree of correlation between the lesion images and generates, for each lesion, a group formed of corresponding ones of the endoscopic images; a representative image selection unit that selects, for each group, a representative image from among the endoscopic images in the group; a saving unit that saves, for each group, the representative image and the endoscopic images that form the group to which the representative image belongs, in association with each other; and a display unit that displays a list of the representative images saved in the saving unit.

Endoscope image processing apparatus, endoscope image processing method, and program

An endoscope apparatus includes: a lesion information acquisition unit that acquires acquired lesion information; a lesion identification unit that compares existing lesion information with the acquired lesion information and determines whether the acquired lesion information corresponds to the same lesion indicated by the existing lesion information; a lesion count unit that counts a lesion count, which is the number of lesions detected in a single endoscopic examination; a display unit; and a control unit that increments the lesion count counted by the lesion count unit by one when the lesion identification unit determines that the acquired lesion information does not correspond to the same lesion indicated by the existing lesion information.

Endoscope apparatus, endoscope operation method, and program

A region of interest located in a first part to be observed is detected from a medical image at a first timing. In a case where a tip portion of an endoscope moves from the first part to be observed toward a second part to be observed and then moves to the first part to be observed again, a return determination unit performs first determination processing for determining whether the tip portion has returned to the first part to be observed, by using a medical image acquired a second timing. A display control unit performs display control of a monitor by using a determination result of the return determination unit.

Endoscope system and medical image processing system

There are provided an endoscopic image processing device, an endoscopic image processing method, an endoscopic image processing program, and an endoscope system that appropriately notify region-of-interest information about a region of interest included in an endoscopic image according to an endoscope operator's action. An endoscope operator's action on a portion to be observed of an endoscopic image is recognized from endoscopic images, and first emphasis display where region-of-interest information is displayed in the endoscopic image at a first emphasis level and second emphasis display where the region-of-interest information is displayed at a second emphasis level relatively lower than the first emphasis level are switched according to a recognition result.

Endoscopic image processing device, endoscopic image processing method, endoscopic image processing program, and endoscope system

Provided are a medical image processing apparatus, a medical image processing method, a program, and a diagnosis support apparatus that report a region of interest without hindering observation of the boundary between the region of interest and a region of non-interest in a medical image. A medical image processing apparatus includes a superimposition processing unit that superimposes, on a medical image, a figure for reporting a region of interest included in the medical image. The superimposition processing unit superimposes the figure on an inside of the region of interest such that at least part of a boundary between the region of interest and a region of non-interest is not superimposed with the figure, thereby reporting the region of interest without hindering observation of the boundary between the region of interest and the region of non-interest.

Medical image processing apparatus, medical image processing method, program, and diagnosis support apparatus

An endoscope device, an endoscope operation method, and a program are provided. An endoscope device includes a motion pattern identification unit that identifies a single or plurality of motion patterns indicating a specific movement of a scope head of a scope unit, an operation storage unit that stores a single or plurality of pieces of operation information for instructing an operation of the endoscope device, and the single or plurality of motion patterns of the scope head in association with each other, an operation information acquisition unit that acquires the operation information corresponding to the motion pattern from the operation storage unit on the basis of the motion pattern identified by the motion pattern identification unit, and an operation execution unit that executes an operation corresponding to the operation information acquired by the operation information acquisition unit.

Growing community of inventors

Kanagawa, Japan

Toshihiro Usuda