Thyroid
Based on region segmentation algorithm
Thyroid occupying lesions Pathological analysis

Thyroid nodule recognition

This software utilizes AI technology to automatically recognize the contour of thyroid nodules in ultrasound, and during ultrasound scanning, real-time reminders are given to the location of nodules in the ultrasound image, greatly improving the accuracy of doctors in scanning nodules.


This software adopts a self-developed UNET+custom network layer semantic segmentation model, and uses 20000 ultrasound data for model training.

Analysis of thyroid nodule characteristics

This software utilizes AI technology to automatically identify the composition, echogenicity, aspect ratio, edge boundaries, and calcification of thyroid nodules, and assists TIRADS grading to improve the accuracy of thyroid scanning by doctors.


This software adopts a self-developed classification network model and uses 40000 ultrasound data for model training.

Usage scenario

This software is mainly used for cardiovascular and cerebrovascular ultrasound examination in physical examination centers, as well as health screening in counties and townships. According to the research data of The Lancet, the global population with carotid atherosclerosis will reach about 2 billion in 2020. Among them, there are approximately 270 million people in China, with carotid artery plaques accounting for approximately 200 million. Cardiovascular and cerebrovascular ultrasound examination, as one of the main items of physical examination, has a wide scope, a large number of people, and a high workload for doctors.


This software helps doctors to quickly complete the examination through automatic plaque recognition, intimal thickness measurement and other functions, assists in analyzing the risk possibility of carotid atherosclerosis, transient ischemic attack, reversible neurological dysfunction and other diseases, and helps prevent high-risk diseases such as ischemic stroke and arterial embolism.

Performance

99
%
Sensitivity
95
%
Specificity
96
%
AUC

*Supports real-time computing(FPS>12)

Deployment

Using ultrasound models: GE/Philips/Mindray/Domestic handheld ultrasound

Available computing units: mobile phone/NVIDIA Xavier/computer/Cloud

Available deployment platforms: Android/Windows/Linux

Maturity

>5000
Number of applied medical institutions
>2500
h
Model training duration

Registration certificate

Yes

logo
Address: Jintie Square, Putuo District, Shanghai
Phone: 021- 01204327
E-mail: danghoang87hl@senz.com
Official Public Account
Official support
Microcheck app
Outpatient app
京ICP备12019481号-5