PFI: AIR-TT: Prototype Scale-up for Traumatic Pelvic and Abdominal Injury Decision Support System (DSS)

这个PFI:AIR技术翻译项目的重点是翻译和加速专利的创伤性骨盆和腹部损伤决策支持系统（DSS）技术的商业化。这将使医生能够快速准确地从所有相关的生物医学图像中提取复杂的患者数据，（例如，具有数百个图像切片的CT扫描）每位患者的创伤评分、诊断、治疗、人口统计和损伤细节。同时整合和分析信息，在患者护理的每个阶段生成预测、警告和治疗建议。该项目不仅有助于减少医疗并发症和提高生存率，而且对优化资源利用也很重要——这是每年减少大约600亿美元治疗骨盆和腹部创伤病例并发症的医疗费用的关键。与现有的DSS工具相比，创伤性骨盆和腹部损伤DSS技术具有以下独特的功能，提供了竞争优势：1)它分割和评估主要腹部器官的损伤；2)为一些具体的、复杂的临床决策提供建议和预测；3)它是全自动的，不需要专家？提供易于使用的软件界面，并可能为创伤患者护理提供更高准确性的相关建议。如果算法和软件通过该项目成功验证，则已经确定了将DSS软件商业化的许可途径。当一个人试图在紧急情况下快速准确地整合复杂类型的患者数据时，临床决策显示出其真正的复杂性。该项目解决了从研究发现到商业应用的几个技术差距。现有的“半自动化”系统只使用部分患者数据，不分析数字图像中包含的详细信息来创建建议；相反，目前主要用于辅助ct（或其他图像）分析的图像处理技术并没有优化到满足创伤和/或DSS工具的需求。该项目将完善和扩大原型，验证其临床应用，并加速其商业化，以帮助临床医生治疗创伤性骨盆和腹部损伤病例。关键技术目标是：1)将器官分割软件模块（目前仅覆盖脾脏）扩展到包括肝脏、肾脏和胰腺；2)改进出血检测算法，发现靠近骨骼的出血；3)使用更大、更全面的数据集进一步验证和改进系统；4)重新编写图形用户界面，以匹配原型的要求，并验证其有效性和临床医生的易用性。本项目生成的关键计算方法包括自动图像处理算法和机器学习方法，以：1)评估骨折和出血的CT扫描并测量其大小；2)多节段主要脏器，识别损伤，定量评估损伤程度；3)预测结果（生存，ICU天数，家庭与康复治疗等），并在治疗的每个步骤为护理人员提供建议。参与该项目的研究生将通过开发DDS工具，测试和验证算法，以及与项目团队，临床医生，业务开发人员，技术转让专业人员和潜在的许可人密切合作，将技术作为可行产品商业化，从而获得创新和技术转化为商业化的经验。

This PFI:AIR Technology Translation project focuses on translating and accelerating commercialization of a patented Traumatic Pelvic and Abdominal Injury Decision Support System (DSS) technology. This will enable physicians to quickly and accurately extract complex patient data from all relevant biomedical images, (e.g., CT scans with hundreds of image slices) trauma scores, diagnoses, treatments, demographics, and injury specifics for each patient?while integrating and analyzing the information to generate prediction, warning, and treatment recommendations at every stage of patient care. This project is important not only to help decrease medical complications and increase survival, but also to optimize resource utilization- a key to reducing the approximately $60B medical cost each year for treating complications in pelvic and abdominal trauma cases. The Traumatic Pelvic and Abdominal Injury DSS technology has the following unique capabilities which provide competitive advantages when compared to the existing state of the art for DSS tools: 1) it segments and assesses damage to major abdominal organs; 2) it provides recommendations and predictions for several specific, complex clinical decisions; 3) it is fully automated and does not require an expert?s supervision in analyzing patient data- providing an easier-to-use software interface and potentially providing higher accuracy in pertinent recommendations for trauma patient care. If the algorithms and software are successfully validated through this project, a licensing pathway has been identified to commercialize the DSS software.Clinical decision making shows its true complexity when one is trying to quickly and accurately integrate complex types of patient data in an emergency setting. This project addresses several technology gaps as it translates from research discovery toward commercial application. Existing "semi-automated" systems use only a portion of patient data and do not analyze detailed information contained in digital images to create recommendations; conversely, current image processing technologies designed mainly to assist in analysis of CTs (or other images) are not optimized to address the needs of trauma and/or DSS tools. This project will refine and scale up the prototype, validate its clinical use, and accelerate its commercialization to assist clinicians in traumatic pelvic and abdominal injury cases. Key technical objectives are to: 1) expand the organ segmentation software module (now covering only the spleen) to include the liver, kidneys, and pancreas; 2) enhance the hemorrhage detection algorithms to find bleeding close to bones; 3) further validate and improve the system using a larger and more comprehensive dataset; and 4) rewrite the graphical user interface to match requirements for the prototype and validate its effectiveness and ease of use by clinicians. Key computational methods generated by this project include automated image processing algorithms and machine learning methods to: 1) assess a CT scan for bone fracture(s) and hemorrhage and measure their sizes; 2) segment more major organs, identify damage, and quantitatively assess level of injury; and 3) predict outcomes (survival, number of ICU days, home vs. rehab, etc.) and form recommendations for care givers at each step of the treatment. The graduate student involved in this project will gain experience in innovation and technology translation towards commercialization through development of the DDS tool, testing and validating the algorithms, and working closing with the project team, clinicians, business developers, tech transfer professionals, and a potential licensee to commercialize the technology as a viable product.