Ground truth generation toolset for microscopic human embryo images

在过去的40年里，由于推迟生育，生育治疗的使用急剧增加。不幸的是，女性的生殖能力在出生后的第二年和第三十年达到顶峰，因此到40岁时生育力显著下降，流产的几率也很高。辅助生殖技术包括体外处理卵子和胚胎的生育治疗。体外受精(IVF)是一种最常见的生育治疗方法，它通过超刺激卵巢产生多个卵子进行体外受精。受精卵(胚胎)孵化到囊胚期(受精后约5天)，然后选择植入。在孵化过程中，根据固有的形态结构，通过分配等级来反映胚泡的质量，从而对最佳候选进行评估。能够量化为什么某些胚泡具有较高的植入潜力将有助于研究人员提高成功率，并将因移植多个胚胎而生多胞胎的可能性降至最低。因此，可靠、准确地识别和测量各种胚胎成分和发育参数是非常必要的。这项研究项目的主要重点是创造半自动技术，帮助和便利专家胚胎学家快速识别和标记感兴趣的区域。更具体地说，我们建议开发一个图像处理工具集，该工具集可以自动处理一些较为繁琐的地面真相识别部分，并允许Expert以最小的人工干预/努力改进对人类胚胎的2D显微图像的结果。

The use of fertility treatments has increased dramatically in the last 4 decades as a result of delayedchildbearing. Unfortunately, female reproductive capacity declines from a peak in the 2nd and 3rd decades oflife so that by the age of 40 there is a significant reduction in fertility with a high chance of miscarriage.Assisted Reproductive Technology includes fertility treatments that involve handling eggs and embryos outsidethe body. In Vitro Fertilization (IVF) is one of the most common fertility treatments in which ovaries are hyperstimulated to produce multiple eggs for external fertilization. The fertilized eggs (embryos) are incubated untilthey reach the blastocyst stage (about 5 days after fertilization) and then are chosen for implantation. Optimalcandidates are evaluated during the incubation process by assigning grades to reflect the quality of blastocystsbased on intrinsic morphological structures. Being able to quantify why certain blastocysts have higherimplantation potentials will help researchers to increase success rates and minimize the chances of multiplebirths due to transferring multiple embryos. Therefore, it is essential to identify and measure various embryocomponents and developmental parameters in a reliable and accurate manner. The main focus of this researchproject is to create semi-automated technologies that can assist and facilitate expert embryologists to identifyand mark regions of interest in a high pace. More specifically, we propose developing an image processingtoolset that can automatically process some of the more tedious parts of ground truth identification and allowexpert to improve results with the minimal manual intervention/effort in 2D microscopic images of humanembryos.