Image diagnosis assistance apparatus, data collection method, image diagnosis assistance method, and image diagnosis assistance program

A diagnostic imaging support method using an apparatus, the diagnostic imaging support method comprising:

• obtaining a digestive-organ endoscopic image of a subject captured by an endoscope or an endoscopic video system;
• estimating a location of a lesion present in the digestive-organ endoscopic image of the subject and information on certainty of the location of the lesion using a convolutional neural network,
• performing control to generate an analysis result image showing the location of the lesion and the certainty of the location of the lesion and to display the analysis result image on the digestive-organ endoscopic image,
• subjecting the convolutional neural network to a learning process based on lesion locations of lesions present in a plurality of digestive-organ tumor endoscopic images of a plurality of subjects, the lesion locations of the lesions being determined in advance through feature extraction of atrophy, intestinal metaplasia, mucosal swelling or depression, and a condition of mucosal color tones,
• changing a display style of lesion location information identifying the location of the lesion in the analysis result image in accordance with the certainty, wherein the information on the certainty of the location of the lesion is indicated as a probability score, and
• changing a display color of lesion location information identifying the location of the lesion in the analysis result image when the probability score is higher than a predetermined score.

A non-transitory computer-readable storage medium storing a diagnostic imaging support program for causing a computer to execute:

• a process of obtaining a digestive-organ endoscopic image of a subject captured by an endoscope or an endoscopic video system;
• a process of estimating a location of a lesion present in the digestive-organ endoscopic image of the subject and information on certainty of the location of the lesion using a convolutional neural network,;
• and
• a process of performing control to generate an analysis result image showing the location of the lesion and the certainty of the location of the lesion and to display the analysis result image on the endoscopic image, wherein the convolutional neural network is subjected to a learning process based on lesion locations of lesions present in a plurality of digestive-organ tumor endoscopic images of a plurality of subjects, the lesion locations of the lesions being determined in advance through feature extraction of atrophy, intestinal metaplasia, mucosal swelling or depression, and a condition of mucosal color tones,
• a process of changing a display style of lesion location information identifying the location of the lesion in the analysis result image in accordance with the certainty, wherein the information on the certainty of the location of the lesion is indicated as a probability score, and
• a process of changing a display color of lesion location information identifying the location of the lesion in the analysis result image when the probability score is higher than a predetermined score.