Artifact-Free Reconstruction of Medical Imaging Information

描述（由申请人提供）：前列腺癌是继皮肤癌之后最常见的癌症，也是美国男性癌症死亡的第二大原因。前列腺癌放射治疗计划的选择受到CT软组织收缩低、MRI前列腺形状扭曲、超声散斑噪声和衰减引起的成像伪影的限制。在过去的几十年里，稳健和自动的前列腺分割只取得了有限的成功，并且仍然是一项具有挑战性的任务。本课题提出了一种新的医学影像信息重建技术。对病理有用和解剖正确的后向散射和衰减场可以重建超声图像，以及自动结构分割。超声优异的软组织收缩性和便携性使其成为准确确定前列腺实际边界的一种有前景的方式，并将其整合到前列腺癌的治疗计划中，在给予足够的靶剂量的同时最大限度地减少对周围正常组织的辐射。提出了四个具体目标。在目标1中，将开发和优化基于变分原理的方法和算法，以补偿衰减伪影，并自动分割医学超声图像中的解剖结构。在目标2中，将开发图像采集协议和系统软件，以构建三维超声体并精确地在空间上配准超声图像。在目标3中，开发的系统的准确性将在边界分割，局部衰减伪影校正和空间校准中进行量化和验证。在目标4中，该方法将通过涉及前列腺癌患者进行临床和定性评估。将根据提出的方法设计治疗方案，并比较前列腺、膀胱和直肠的剂量覆盖率。将评估使用所建议的方法减少规划余量的意义。本项目利用经腹超声成像在前列腺癌治疗计划中的潜力，比目前在日常验证中实现的潜力更大。所提出的方法将改善衰减伪影校正，揭示隐藏/额外的临床重要信息，自动描绘解剖结构，提高癌症治疗的准确性，并降低正常组织的毒性。

DESCRIPTION (provided by applicant): Prostate cancer is the most common cancer after skin cancer and the second leading cause of cancer death in men in the United States. The options for radiotherapy treatment planning of prostate cancer are limited by CT's low soft tissue contract, MRI's distortion of prostate shape, and ultrasound's speckle noise and attenuation-induced imaging artifacts. Robust and automatic prostate segmentation has only achieved limited success in the past decades and remains a challenging task. A novel technique for reconstruction of medical imaging information is proposed in this project. Pairs of pathological useful and anatomically correct backscatter and attenuation fields can be reconstructed from ultrasound images, along with automatic structure segmentation. The excellent soft tissue contract and portability of ultrasound make it a promising modality for accurate determination of the actual prostate boundary, and to be integrated into prostate cancer treatment planning to give adequate target dose while minimizing radiation to surrounding normal tissues. Four specific aims are proposed. In aim 1, the method and algorithm based on variational principle will be developed and optimized to compensate for attenuation artifacts, and automatically segment anatomic structures in medical ultrasound images. In aim 2, image acquisition protocol and system software will be developed to construct 3D ultrasound volumes and accurately register ultrasound images spatially. In aim 3, the accuracy of the developed system will be quantified and verified in boundary segmentation, localized attenuation artifact correction, and spatial calibration. In aim 4, the approach will be evaluated clinically and qualitatively by involving prostate cancer patients. Treatment plans will be designed based on the proposed method, and compared on dose coverage in the prostate, bladder and rectum. The significance of reduction in planning margin using the proposed method will be evaluated. This project exploits a greater potential of trans-abdominal ultrasound imaging in prostate cancer treatment planning than is currently being realized in daily verification. The proposed method will improve attenuation artifact correction, reveal hidden/additional clinic-important information, automatically delineate anatomic structures, increase cancer treatment accuracy, and reduce normal tissue toxicity. PUBLIC HEALTH RELEVANCE: Robust and automatic prostate segmentation has only achieved limited success in the past decades and remains a challenging task. This project exploits a greater potential of trans-abdominal ultrasound imaging in prostate cancer treatment planning than is currently being realized in daily verification. The proposed method will improve attenuation artifact correction, reveal hidden/additional clinic-important information, automatically delineate anatomic structures, increase cancer treatment accuracy, and reduce normal tissue toxicity.