Evaluation of TCM Tongue Diagnosis Performance For The Assessment Of Liver Fibrosis Based On Deep Learning

注册号:

Registration number:

ITMCTR2100005337

最近更新日期:

Date of Last Refreshed on:

2021-11-27

注册时间:

Date of Registration:

2021-11-27

注册号状态:

Registration Status:

预注册

Prospective registration

注册题目:

基于深度学习探索中医舌诊评估肝纤维化程度的诊断效能

Public title:

Evaluation of TCM Tongue Diagnosis Performance For The Assessment Of Liver Fibrosis Based On Deep Learning

注册题目简写:

English Acronym:

研究课题的正式科学名称:

The First Affiliated Hospital of Sun Yat-sen University

Scientific title:

Evaluation of TCM Tongue Diagnosis Performance For The Assessment Of Liver Fibrosis Based On Deep Learning

研究课题的正式科学名称简写:

Scientific title acronym:

研究课题代号(代码):

Study subject ID:

在二级注册机构或其它机构的注册号:

The registration number of the Partner Registry or other register:

ChiCTR2100053676 ; ChiMCTR2100005337

申请注册联系人:

吕小州

研究负责人:

陈泽雄

Applicant:

Xiaozhou Lv

Study leader:

Zexiong Chen

申请注册联系人电话:

Applicant telephone:

13560354460

研究负责人电话:

Study leader's telephone:

13711191912

申请注册联系人传真 :

Applicant Fax:

研究负责人传真:

Study leader's fax:

申请注册联系人电子邮件:

Applicant E-mail:

lvxzh3@mail.sysu.edu.cn

研究负责人电子邮件:

Study leader's E-mail:

13711191912@163.com

申请单位网址(自愿提供):

Study leader's website(voluntary supply):

研究负责人网址(自愿提供):

Study leader's website
(voluntary supply):

申请注册联系人通讯地址:

广州市越秀区中山二路58号

研究负责人通讯地址:

广州市越秀区中山二路58号

Applicant address:

Guangzhou, Yuexiu District, Zhongshan 2nd Road

Study leader's address:

Guangzhou, Yuexiu District, Zhongshan 2nd Road

申请注册联系人邮政编码:

Applicant postcode:

研究负责人邮政编码:

Study leader's postcode:

申请人所在单位:

中山大学附属第一医院

Applicant's institution:

The First Affiliated Hospital of Sun Yat-sen University

是否获伦理委员会批准:

Approved by ethic committee:

伦理委员会批件文号:

Approved No. of ethic committee:

伦审[2021]464号

伦理委员会批件附件:

Approved file of Ethical Committee:

View

批准本研究的伦理委员会名称:

中山大学附属第一医院临床科研和实验动物伦理委员会

Name of the ethic committee:

IEC for Clinical Research and Animal Trials of the First Affiliated Hospital of Sun Yat-sen&#3

伦理委员会批准日期:

Date of approved by ethic committee:

2021/6/28 0:00:00

伦理委员会联系人:

Contact Name of the ethic committee:

伦理委员会联系地址:

Contact Address of the ethic committee:

伦理委员会联系人电话:

Contact phone of the ethic committee:

伦理委员会联系人邮箱:

Contact email of the ethic committee:

研究实施负责(组长)单位:

中山大学附属第一医院

Primary sponsor:

The First Affiliated Hospital of Sun Yat-sen University

研究实施负责(组长)单位地址:

广州市越秀区中山二路58号

Primary sponsor's address:

No.58 Zhongshan Second road, Guangzhou

试验主办单位(项目批准或申办者):

Secondary sponsor:

国家:

中国

省(直辖市):

广东

市(区县):

广州

Country:

China

Province:

Guangdong

City:

Guangzhou

单位(医院):

中山大学附属第一医院

具体地址:

广州市中山二路58号

Institution
hospital:

The First Affiliated Hospital of Sun Yat-sen University

Address:

Guangzhou Zhongshan Second Road No.58

经费或物资来源:

构建区块链融合联邦学习的肝癌早期筛查粒计算模型

Source(s) of funding:

Build a granular computing model for liver cancer early screening based on fusion of blockchain federated learning

研究疾病:

肝纤维化

研究疾病代码:

Target disease:

Liver Fibrosis

Target disease code:

研究类型:

Study type:

观察性研究

Observational study

研究设计:

Study design:

单臂

Single arm

研究所处阶段:

Study phase:

其它

Others

研究目的:

探索舌诊评估肝纤维化程度的效能,进一步发挥中医舌诊在评估疾病进展、预后及治疗效果的作用。

Objectives of Study:

To explore the effectiveness of tongue diagnosis in evaluating the degree of liver fibrosis, and to further exert the role of tongue diagnosis in evaluating the progress, prognosis and therapeutic effect of disease.

