多模态超声组学结合超声原始RF信号对甲状腺癌风险评估和预后判断研究

甲状腺癌预后呈两极分化特点，即大部分预后较好，但一部分患者癌灶侵袭性、转移性高且预后差。目前，包括TNM（AJCC）、MACIS、GAMES等在内的评估方法，并未针对上述两极分化的预后特点建立相应判断模型，导致其临床上易出现过度治疗或治疗不足等情况。超声是甲状腺癌首选检查方式，如何在术前综合分析包括超声影像、超声造影等静态、动态图像，同时结合超声原始射频（RF）信号等原始超声数据，在大样本层面对多种模态的信息进行有效融合，并建立相应预后判断风险预测模型，是甲状腺癌相关诊疗技术发展的未来趋势。同时，该研究对规范治疗手段、精准判断预后、减少患者负担、提高患者生活质量等问题都具有重要临床价值。因此，本申请提出基于多模态超声数据及新兴人工智能、大数据和超声影像组学等技术，对甲状腺癌进行多维度术前病理亚型分析、临床分期预测和预后判断研究，为甲状腺癌的精准诊疗提供新的参考依据和评价手段。

Prognosis of thyroid carcinoma is characterized by polarization, and it means most of the patients have good prognosis, but some of them who have aggressive, metastatic cancer focus have poor one. Up to now, corresponding judgment models for the prognostic characteristics of the above-mentioned polarization have not been established in a flood of assessment methods including TNM (AJCC), MACIS, GAMES, etc., thereby possibly leading to clinical over-treatment or under-treatment. Ultrasound is the first choice for thyroid cancer examination, and the future trend in the development of thyroid carcinoma-related medical technology includes how to comprehensively analyze static and dynamic images including ultrasonic images and ultrasound contrast before surgery, and combine original ultrasound data such as raw radio frequency (RF) signals to analyze the information with multiple modalities at the large sample level and to establish a corresponding prognosis judgment risk prediction model. Meanwhile, the study has important clinical value for standardizing treatment methods, accurately predicting prognosis, reducing patient burden and improving quality of life. Therefore, based on multi-modal ultrasound data and emerging artificial intelligence, this application is proposed to conduct the analysis of multi-dimensional preoperative pathological subtype and prediction of clinical stage and prognosis, which will provide a new reference basis and evaluation method for precise diagnosis and treatment of thyroid carcinoma.