Fig.1 LD-2P-FLIM enables large FOV, high-resolution two-photon imaging by AO correction using off-the-shelf optics. (A) Segment-scanning sensorless AO strategy for breaking the nominal FOV of objectives. (B) Full FOV comparison between AO off and AO on by imaging of a structured fluorescent sample with 10 lines per mm. The image shows 30 lines, which suggests that the FOV diameter can be extended to 3 mm. (C) Dendritic spines can be resolved through in vivo imaging of Thy1-GFP-M mice at nominal and extended FOVs. (D) A 3 × 3 mm2 FOV can encompass multiple functional cortical areas in the mouse brain. (E) Full-FOV (3 × 3 mm2) imaging of GCaMP6s-labeled neurons in vivo, encompassing the left and right hemispheres of the mouse brain with a clearly visualized sagittal sinus (red arrow). (F) Zoomed-in images indicated by the boxes in E from the nominal and extended FOVs. (G) Zoomed-in images indicated by the dashed boxes in F. (H) Calcium dynamics of the imaged neurons in E, suggesting that more than 3,000 neurons can be detected simultaneously. (I) The corresponding inferred spikes in G.#

针对大视野扫描带来的时间分辨率下降问题，研究团队设计了灵活的双区域时间复用成像架构，如图2所示。利用偏振光学器件将单束脉冲激光拆分为具有6.25 ns时间延迟的双束光，分别接入两套独立的扫描引擎，在大视野内任意选择两个目标区域进行平行扫描。无论是横向相距800 μm的两个独立脑区（如初级运动皮层M1与初级躯体感觉皮层S1），还是轴向相距80 μm的不同皮层深度（如S1的第1层与第2/3层），系统均能实现无缝同步记录。在足底电刺激实验中，双区域同步成像清晰展示了M1与S1神经元几乎同步的钙响应，而在轴向双平面成像中则捕捉到了感觉信号从L1向L2/3传导的时间延迟。

Fig.2 LD-2P-FLIM enables simultaneous and flexible dual-region imaging of neural activities with subcellular resolution over a large FOV. (A) Optical schematic of the LD-2P-FLIM system via TM. By designing two independent SEs, consisting of a DM, a resonant scanner, and a pair of galvanometer mirrors, simultaneous dual-region imaging can be performed in two arbitrarily selected regions (settable location and size). (B) Simultaneous calcium imaging from two lateral regions in M1 and S1 under ES. (C) Representative examples of stimuli-related neural activities from B. Under periodic stimuli, the neural activity clearly (>3 ΔF/F) demonstrated a periodic response in M1 and S1. (D) Simultaneous calcium imaging from dual planes in S1 L1 and L2/3 under ES. (E) Representative examples of stimuli-related neural activities from D. In contrast to the results of M1 and S1, the calcium responses were significantly delayed in S1 L2/3 compared with those in S1 L1 under the same stimuli. PBS: polarizing beam splitter, FPGA: field programmable gate arrays.#

在荧光寿命成像速度方面，定制的FPGA解复用模块实现了超高吞吐量的实时处理，如图3所示。系统以3.2 GS/s的高采样率对PMT探测到的混合荧光信号进行同步采集，并利用激光脉冲的时间相位精确分离出两个区域的光子衰减曲线。该模块支持双区域同时以30 Hz（512 × 512像素）的高帧率输出荧光寿命图像，像素吞吐量达到15.73 Mpixels/s，比传统TCSPC提升了200倍以上，寿命测量误差小于0.1 ns。借助高吞吐能力，研究团队利用单激光（920 nm）同时激发荧光寿命差异显著的三种绿色荧光蛋白，实现了高激发效率的多参数成像，并成功在轴向双平面上实时追踪了光损伤诱发单个神经元凋亡后，小胶质细胞突起向损伤位点快速迁移的动态过程。

Fig.3 LD-2P-FLIM provides fast and high-throughput two-region FLIM. (A) Schematic of the developed FPGA module, which can demultiplex the fluorescence signals from two regions and achieve fast FL imaging. (B) Decay curves obtained by imaging a FITC solution between LD-2P-FLIM and conventional TCSPC. (C) Comparison of the measured FLs between LD-2P-FLIM and conventional TCSPC. The results suggest that LD-2P-FLIM can provide quantitative lifetime results with an error of less than 0.1 ns. (D) Lifetime measurement for EGFP, BrUSLEE, and NowGFP in vitro, verifying the obvious lifetime difference among the three green fluorescent labels. (E) Schematic of the imaging of the response of microglia to local neuron injury caused by laser ablation. (F) Time-lapse FL images from the axial dual plane, suggesting that microglial processes move rapidly toward the site of the injured neuron induced by the two-photon laser. (G) Zoomed-in images from the damaged plane indicated by the solid and dashed red boxes in F. (H) Detailed kinetics of the microglial response over 1 h within the dual planes.#

