Background

Various distance as well as angle measurements of facial structures are evidence of proper ocular plastic and eyelid disease surgery plans. Surgeons nowadays make decisions based on so called experience, which is non-quantitative, resulting in low dissemination efficiency.

Purpose

To establish a method for eyelid parameter automatic quantification without mega data labeling and training, facilitating the popularization of eyelid related surgery design.

Methods

An automatic measurement of eyelid parameters is described. It is composed of obtaining of absolute distance between left inner canthus and right inner canthus by 3D models of human faces, calculation of 22 eyelid parameters based on face landmarks, converting 18 distance-related eyelid parameters’ values which correspond to relative distance to real distance.

Results

An auto-measurement of 80 normal patients and 60 TAO patients is conducted. By comparison to manual-measurement, we achieved relative errors less than 10% and 13% for regular eyes and diseased eyes.

Conclusions

Eyelid related characteristics are quantified in an easy, portable and precise way. Based on which, series of approaches in quantitative evaluation, operation selection and extensive dissemination may develop in the fields of eyelid disease and ocular plastics.

中文翻译：

---

一种基于3D扫描的眼睑参数自动测量方法

背景

面部结构的各种距离和角度测量是正确的眼部整形和眼睑疾病手术计划的证据。现在的外科医生根据所谓的经验做出决定，这是非量化的，导致传播效率低下。

目的

建立无需大数据标注和训练的眼睑参数自动量化方法，促进眼睑相关手术设计的普及。

方法

描述了眼睑参数的自动测量。它包括通过人脸3D模型获取左内眦与右内眦之间的绝对距离，基于人脸地标计算22个眼睑参数，将18个与距离相关的眼睑参数值转换为相对距离对应的真实距离.

结果

对 80 名正常患者和 60 名 TAO 患者进行自动测量。与手动测量相比，我们实现了普通眼和患病眼的相对误差分别小于 10% 和 13%。

结论

眼睑相关特征以简单、便携和精确的方式量化。在此基础上，在眼睑疾病和眼部整形领域可能会发展出一系列定量评估、手术选择和广泛传播的方法。