Quantitative, non-invasive characterization of urinary stone composition and fragility using multi-energy CT and machine learning techniques

使用多能量 CT 和机器学习技术对尿路结石成分和脆性进行定量、非侵入性表征

• 批准号: 10377461
• 负责人: Cynthia H McCollough
• 金额: $ 35.75万
• 依托单位: MAYO CLINIC ROCHESTER
• 依托单位国家: 美国
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-06-01 至 2023-11-16
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10377461
• 关键词: Affect; Alkalies; Bilateral; Biological Markers; Biometry; Calcium Oxalate; Characteristics; Classification; Clinical Research; Clinical Treatment; Cost Effective Management; Coupled; Data; Disease; Dose; Economic Burden; Effectiveness; Excision; Future; Generations; Goals; Image; Imaging Techniques; Imaging technology; Injury to Kidney; Kidney Calculi; Knowledge; Location; Machine Learning; Measurement; Measures; Medical; Medical Care Costs; Methodology; Methods; Minerals; Morphology; Outcome; Patients; Percutaneous Nephrolithotomy; Phenotype; Physicians; Population; Prevalence; Procedures; Publishing; Recovery; Research; Research Personnel; Research Subjects; Resolution; Risk; Scientific Advances and Accomplishments; Scientist; Series; Shapes; Source; Structure; Surface; Techniques; Technology; Testing; Texture; Time; United States; Ureteroscopy; Uric Acid; Urinary Calculi; Validation; X-Ray Computed Tomography; attenuation; base; calcium phosphate; clinical investigation; cost; deep learning; evidence base; health management; high risk; imaging modality; in vivo; innovation; learning strategy; novel; photon-counting detector; prevent; risk minimization; statistical learning; treatment strategy

PROJECT SUMMARY Symptomatic urinary stone disease (USD) affects >8% of the United States population, resulting in an estimated annual medical cost exceeding $10 billion. Computed tomography (CT) is the established method for imaging urinary calculi and can provide accurate sub-millimeter details of the size and location of renal stones. However, in vivo characterization of more than just size and location is critical for quantifying stone characteristics important for optimal patient health management and essential for clinical research. A complete characterization of renal stones, including stone composition and fragility, is needed for safe and cost effective management of USD, as well as for phenotyping of research subjects. Our proposal meets these needs by developing methods to accurately and non-invasively characterize stones using low-dose, multi-energy CT. Our long-term goal is to use advanced CT methodologies to characterize urinary calculi for the purpose of directing clinical treatment and facilitating clinical investigation. Our objectives in this application are to develop and validate in vivo quantitative techniques for characterizing mixed and non-uric-acid stone types, as well as for predicting the likelihood of successful stone comminution, a novel concept we refer to as stone fragility. These image-based stone biometrics will enable evidence-based identification of treatment strategies that maximize effectiveness while minimizing risk, as well as accurate and non-invasive classification of research subjects to accelerate scientific advances in the understanding and treatment of USD. We will meet these objectives by accomplishing the following specific aims: Specific Aim 1: Develop and validate CT techniques to characterize mixed and non-uric-acid stone types. Specific Aim 2: Develop and validate CT techniques to predict stone fragility. Current state-of-the-art stone imaging technology cannot accurately identify the composition of mixed and non- uric-acid stone types, nor can it provide quantitative indications of the likelihood of efficient comminution using the lowest risk technique. The innovation of this proposal lies in the use of newly developed statistical, deep learning and texture analysis techniques to quantitatively describe essential characteristics of urinary calculi, namely composition and fragility. The significance of this proposal is that the knowledge derived from using such techniques represents unique quantitative biomarkers that will allow physicians and researchers to more effectively manage and study USD. The developed methods respond to critical needs in the field of stone disease and will advance the ability of physicians to optimally direct patient therapy and scientists to phenotype research subjects.

项目摘要 症状性泌尿系结石病（USD）影响>8%的美国人口， 估计每年的医疗费用超过100亿美元。计算机断层扫描（CT）是确定的方法， 成像泌尿系结石，并可以提供准确的亚毫米级的肾结石的大小和位置的细节。 然而，体内表征不仅仅是大小和位置对于量化结石至关重要 这些特征对于优化患者健康管理非常重要，对于临床研究至关重要。一个完整 肾结石的特征，包括结石成分和脆性，需要安全和成本效益 USD的管理，以及研究对象的表型。我们的建议符合这些需要， 开发使用低剂量、多能量CT准确、无创地表征结石的方法。 我们的长期目标是使用先进的CT方法来表征尿路结石， 指导临床治疗并促进临床研究。我们在此应用程序中的目标是开发 并验证用于表征混合型和非尿酸结石类型的体内定量技术，以及 用于预测结石粉碎成功的可能性，这是我们称为结石脆性的新概念。 这些基于图像的结石生物识别技术将使治疗策略的循证识别成为可能， 最大限度地提高效率，同时最大限度地降低风险，以及准确和非侵入性的研究分类 主题，以加快科学进步的理解和治疗USD。我们将满足这些 实现以下具体目标： 具体目标1：开发和验证CT技术，以表征混合和非尿酸 石头类型 具体目标2：开发和验证CT技术以预测结石脆性。 目前最先进的石材成像技术不能准确地识别混合和非混合石材的成分。 尿酸结石类型，也不能提供使用有效粉碎的可能性的定量指标 风险最低的技术。该建议的创新之处在于使用新开发的统计、深度 学习和纹理分析技术来定量描述泌尿系结石的基本特征， 即组成和脆弱性。这一建议的意义在于， 这些技术代表了独特的定量生物标志物，将使医生和研究人员能够更多地 有效地管理和研究美元。开发的方法响应了石材领域的关键需求 疾病，并将提高医生的能力，以最佳方式指导患者治疗和科学家的表型 研究对象。

项目成果

Automated radiomic analysis of CT images to predict likelihood of spontaneous passage of symptomatic renal stones.

• DOI: 10.1007/s10140-021-01915-4
• 发表时间: 2021-08
• 期刊: Emergency radiology
• 影响因子: 2.2
• 作者: Mohammadinejad P;Ferrero A;Bartlett DJ;Khandelwal A;Marcus R;Lieske JC;Moen TR;Mara KC;Enders FT;McCollough CH;Fletcher JG
• 通讯作者: Fletcher JG

Multi-Institutional Prospective Randomized Control Trial of Novel Intracorporeal Lithotripters: ShockPulse-SE vs Trilogy Trial.

• DOI: 10.1089/end.2020.1097
• 发表时间: 2021-09
• 期刊: Journal of endourology
• 影响因子: 2.7
• 作者: Large T;Nottingham C;Brinkman E;Agarwal D;Ferrero A;Sourial M;Stern K;Rivera M;Knudsen B;Humphreys M;Krambeck A
• 通讯作者: Krambeck A