Temperature controlled retinal laser treatment

视网膜激光光凝已成为治疗各种眼病的标准工具。主要的想法是使用强激光脉冲导致温度升高，最后在视网膜组织凝固。例如，光凝用于防止视网膜孔洞形成后脱离，或用于减少视网膜周围高耗氧光感受器的数量，以保护糖尿病患者的黄斑功能。在治疗糖尿病黄斑水肿的过程中，既要防止凝血过强对神经视网膜造成不可逆的损伤，又要避免剂量过少，保证达到预期的治疗效果。光凝的一个主要挑战是调整适当的光剂量，这是由于视网膜上强烈的吸收变化，改变眼睛内的光散射和小的不自主眼球运动（微眼跳）。在目前的临床实践中，治疗医师在典型的50-200 ms照射时间后，根据先前病变的可见性来选择后续病变的激光功率。然而，这是一个非常繁琐和耗时的过程，往往导致相当不满意的结果。因此，基于实时测量对预期温升（感应热）的激光功率进行非常精确的控制是非常重要的。在我们之前的工作中，基于光声效应开发了第一个也是迄今为止唯一的实时视网膜温度测定方法。本课题的主要目标是基于该方法开发闭环自动控制策略，以保证眼科医生预定的所需治疗温度的准确实现。特别是，我们计划开发一种新的实验设置，摆脱了以前使用的治疗和探测激光的组合，只使用一个高重复率激光，同时激发光声压波并产生足够的热用于凝固。在算法方面，我们将(i)开发适合控制器设计的不同粒度的模型，（ii）设计观察/估计策略，以从可用的测量（压力）中重建所需的数量（特别是视网膜温度分布），以及（iii）开发和评估不同复杂性的合适控制策略，以便自动调整所需的激光功率。关于后一点，我们将特别考虑不同的模型预测控制方法，允许约束满足的硬保证，这在医学应用中是不可或缺的。总之，该项目将为视网膜激光治疗的显著改进和安全应用迈出关键一步，并进一步有助于研究尚不完全了解的组织短脉冲热损伤。

Retinal laser photocoagulation has become a standard tool for the treatment of various eye diseases. The main idea is to use intensive laser pulses resulting in a temperature increase and finally in coagulation of the retinal tissue. Photocoagulation is e.g. used to prevent retinal detachment after hole formation, or to reduce the number of highly oxygen consuming photoreceptors in the retinal periphery to preserve the macular function in diabetic patients. Mostly, minimally invasive uniform lesions are demanded: For the treatment of diabetic macular oedema irreversible damages of the neural retina by too strong coagulation must be prevented while guaranteeing the desired therapeutic effect by avoiding underdose.A major challenge in photocoagulation is the adjustment of a proper light dosing owing to strong absorption variations across the retina, changing light scattering within the eye and small involuntary eye movements (microsaccades). In current clinical practice, the treating physician chooses the laser power for the subsequent lesions according to the visibility of the previous lesions after the typical 50-200 ms irradiation time. However, this is a very cumbersome and time consuming procedure and often leads to quite unsatisfactory results. Hence, a very accurate control of the laser power for the intended temperature rise (induced heat) based on real-time measurements is of great importance.In our previous work a first and so far only method for real-time retinal temperature determination was developed based on the optoacoustic effect. The main goal of the present proposal is to develop closed-loop automatic control strategies based on this methodology, which can guarantee an accurate realization of the desired treatment temperature predetermined by the ophthalmologist. In particular, we plan to develop a new experimental setting getting rid of the previously used combination of treatment and probe lasers by using only one high repetition rate laser, which simultaneously serves to excite optoacoustic pressure waves and produces sufficient heat for coagulation. On the algorithmic side, we will (i) develop models of different granularity suitable for controller design, (ii) design observation/estimation strategies to reconstruct the desired quantities (in particular the retinal temperature distribution) from the available measurements (pressure), and (iii) develop and evaluate suitable control strategies of different complexity in order to automatically adjust the required laser power. With respect to the latter point, we will in particular consider different model predictive control approaches allowing for hard guarantees of constraint satisfaction, which is indispensable in medical applications.In conclusion, the proposed project will constitute a crucial step towards significantly improved and safe applications of retinal laser therapy and further helps to investigate short pulsed thermal damage of tissue not fully understood yet.