yharyarias/tirads_tiroides

Thyroid ultrasound images, classified into 5 classes that correspond to the European EU-TIRADS scale, this consists of:

EU-TIRADS 1: no nodule
EU-TIRADS 2: benign
EU-TIRADS 3: low risk (oval, smooth margin, iso / hyperechoic, no high risk features)
EU-TIRADS 4: intermediate risk (oval, smooth margin, mildly hypoechoic, no high risk features)
EU-TIRADS 5: any high risk features (non-oval, irregular margin, microcalcifications, marked hypoechogenicity)

Ultrasound images of the thyroid that were taken from the ultrasound scanners of the FOSCAL/FOSUNAB clinic, as a final master's project for the Polytechnic University of Valencia, in collaboration with doctors Federico Lubinus and Boris Marconi, who together with Yhary Arias have worked on the classification of said ultrasounds that are saved in .DICOM format and then transformed to PNG to make the process lighter. The strategy that was carried out for the collection of images and later their labeling was: for each examination that was carried out on patients with or without a possible diagnosis, only the images without personal or sensitive information were kept, all this on a hard drive. , then a pre-processing of the images was done, their format was changed and finally they were mounted on a web page with a single view to facilitate the classification of the doctors who were in charge of this arduous task. Ultrasounds were classified into 5 classes that correspond to the European EU-TIRADS scale, this consists of:

EU-TIRADS 1: no nodule
EU-TIRADS 2: benign
EU-TIRADS 3: low risk (oval, smooth margin, iso / hyperechoic, no high risk features)
EU-TIRADS 4: intermediate risk (oval, smooth margin, mildly hypoechoic, no high risk features)
EU-TIRADS 5: any high risk features (non-oval, irregular margin, microcalcifications, marked hypoechogenicity)

Risk of malignancy
EU-TIRADS 1: n/a
EU-TIRADS 2: 0%
EU-TIRADS 3: low risk (2-4%)
EU-TIRADS 4: intermediate risk (6-17%)
EU-TIRADS 5: high risk (26-87%)

References
1. Gilles Russ, Steen J. Bonnema, Murat Faik Erdogan, Cosimo Durante, Rose Ngu, Laurence Leenhardt. European Thyroid Association Guidelines for Ultrasound Malignancy Risk Stratification of Thyroid Nodules in Adults: The EU-TIRADS. (2019) European ThyroidJournal. 6 (5): 225. doi:10.1159/000478927 - Pubmed
2. Gilles Russ, Bénédicte Royer, Claude Bigorgne, Agnès Rouxel, Marie Bienvenu-Perrard, Laurence Leenhardt. Prospective evaluation of thyroid imaging reporting and data system on 4550 nodules with and without elastography. (2013) European Journal of Endocrinology. 168 (5): 649. doi:10.1530/EJE-12-0936 - Pubmed
3. Jung Hyun Yoon, Kyunghwa Han, Eun-Kyung Kim, Hee Jung Moon, Jin Young Kwak. Diagnosis and Management of Small Thyroid Nodules: A Comparative Study with Six Guidelines for Thyroid Nodules. (2016) Radiology. 283 (2): 560-569. doi:10.1148/radiol.2016160641 - Pubmed
4. Ting Xu, Ya Wu, Run-Xin Wu, Yu-Zhi Zhang, Jing-Yu Gu, Xin-Hua Ye, Wei Tang, Shu-Hang Xu, Chao Liu, Xiao-Hong Wu. Validation and comparison of three newly-released Thyroid Imaging Reporting and Data Systems for cancer risk determination. (2019). Endocrine. 64 (2): 299. doi:10.1007/s12020-018-1817-8 - Pubmed
5. Ting Xu, Ya Wu, Run-Xin Wu, Yu-Zhi Zhang, Jing-Yu Gu, Xin-Hua Ye, Wei Tang, Shu-Hang Xu, Chao Liu, Xiao-Hong Wu. Validation and comparison of three newly-released Thyroid Imaging Reporting and Data Systems for cancer risk determination. (2019). Endocrine. 64 (2): 299. doi:10.1007/s12020-018-1817-8 - Pubmed
6. Grani, Giorgio, Lamartina, Livia, Ascoli, Valeria, Bosco, Daniela, Biffoni, Marco, Giacomelli, Laura, Maranghi, Marianna, Falcone, Rosa, Ramundo, Valeria, Cantisani, Vito, Filetti, Sebastiano, Durante, Cosimo. Reducing the Number of Unnecessary Thyroid Biopsies While Improving Diagnostic Accuracy: Toward the “Right” TIRADS. (2019) The Journal of Clinical Endocrinology & Metabolism. 104 (1): 95. doi:10.1210/jc.2018-01674 - Pubmed
7. Giorgio Grani, Livia Lamartina, Vito Cantisani, Marianna Maranghi, Piernatale Lucia, Cosimo Durante. Interobserver agreement of various thyroid imaging reporting and data systems. (2018) Endocrine Connections. 7 (1): 1. doi:10.1530/EC-17-0336 - Pubmed

Taken from: https://radiopaedia.org/articles/european-thyroid-association-tirads

*Citation Information*

@yharyarias{tirads_tiroides:2022,
author = {Yhary Arias, Federico Lubinus, Boris Marconi},
title = {Common Voice: Thyroid Ultrasound Imaging Dataset},
thesistitle = {Sistema para la clasificación y reconocimiento de imágenes de ultrasonido en
tiroides, basado en técnicas de aprendizaje profundo para el apoyo en el proceso
de diagnóstico según la escala EU-TIRADS},
year = 2022
}

Bucaramanga, Santander, 2022

原始信息汇总

数据集概述

数据集内容

• 类别数量：5类
• 类别描述：
  • EU-TIRADS 1: 无结节
  • EU-TIRADS 2: 良性
  • EU-TIRADS 3: 低风险（椭圆形，平滑边缘，等/高回声，无高风险特征）
  • EU-TIRADS 4: 中等风险（椭圆形，平滑边缘，轻度低回声，无高风险特征）
  • EU-TIRADS 5: 任何高风险特征（非椭圆形，不规则边缘，微钙化，显著低回声）

数据来源

• 采集地点：FOSCAL/FOSUNAB诊所
• 项目背景：作为瓦伦西亚理工大学硕士学位项目，与Federico Lubinus和Boris Marconi医生合作完成。
• 数据处理：原始DICOM格式转换为PNG格式，以减轻处理负担。

风险评估

• 风险等级：
  • EU-TIRADS 1: 不适用
  • EU-TIRADS 2: 0%
  • EU-TIRADS 3: 低风险（2-4%）
  • EU-TIRADS 4: 中等风险（6-17%）
  • EU-TIRADS 5: 高风险（26-87%）

引用信息

• 作者：Yhary Arias, Federico Lubinus, Boris Marconi
• 标题：Common Voice: Thyroid Ultrasound Imaging Dataset
• 论文标题：基于深度学习技术的甲状腺超声图像分类与识别系统，支持EU-TIRADS标准诊断过程
• 年份：2022

搜集汇总

背景与挑战

背景概述

该数据集是一个甲状腺超声图像数据集，按照欧洲EU-TIRADS标准将图像分为5个风险类别，从无结节到高风险特征，用于支持甲状腺结节的诊断和分类。图像来源于临床超声扫描仪，经过预处理和格式转换（从.DICOM到PNG），是瓦伦西亚理工大学硕士项目的成果，与医生合作完成。