Computer-based screening of the placenta to identify postnatal women likely to develop cardiovascular disease in their lifetime

胎盘是妊娠的关键器官，负责母亲和胎儿之间所有气体和营养物质的交换。当胎盘的发育或功能受损时，可能会出现常见的妊娠疾病（例如，妊娠期高血压、胎儿生长受限）。现在已知，这些疾病是由胎盘发育不良引起的，是日后发生心血管疾病（CVD）的重要危险因素。CVD是女性死亡的主要原因。因此，几家医院为患有胎盘疾病的怀孕妇女开设了专门的后续诊所，在那里对她们的心血管健康进行筛查，并提供预防方案。这些诊所是治疗女性心血管疾病的一种很好的方法，但是，它们运行起来非常昂贵，并且可能对所有患有胎盘疾病的女性都是不必要的，因为约50%的女性被转介到这些诊所，后来被认为是心血管疾病的低风险。我们以前的工作已经证明，在显微镜下可以看到的胎盘中的特殊类型的损伤与随访时的CVD风险密切相关。不幸的是，由专业病理学家进行的视觉筛查是不切实际的，因为它成本高昂，而且在较小的医院和医疗机构中不可用。目前的研究提案将解决这一挑战，使用机器学习方法使计算机能够学习如何自动检测具有高CVD风险的女性胎盘图像中的这些视觉模式。如果成功的话，开发的技术将有助于简化高优先级妇女的识别和转介，以具有成本效益和及时的方式进行心血管筛查和干预计划，最终改善这些妇女的心血管健康。

The placenta is a critical organ of pregnancy, responsible for the exchange of all gases andnutrients between the mother and fetus. When the development or function of the placentabecomes compromised, common diseases of pregnancy can arise (e.g., hypertension inpregnancy, fetal growth restriction). It is now known that these diseases, which result from adamaged placenta, are significant risk factors for the development of cardiovascular disease(CVD) later in life. CVD is a leading cause of death for women. As such, several hospitalsnow run specialized follow up clinics for women following a pregnancy with placental disease,where their cardiovascular health is screened and prevention programs are offered. Theseclinics are a wonderful approach to tackle CVD in women, however, they are very expensiveto run and are probably unnecessary for all women with placental disease, as ~50% ofwomen referred to these clinics are later deemed to be at low risk for CVD. Our previous workhas demonstrated that particular types of damage in the placenta, which can be seen under amicroscope, are strongly associated with CVD risk at follow up. Unfortunately, visualscreening by specialized pathologists is impractical as it is costly and unavailable in smallerhospitals and health care settings. The current research proposal will tackle this challenge,using machine learning methods to enable a computer to learn how to automatically detectthese visual patterns in images of placentas from women who have high CVD risk. Ifsuccessful, the developed technology will help streamline the identification and referral ofhigh-priority women to cardiovascular screening and intervention programs in a cost- effectiveand timely fashion, ultimately improving the cardiovascular health of these women.