Won a national award!

Recently, the State Intellectual Property Office issued the "Decision on the Award of the 23rd China Patent Award", and the inferred medical invention patent "Network Model Training Method and Device, Image Segmentation Method and Device" won the 23rd China Patent Excellence Award !

 inferring that the medical invention patent "network model training method and device, image segmentation method and device" won the 23rd China Patent Excellent prize!

As the highest honor in the field of intellectual property rights in China, the "China Patent Award" is jointly selected and awarded by the State Intellectual Property Office and the World Intellectual Property Organizatio.

It is the only government department award recognized by the United Nations World Intellectual Property Organization in Chin.

 

In medical imaging, accurate segmentation of lesion boundaries is of great significance for the calculation of clinical indicators and the early diagnosis of disease.

However, it is very challenging to accurately and automatically segment lesions due to their uneven representation and lack of background contras.

On the other hand, medical images such as CT and MRI belong to volumetric data with higher data dimension.

Although the introduction of high-dimensional information can help improve the accuracy of lesion segmentation, it often requires more computing resources and higher computer hardware requirement.

In the process of industrializing AI products, how to achieve a balance between the two is also an important issu.

In order to solve this problem, the inference medical research and development of "network model training method and device, image segmentation method and device" patent (hereinafter referred to as "the patent") technolog.

The advantage of this patent is that it does not perform deep learning supervision on pixel-by-pixel recognition tasks in high-dimensional images, does not require segmentation and calculation of high-dimensional data, and does not require a large number of image morphological feature design.

The hyperplane fitting on the surface of the region of interest is used as a supervision, and the calculation of high-dimensional data is naturally converted into a low-dimensional calculation proces.

At the same time, it meets the needs of low computing cost and high segmentation performance in product applications, and solves the key and common medical engineering technical problems in the field of lesion segmentation technology in the AI ​​medical imaging industr.

In terms of practical application, the patented technical solution is an important technical path in the iterative upgrade process of the lesion segmentation function of the two products of InferRead CT Lung and InferRead CT Pneumonia, and provides the calculation of clinical indicators by accurately segmenting the lesion boundar.

and early diagnosis of the disease are important parameter.

This function can greatly save the time and energy of doctors in reading images, delineating the scope of lesions, and calculating various parameters of the lesions, and it is convenient for doctors to apply more time and energy to formulating follow-up follow-up/treatment plans for patient.

At present, these two products have obtained the four major global market access certifications of the US FDA, EU CE, Japan PMDA and China NMPA, and have been used in the auxiliary diagnosis of new coronary pneumonia and early lung cancer screening in many countries, and have been approved by Chinese and foreign medical institutions and Doctors alik.

At present, the patent has been applied to the calculation of lesion segmentation in the field and range of lung CT image.

Recently, Inference Medical has launched innovative solutions in the precise diagnosis and treatment of lung cancer, namely AI target reconstruction of pulmonary nodules and AI-4D chest intelligent visualization syste.

Dozens of related patented technologies have been publishe.
By accurately segmenting the regions of interest in the lungs, the location, volume, density and other information of the lesions can be further determined, which has important clinical value in the analysis and calculation of clinical indicators, early diagnosis of diseases, preoperative planning and intraoperative navigatio.

 

AI Target Reconstruction Solution for Lung Nodules

AI target reconstruction solution for pulmonary nodules AI target reconstruction solution for pulmonary nodules

The patented technology is not limited to CT images, but can also be applied to high-dimensional image data such as MR.
It is not limited to chest images, but can also be applied to lesion segmentation in other parts, such as breast, liver, brain, coronary artery,et.

The "China Patent Award" is an important symbol that reflects the ability of independent innovation and the transformation efficiency of scientific and technological achievement.
The patent of Inference Medical has won the China Patent Excellence Award, which means that Inference Medical's innovative scientific and technological strength has been recognized at the national leve.
As a demonstration unit of intellectual property rights in Beijing, Inference Medical currently has more than 320 patents that have been approved or are in the process of applyin.
In the future, Inference Medical will continue to intensify research and development of innovative technologies, accelerate the transformation of technological achievements into products, and use advanced technology and patented product.
Solve the problem of people's livelihood and healt.

 

Midea takes the lead in medical treatment to build a full map from life to life