甲状腺结节是常见的临床疾病，超声检查是其筛查与定性诊断的首选影像学方法。但各超声模态单独用于甲状腺结节进行定性诊断研究均未获得理想结果，且目前尚未形成甲状腺疾病定性诊断绝对可靠的标准，其诊断准确性受医生诊断水平的严重限制。.本项目将模式识别研究与医学临床信息相结合，欲建立国际共享甲状腺超声资料库，以灰阶超声、彩色多普勒血流成像、实时弹性成像、三维能量多普勒血管成像的多模态超声资料作为研究对象，采用模式识别方法对超声资料进行预处理、特征提取、分类，拟获得甲状腺疾病定性诊断的多模态超声综合定量指标。.本研究将多模态资料综合分析，诊断信息更加丰富，使超声检查在甲状腺疾病的诊断过程中发挥最大作用，多模态超声综合定量指标将成为甲状腺疾病超声定性诊断更为详尽、有效的新标准，为甲状腺疾病超声诊断带来突破性进展。

Thyroid nodule is a common clinical disease, and ultrasound examination is the first of choice for its screening and qualitative diagnosis. However, a single module ultrasound examination could not obtain satisfactory results for differentiaton malignant from benign thyrid nodules, and no absolutely reliable criteria for thyroid disease has been found to date, making diagnostic accuracy severly affected by physicians'experience..This project aims to establish world-wide thyroid ultrasound database, combining pattern recognition and clnical information, and to perform pattern recongition technique on multimodality images such as gray scale ultrasound, color Doppler flow imaging, Realtime Elastography, Three-dimensional Power Doppler Angiography. We would obtain quantitive indexes of multimodality ultrasound images for qualitative diagnosis of thyroid disease through image processing, feature extraction, and classfication..This study would make comprehensive analysis on database, provide abudent diagnostic information, maxmize the effect of ultrasound examination on thyroid disease, and utilize those indexes as more accurate and effective criteria for thyroid disease qualitative diagnosis, which could bring ultrasound diagnosis on thyroid disease with breakthroughs.