Image Analysis Tools for mpMRI Prostate Cancer Diagnosis Using PI-RADS

使用 PI-RADS 进行 mpMRI 前列腺癌诊断的图像分析工具

• 批准号: 10256757
• 负责人: John Aaron Onofrey
• 金额: $ 80.31万
• 依托单位: EIGEN HEALTH SERVICES LLC
• 依托单位国家: 美国
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-05-01 至 2023-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10256757
• 关键词: Accounting; Address; Agreement; Algorithms; Atlases; Benign; Biopsy; Cancer Diagnostics; Cancer Etiology; Categories; Cessation of life; Classification; Clinical; Collaborations; Communication; Computer Assisted; Consumption; Data; Data Collection; Databases; Diagnosis; Diagnostic; Diffusion; Diffusion Magnetic Resonance Imaging; Environment; Goals; Image; Image Analysis; Image Enhancement; Information Systems; Interobserver Variability; Intraobserver Variability; Label; Lesion; Letters; Localized Lesion; Magnetic Resonance Imaging; Malignant - descriptor; Malignant Neoplasms; Malignant neoplasm of prostate; Manuals; Maps; Meta-Analysis; Methodology; Methods; Modeling; Monitor; Multimodal Imaging; Performance; Phase; Procedures; Process; Prostate; Protocols documentation; Psyche structure; Radiologic Finding; Radiology Specialty; Reader; Reading; Reporting; Reproducibility; Research; Site; Source; Specific qualifier value; Specificity; Staging; Standardization; System; T2 weighted imaging; Time; Training; Validation; Variant; Work; Workload; base; cancer diagnosis; clinical decision support; clinical diagnostics; clinical imaging; clinical practice; clinical research site; commercialization; contrast enhanced; deep learning; design; disease heterogeneity; image guided; image registration; imaging biomarker; improved; innovation; machine learning algorithm; machine learning method; medical specialties; men; mortality; multimodality; novel; prostate biopsy; radiological imaging; radiologist; research clinical testing; support tools; tool

Project Summary Prostate cancer is one of the most commonly occurring forms of cancer, accounting for 21% of all cancer in men. The Prostate Imaging Reporting and Data System (PI-RADS) aims to standardize reporting of prostate cancer using multi-parametric magnetic resonance imaging (mpMRI). However, the in-depth analysis, as demanded by PI-RADS, remains challenging due to the complexity and heterogeneity of the disease, and it is a clinically burdensome task subject to both significant intra- and inter-reader variability. Auxiliary tools based on machine learning methods such as deep learning can reduce diagnostic variability and increase workload efficiency by automatically performing tasks and presenting results to a radiologist for the purpose of decision support. In particular, automated identification and classification of lesion candidates using imaging data can be performed with respect to PI-RADS scoring. In Phase I of this project, we developed two automated methods to reduce the intra- and inter-observer variability while interpreting mpMRI images using the PI-RADS protocol: (i) a method to co-register mpMRI data, and (ii) a method to geometrically segment the prostate gland into the PI-RADS protocol sector map. The overarching goal of this Phase II project is to develop machine learning algorithms that incorporate both co-registered multi-modal imaging biomarkers and PI-RADS sector map information into an automated clinical diagnostic aid. The innovation in this project lies in the use of deep learning to automatically predict PI-RADS classification. This project is significant in that it has the potential to improve clinical efficiency and reduce diagnostic variation in prostate cancer diagnosis. In Aim 1 of this project, we will develop a deep learning approach to localize and classify lesions in mpMRI. In Aim 2, we will integrate this diagnostic tool into the ProFuseCAD system and perform rigorous multi-site validation to quantify PI-RADS classification performance. Both aims will utilize a database of over 1,000 existing mpMRI images from multiple clinical sites to develop and validate the algorithms. Ultimately, enhancements from this project will create a novel feature for Eigen's (the applicant company's) FDA 510(k)-cleared imaging product, ProFuseCAD, in order to improve the diagnosis and reporting of prostate cancer.

项目摘要 前列腺癌是最常见的癌症之一，占男性所有癌症的21%。 前列腺成像报告和数据系统（PI-RADS）旨在规范前列腺癌的报告 使用多参数磁共振成像（mpMRI）。然而，根据要求， 由PI-RADS，仍然具有挑战性，由于疾病的复杂性和异质性，它是一个临床 繁重的任务受到读者内部和读者之间的显著差异的影响。基于机器的辅助工具 深度学习等学习方法可以通过以下方式减少诊断变异性并提高工作负载效率： 自动执行任务并将结果呈现给放射科医师以用于决策支持的目的。在 特别地，可以使用成像数据执行病变候选者的自动识别和分类 关于PI-RADS评分。在该项目的第一阶段，我们开发了两种自动化方法来减少 使用PI-RADS协议解释mpMRI图像时观察者内和观察者间的变异性：（i）一种方法 共配准mpMRI数据，以及（ii）将前列腺几何分割成PI-RADS的方法 协议扇区映射。这个第二阶段项目的首要目标是开发机器学习算法， 将共同配准多模态成像生物标记和PI-RADS扇区映射信息结合到 自动化临床诊断辅助设备。该项目的创新在于使用深度学习自动 预测PI-RADS分类。该项目的重要性在于它有可能提高临床效率 并减少前列腺癌诊断中的诊断变异。在本项目的目标1中，我们将开发一个深入的 学习mpMRI中病灶定位和分类的方法。在目标2中，我们将把这个诊断工具集成到 ProFuseCAD系统，并执行严格的多站点验证，以量化PI-RADS分类性能。 这两个目标都将利用来自多个临床研究中心的1,000多个现有mpMRI图像的数据库来开发和 验证算法。最终，该项目的增强功能将为Eigen的（ 申请人公司）FDA 510（k）批准的成像产品ProFuseCAD，以改善诊断和 前列腺癌的治疗方法