20240005498. METHOD OF GENERATING TRAINED MODEL, MACHINE LEARNING SYSTEM, PROGRAM, AND MEDICAL IMAGE PROCESSING APPARATUS simplified abstract (FUJIFILM Corporation)

METHOD OF GENERATING TRAINED MODEL, MACHINE LEARNING SYSTEM, PROGRAM, AND MEDICAL IMAGE PROCESSING APPARATUS

Organization Name

FUJIFILM Corporation

Inventor(s)

Akira Kudo of Tokyo (JP)

METHOD OF GENERATING TRAINED MODEL, MACHINE LEARNING SYSTEM, PROGRAM, AND MEDICAL IMAGE PROCESSING APPARATUS - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240005498 titled 'METHOD OF GENERATING TRAINED MODEL, MACHINE LEARNING SYSTEM, PROGRAM, AND MEDICAL IMAGE PROCESSING APPARATUS

Simplified Explanation

The abstract describes a learning model that uses a generative adversarial network to generate images in a different domain based on medical images. The model includes a generator and a discriminator, both implemented using convolutional neural networks. The computer acquires training data consisting of medical images from two domains and performs training using this data.

• The learning model uses a generative adversarial network structure.
• The generator is a convolutional neural network that takes a medical image from one domain as input and generates an image in a different domain.
• The discriminator is another convolutional neural network that receives input data consisting of the generated image or a medical image from the second domain, along with coordinate information of the human body.
• The discriminator's role is to determine the authenticity of the input image.
• The computer acquires training data containing medical images from both domains.
• The computer performs training processing using the acquired training data.

Potential Applications

• Medical image translation: The model can be used to generate images in a different domain, such as converting X-ray images to MRI images or vice versa.
• Data augmentation: The generated images can be used to augment existing medical image datasets, increasing the diversity and size of the training data.
• Image synthesis: The model can be used to synthesize realistic medical images for various purposes, such as training other machine learning models or simulating medical scenarios.

Problems Solved

• Limited availability of labeled medical image data: The model can generate synthetic images in a different domain, allowing for the creation of larger and more diverse datasets for training machine learning models.
• Domain adaptation: The model can bridge the gap between different domains of medical images, enabling the transfer of knowledge and insights from one domain to another.
• Image synthesis challenges: The model addresses the challenge of generating realistic and high-quality medical images, which can be useful for various medical imaging applications.

Benefits

• Improved data availability: The model enables the generation of synthetic medical images, reducing the reliance on scarce and expensive labeled data.
• Enhanced training and performance: By augmenting existing datasets and bridging domain gaps, the model can improve the training and performance of machine learning models in medical imaging tasks.
• Cost and time savings: Generating synthetic images can save time and resources compared to collecting and labeling new medical image data.

Original Abstract Submitted

by using a learning model having a structure of a generative adversarial network including a first generator configured using a first convolutional neural network that receives an input of a medical image of a first domain and that outputs a first generated image of a second domain, and a first discriminator configured using a second convolutional neural network that receives an input of data including first image data, which is the first generated image or a medical image of the second domain included in a training dataset and coordinate information of a human body coordinate system corresponding to each position of a plurality of unit elements configuring the first image data, and that discriminates authenticity of the input image, a computer acquires a plurality of pieces of training data including the medical image of the first domain and the medical image of the second domain; and performs training processing.