Computerized Detection of Polyps in CT Colonography

CT 结肠成像中息肉的计算机化检测

• 批准号: 7126460
• 负责人: HIROYUKI YOSHIDA
• 金额: $ 37.7万
• 依托单位: MASSACHUSETTS GENERAL HOSPITAL
• 依托单位国家: 美国
• 财政年份: 2003
• 资助国家: 美国
• 起止时间: 2003-07-03 至 2008-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7126460
• 关键词: bioimaging /biomedical imaging; clinical research; colon; colon polyp; computed axial tomography; computer assisted diagnosis; computer data analysis; diagnosis design /evaluation; early diagnosis; endoscopy; gastrointestinal imaging /visualization; health care personnel performance; human data; image enhancement; mass screening; mathematical model; neoplasm /cancer radiodiagnosis; patient oriented research

DESCRIPTION (provided by applicant): Colon cancer, the second leading cause of cancer deaths for men and women in the United States, can be prevented if precursor colonic polyps are detected and removed. The long-term goal of the proposed project is to advance the early detection of colon cancer. Computed tomographic colonography (CTC) has been proposed as a promising technique for colon cancer screening, but for CTC to be a practical screening tool, many images must be interpreted rapidly and accurately. To this end, the short-term goal of the project is to develop a high-performance computer-aided diagnosis (CAD) scheme for the automated detection of polyps in CTC to assist radiologists in detecting polyps quickly and accurately, by providing them with a "second opinion" regarding the locations of suspicious polyps. We hypothesize that a CAD scheme can significantly reduce radiologists' interpretation time and improve their performance in detecting colonic polyps in CTC. To explore this hypothesis, we propose the following specific aims: Specific Aim 1. Establish a large CTC database of polyps to develop and evaluate a CAD scheme: (1) Collect new CTC cases of polyps retrospectively and prospectively. Specific Aim 2. Develop methods for the detection of polyp candidates: (1) Generate isotropic volumetric data from axial CT images; (2) Develop methods for automated extraction of the colon based on apriori knowledge of the abdominal anatomy; (3) Develop methods for extraction of polyp candidates based on geometric features. Specific Aim 3. Develop methods for the reduction of false positives: (1) Develop methods of clustering of polyp candidates to merge polyp candidates and to remove false positives due to noise; (2) Develop methods based on 3-dimensional volumetric features that differentiate polyps from false positives due to normal anatomic structures; (3) Use diseriminant analysis, artificial neural networks, and genetic algorithms to merge volumetric features for the reduction of false positives. Specific Aim 4. Evaluate the performance and benefit of the overall CAD scheme: (1) Evaluate the performance of the CAD scheme in the detection of polyps in CTC. (2) Evaluate the benefit of the CAD scheme in reducing radiologists' interpretation time and improving diagnostic performance in the detection of polyps by means of an observer study. Successful development of such a CAD scheme will advance the clinical implementation of CT-based colon cancer screening, promote early diagnosis of colon cancer, and ultimately reduce mortality due to colon cancer.

描述（由申请人提供）：结肠癌是美国男性和女性癌症死亡的第二大原因，如果检测并切除前体结肠息肉，则可以预防结肠癌。该项目的长期目标是促进结肠癌的早期检测。计算机断层结肠成像 (CTC) 被认为是一种很有前途的结肠癌筛查技术，但要使 CTC 成为实用的筛查工具，必须快速、准确地解读许多图像。为此，该项目的短期目标是开发一种用于CTC息肉自动检测的高性能计算机辅助诊断（CAD）方案，协助放射科医生快速、准确地检测息肉，为放射科医生提供关于可疑息肉位置的“第二意见”。我们假设 CAD 方案可以显着减少放射科医生的解释时间并提高他们在 CTC 中检测结肠息肉的性能。为了探讨这一假设，我们提出以下具体目标： 具体目标 1. 建立大型息肉 CTC 数据库，以开发和评估 CAD 方案： (1) 回顾性和前瞻性收集新的息肉 CTC 病例。具体目标 2. 开发息肉候选者的检测方法： (1) 从轴向 CT 图像生成各向同性体积数据； (2) 基于腹部解剖学的先验知识开发自动提取结肠的方法； (3)开发基于几何特征的息肉候选提取方法。具体目标 3. 开发减少误报的方法： (1) 开发息肉候选者的聚类方法，以合并息肉候选者并消除由于噪声导致的误报； (2) 开发基于 3 维体积特征的方法，将息肉与正常解剖结构导致的误报区分开来； (3)使用判别分析、人工神经网络和遗传算法来合并体积特征以减少误报。具体目标 4. 评估整个 CAD 方案的性能和效益： (1) 评估 CAD 方案在 CTC 息肉检测中的性能。 (2) 通过观察者研究评估 CAD 方案在减少放射科医师解释时间和提高息肉检测诊断性能方面的益处。该CAD方案的成功开发将推动基于CT的结肠癌筛查的临床实施，促进结肠癌的早期诊断，最终降低结肠癌死亡率。