药物成份或治疗方案详述:

1、数据预处理:建立分类文件,给舌图像集、指甲图像集、手掌图像集分别建立训练集、测试集。 2、训练与模型评估:(1)图像预处理:随机将数据集分类80%训练集与20%验证集。将图像进行随机旋转、平移得到数据扩增。(2)将扩增的数据进行训练,其中进行Mask R-CNN舌图像分割模型的训练,使用ResNet50对图像进行细粒度图像分类训练,得到参数。对训练后的模型进行验证,评估诊断效能。分割评估使用均像数准确度、均交并比。分类评估使用F1分数以及受试者工作特征曲线和曲线下面积。 3、将验证后的图像特征参数与患者肝纤维化程度(瞬时弹性彩超结果)、谷丙转氨酶、谷草转氨酶、总胆红素、直接胆红素、间接胆红素、凝血酶原时间进行相关性分析,进一步提取可以评估肝纤维化程度的图像特征。 4、建立测试集,读取测试图片,进行肝纤维化程度的预测,评估舌象、指甲与手掌图像评估肝纤维化程度的效能。

Description for medicine or protocol of treatment in detail:

1. Data preprocessing: establish classification files, establish training sets and test sets for tongue image sets, nail image sets and palm image sets respectively. 2. Training and model evaluation: (1) Image preprocessing: randomly classify 80% training set and 20% validation set. The image was randomly rotated and translated to obtain data amplification. (2) the amplified data are trained, including the segmentation model of Mask r-cnn tongue image, and the fine-grained image classification training of ResNet50 is used to obtain the parameters. The trained model was validated to evaluate the diagnostic efficiency. Mean image number accuracy and mean intersection ratio were used for segmentation evaluation. The classification was evaluated using F1 scores as well as subject operating characteristic curves and area under curves. 3. Correlation analysis was conducted between the verified image feature parameters and the degree of liver fibrosis (instantaneous elastic color ultrasound results), alanine aminotransferase, aspartate aminotransferase, total bilirubin, direct bilirubin, indirect bilirubin and prothrombin time of the patients, and further extraction of image features that can evaluate the degree of liver fibrosis. 4. Establish test sets, read test pictures, predict the degree of liver fibrosis, and evaluate the effectiveness of tongue, nail and palm images in evaluating the degree of liver fibrosis.

纳入标准:

1. 在中山大学附属第一医院超声科行肝脏瞬时弹性检测;2. 具有相关的临床资料、实验室指标;3. 同意被获取舌象、指甲、手掌图像。

Inclusion criteria

1. Transient elastic detection of liver was performed in ultrasound Department of the First Affiliated Hospital of Sun Yat-sen University. 2. Relevant clinical data and laboratory indicators; 3. Agree to be acquired tongue image, fingernail and palm image.

排除标准:

1. 临床资料不全;2. 拒绝拍照。

Exclusion criteria:

1. Incomplete clinical data; 2. Refuse to take pictures.

研究实施时间:

Study execute time:

From 2021-12-01

To      2023-12-31

征募观察对象时间:

Recruiting time:

From 2021-12-01

To      2023-12-31

干预措施:

Interventions:

组别:

1

样本量:

450

Group:

one

Sample size:

干预措施:

干预措施代码:

Intervention:

None

Intervention code:

样本总量 Total sample size : 450

研究实施地点:

Countries of recruitment
and research settings:

国家:

中国

省(直辖市):

广东

市(区县):

广州

Country:

China

Province:

Guangdong

City:

Guangzhou

单位(医院):

中山大学附属第一医院

单位级别:

三甲医院

Institution/hospital:

The First Affiliated Hospital of Sun Yat-sen University

Level of the institution:

Tertiary A Hospital

测量指标:

Outcomes:

指标中文名:

肝脏弹性值

指标类型:

主要指标

Outcome:

Liver elasticity

Type:

Primary indicator

测量时间点:

测量方法:

Measure time point of outcome:

Measure method:

指标中文名:

谷丙转氨酶

指标类型:

主要指标

Outcome:

glutamic-pyruvic transaminase

Type:

Primary indicator

测量时间点:

测量方法:

Measure time point of outcome:

Measure method:

采集人体标本:

Collecting sample(s)
from participants:

标本中文名:

舌象图片

组织:

Sample Name:

Tongue pictures

Tissue:

人体标本去向

其它

说明

Fate of sample 

Others

Note:

征募研究对象情况:

尚未开始

Not yet recruiting

年龄范围:

最小 18
Min age years
最大 80
Max age years

Recruiting status:

Participant age:

性别:

Gender:

男女均可

Both

随机方法(请说明由何人用什么方法产生随机序列):

无分组,无随机方法

Randomization Procedure (please state who generates the random number sequence and by what method):

Groups are not devided

盲法:

Blinding:

是否共享原始数据:

IPD sharing:

Yes

共享原始数据的方式(说明:请填入公开原始数据日期和方式,如采用网络平台,需填该网络平台名称和网址):

在ResMan的IPD共享平 台上传临床试验结果数据

The way of sharing IPD”(include metadata and protocol, If use web-based public database, please provide the url):

Upload clinical trial results data to ResMan's IPD sharing platform

数据采集和管理(说明:数据采集和管理由两部分组成,一为病例记录表(Case Record Form, CRF),二为电子采集和管理系统(Electronic Data Capture, EDC),如ResMan即为一种基于互联网的EDC:

研究数据的纪录及保存方式(原始记录、电子CRF等):纸质数据专人管理收集;电子数据专用电脑专人设置密码的EXCEL电子表格管理数据。数据质量保障措施:电子数据一式两份,两名数据管理员各自录入和保存数据,并定期核对和复合数据。

Data collection and Management (A standard data collection and management system include a CRF and an electronic data capture:

Research data recording and preservation methods (original records, electronic CRF, etc.): Paper data are managed and collected by special personnel; Electronic data dedicated computer specially-assigned password EXCEL spreadsheet management data. Data quality assurance measures: Electronic data in duplicate, two data managers input and save data separately, and check and compound data regularly.

数据管理委员会:

Data Managemen Committee:

研究计划书或研究结果报告发表信息
(杂志名称、期、卷、页,时间;或网址):

Publication information of the protocol/research results report
(name of the journal, volume, issue, pages, time; or website):

ITMCTR BJ-ICP:07032215-5 Tip: IE8 is recommended Use the system with widescreen display resolution above