AI-SONIC-TH flagged suspicious thyroid nodules in China's review

In the maker's trial that China's CMDE reviewed, AI-SONIC-TH flagged patients with thyroid nodules of TI-RADS 4 or higher with 97.07% sensitivity and 90.54% specificity.

  • Not rated an organization

    The accuracy was measured in the maker's pre-market trial, and the report does not say whether the prompts changed any patient's care.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Intended benefit.
Evidence

foundstudiedreportedclaimed

An independent regulatory document supports the benefit.

Compared with

Targets for the lower 95% confidence bounds of per-patient sensitivity and specificity (the report does not give the values). Measured.

The system’s part
Not isolated. The trial measured the software's own prompts per patient and also scored doctors reading with it. It gave no figures for doctors reading alone, so the software's effect on their judgment is not isolated. The trial was the maker's own. The report says that the diagnosis module contains a deep-learning algorithm.
Not yet checked
LimitsCosts and harmsLearns from dataLater outcome

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
AI-SONIC-TH (甲状腺结节超声影像辅助诊断软件)
Made by
Zhejiang Deshang Yunxing Medical Technology Co., Ltd. (浙江德尚韵兴医疗科技有限公司)
Used by
Zhejiang Deshang Yunxing Medical Technology (prospective registration trial with 1,043 patients at three sites)
Where
Three trial sites that the report does not name, for a registration in China
Stage
In a pilot
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Identifying Objects, Actions, and Events
Collections
Scrutiny and findings
Last revised
1 November 2026

Sources

  1. Regulatory document, by authors with no tie to the system.

    Center for Medical Device Evaluation (CMDE), National Medical Products Administration, China. 医疗器械产品注册技术审评报告：甲状腺结节超声影像辅助诊断软件 (technical review report for the registration of thyroid nodule ultrasound image diagnosis-aid software), acceptance no. CQZ2200673, applicant 浙江德尚韵兴医疗科技有限公司. 3 November 2023. Section 三、临床评价概述 (clinical evaluation).

    以患者为单位，器械独立阅片的灵敏度为97.07%（95%CI：95.06%，98.28%） ，特异度为 90.54%（95%CI：87.91%，92.64%） ，95%置信区间下限均大于目标值
    认为该产品的上市为适用人群带来的受益大于风险。

Details

AI system

Reported configuration

Release 1 (models AI-SONIC-TH-B to AI-SONIC-TH-I)

Workflow

A sonographer outlined each nodule by hand on a grayscale ultrasound image imported into the software. The deep-learning algorithm estimated the probability of each ACR TI-RADS category and suggested category 3 or lower or category 4 or higher.

Follow-up

Conditions for replication

Patients have had no thyroid nodule surgery, ablation, radiotherapy or chemotherapy. Images come from the specified ultrasound devices and probes. Sonographers make the final imaging diagnosis, and clinicians do not decide on the software's result alone.

Entry details

Sector

Healthcare.

Purpose

Zhejiang Deshang Yunxing Medical Technology built AI-SONIC-TH to tell sonographers whether a thyroid nodule that they outline on an ultrasound image is suspected ACR TI-RADS category 4 or higher.

30 more questions not yet checked

model, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). AI-SONIC-TH flagged suspicious thyroid nodules in China's review (AIBD-1089). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-1089.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)

BibTeX Data Field definitions