高吞吐量的荧光寿命成像还解决了神经血管偶联（NVC）过程中多色标记激发效率低的问题，如图4所示。通过使用同为绿色荧光的GCaMP6s标记神经元和FITC标记血管，利用两者显著的寿命差异（2.63 ns与4.52 ns）进行无串扰信号分离。在周期性电刺激下，双区域FLIM实时记录到了神经元激活后伴随的局部微血管扩张过程。在 kainic acid 诱发的癫痫模型中，系统成功捕捉到了癫痫发作时神经元异常放电的时空传播路径，以及血管在发作期剧烈扩张、发作后收缩的完整生理与病理动态。

Fig.4 Synchronous imaging of neurovascular dynamics in vivo under ES and epilepsy. (A) FL measurement of FITC and GCaMP6s in vivo, verifying the obvious lifetime difference between the two green fluorescent labels. (B) Simultaneous two-region FL imaging under the resting state and ES. Owing to the fast FL imaging ability over a large FOV, the dynamic changes in blood vessels and local neural activity can be clearly distinguished and simultaneously observed at high spatiotemporal resolution. (C) NVC under ES. Vascular dilations are measured along the white lines shown in B, and the calcium signals are the averaged fluorescence signals of the neurons indicated as white arrows in B. (D) Sequential FL images of four ROIs (white dashed boxes in B) during the last ES period (t: from 373 s to 419 s), corresponding with the calculated values of Δd/d and ΔF/F. (E) Simultaneous two-region FL imaging under the resting state and during seizures. (F) The corresponding spatiotemporal maps of successive neuron discharge in the two regions. (G) NVC during seizures. Vascular dilations are measured along the white lines shown in a, and the calcium signals are the average fluorescence signals inside the white dashed boxes in E. (H and I) Color-coded seizure-spreading events and intensity changes in pixels vs. time along the dashed arrows.#

最重要的是，LD-2P-FLIM将活体脑功能成像提升到了绝对浓度的定量水平，如图5所示。研究团队利用钙离子敏感的荧光寿命传感器jRCaMP1b，在体外建立细胞校准曲线后，实现了对活体神经元群体钙浓度的定量监测。相比传统的荧光强度变化率，荧光寿命变化率（Δτ/τ）在检测神经元自发活动时的敏感度提升了1.8倍。在癫痫发作过程中，系统定量测得神经元内游离钙浓度从自发状态下的81.15 nM剧增至发作期的856.80 nM，这一定量结果与同步记录的脑电图（EEG）电压幅值变化呈现高度一致性，验证了该技术在活体动物大群体神经元定量成像中的可靠性。

Fig.5 LD-2P-FLIM allows quantitative calcium imaging during spontaneous activities and epilepsy. (A) Calibration curve of the lifetimes in situ through imaging cultured cells. The x axis represents the logarithms of the free calcium concentrations using the Calcium Calibration Buffer Kit #1. (i) In vivo calcium imaging in the resting state and during spontaneous activity. (C) Normalized changes in lifetime and intensity from one neuron (indicated as yellow arrows in B). The change rates were calculated by ΔF/F or Δ_τ_/τ. (D) Comparison of change rates between intensity and lifetime imaging during spontaneous activities. The results (calculated from more than 10 neurons) suggested that, compared with those of intensity imaging, the change rates of lifetime imaging were improved by 1.8-fold during spontaneous activity. (E) In vivo calcium imaging of the two regions in the resting state and during seizures. (F) Zoomed-in FL images and the corresponding intensity images from the white dashed box in E. (G) Normalized changes in lifetime and intensity from one neuron (indicated as yellow arrows in F). (H) Comparison of change rates between intensity and lifetime imaging during seizures in the two regions. Similar to spontaneous activities, the change rates of lifetime imaging were obviously greater than those of intensity imaging during seizures in the two regions. (I) Comparison of the measured calcium concentrations between spontaneous calcium activities and seizures by LD-2P-FLIM, indicating its feasibility for in vivo quantitative calcium imaging. (J) Two representative EEGs from different mice during spontaneous activities and seizures. (K) The statistical results of measured voltages between spontaneous calcium activities and seizures by EEG.#