LaVision双光子显微镜-多线扫描双光子成像（三）
2.2.多线TPLSM中通过成像检测释放光
在单光束TPLSM中，光电倍增管PMT或者雪崩二极管APD可以很方便地用于释放光检测，由于双光子激发的原理，激发只发生在激光焦点处。因此，用于屏蔽离焦光线的共焦小孔变得不必要，并且可以使用NDD检测。这意味着激发光不会被送回扫描镜，而是直接进入位于靠近物镜处的具有大的活性面积的PMT。但是，这种简单检测模式在多线PTLSM中却很难实现，因为必须要分离出每一个激光焦点发出的荧光。因此，我们首先设想了一种使用多APD的检测模式在非扫描模式（descanned mode）下操作。这种检测模式的灵敏性是阳性的，在仅用一个激光束和单个APD的预实验中(Kalb et al., 2004).但是我们抑制了安装额外多个APD用于多线检测的想法，因为检测器的调整非常关键。取而代之的是，受到芯片信号放大的超灵敏度CCD（所谓电子增强CCD）的启示，我们决定在non-descanned mode下使用成像设备。在这种模式下，使用同步CCD读出光束一次或多次扫过样品来获取整幅图像。这相对于使用PMT或APD的点扫描检测，它们在每个激光焦点位置取一个值，图像通过将这些值与激光的位置进行关联来重构。在散射性组织中，使用成像检测取代点扫描检测在一定程度上不可避免地降低了空间分辨率。如我们所示，多线TPLSM也可以进行点扫描检测，但只能以远低于成像检测的时间分辨率进行（见本段后面内容和section3对两中检测模式进行关键评估）。
Fig. 2. Schematic of multiline TPLSM. Top: principle of the mirror-based beam multiplexer as first described by Nielsen et al. (2001). A linear array of in this case eight laser beams is generated by repeated separation at a 50%-beamsplitter mirror (BS) and reflection at high-reflectivity mirrors (HR). Bottom: schematic of the components in the setup for multiline TPLSM. The laser beam of a tunable Ti:sapphire laser passes a shutter, an attenuator, a beam telescope to adjust the size of the beam, and a dispersion compensation consisting of a pair of prisms. The latter prevents elongation of laser pulses by compensating for the group velocity dispersion introduced by the glass in the optical path of the laser beam. The laser beam then enters the beam multiplexer which divides the laser beam into 64 beams. In the commercially available version of the beam multiplexer (LaVision BioTec, Bielefeld) used now in our setup the maximal number of beams can be reduced by software-controlled replacement of part of the beamsplitter mirror with a high-reflectivity mirror. However, in order to generate a beam array with widely separated laser foci (as used for the recording shown in Fig. 3B), manual blocking of some of the beams had to be used. Scan movements of the beam array are controlled by nonresonant xy-galvo-scan mirrors. Before entering the microscope (Olympus IX70), a second beam telescope is passed. The microscope long-distance water-immersion 40×/N.A. 0.8 objective was mounted on a z-piezo to control the focal plane for the acquisition of image stacks. The fly is fixed with bees wax at a small glass plate and mounted on the microscope stage to control its xy-position. Moving patterns are presented in the visual field of the fly at high contrast and fast update rate with a custom-built LED board. The microscope was not equipped with conventional epifluorescence illumination. However, delivery of excitation light (mercury lamp band-pass filtered at 420–480 nm) to the sample via a lightguide (not shown in the diagram) turned out more practical than TPLSM to localize the dye-filled neuron at the beginning of an experiment. In the first case, a 465 nm beamsplitter and a 515 nm long-pass emission filter was used. In the case of TPLSM a 680 nm beamsplitter in combination with a TP-emission filter (short pass 700 nm) was used. An EMCCD camera (Andor Ixon DV887BI) was installed for the detection of emission light.
由于技术限制，直到现在，TPLSM中的成像检测主要限制在要么提供高水平的发射光，要么不要求高的时间分辨率(see e.g. Bewersdorf et al., 1998). 为了获取更高的可能灵敏度（进而到时间分辨率），即使在正常动物原位成像过程中的暗淡激发光情况下，我们使用了一个背景照明CCD，在500-650nm处量子效率>90%（例如，在一个包含了大多数钙离子染料的激发峰的范围内）和高达1000的可变电子倍增增益因子(Andor iXon DV887BI)(Coates et al., 2004).
