We have assembled a resonant scanning 2-photon system based on previous systems designed by Ian Parker (UC Irvine) (1) and Mike Sanderson (UMass Worcester) (2). The original design functioned as a confocal system but more recent incarnations extend the technology (and actually simplify the design) for 2-photon excitation (3) and 4-channel collection using the Raven video card (4). In addition to minor modifications in the overall design of the Parker/Sanderson prototypes, our instruments are expanded to include 4 PMTs and also place those PMTs within the infinity space of the objective. Furthermore, as we are essentially biologists first, we were lucky to find a collection of scan-head parts available from Sutter Instruments. This availability dramatically reduces the need for novices to construct parts.

The primary advantages of this system are:
1. Cost (approximately 1/2 the cost of buying a turnkey system).
2. Four channel acquisition.
3. 60fps maximal acquisition speeds. (Although many deep-tissue applications require frame averaging to increase the signal/noise, there is the potential to do > video rate work when the signal is sufficien).
4. Decreased dispersion. Because the system only has about 1 inch of optics from laser to sample, dispersion is very low meaning the two-photon effect is maximized.
5. Service. This is a relatively easy system to builde. Although a complete optical neophyte will want to hire a consultant to help build this, once you have the fundamentals down, you don’t require an expensive service contract and when breakdowns occur, you are in control of fixing only the parts that are broken.

Software Overview:
When building our instrument, we were greatly aided by an incarnation of a distortion correction filter that was built into Video Savant with using base code from Mike Sanderson. In conjunction with Bitflow Raven data acquisition cards, our system can process up to 60 full frames per second (we typically use a 30 fps however). Video Savant is a software package that acquires images from PMTs or cameras. In this case, the programmers at Video Savant were extremely helpful in writing a custom GUI to control data acquisition as well as driving a Prior XY stage and a PIFOC high-speed piezzo-electric Z-drive of the type we first used to obtain 4D data in widefield and confocal applications). This code simplifies data handling and creates indexed files that can be read in ImageJ, Bitplane or Metamorph (or standard Windows image viewers) for analysis. See 'Software Details' for more details.

Resonant Scanning:
Commercial instruments typically scan the sample on-demand. For most commercial systems, the combination of control aspects for these servo mirrors and other factors conspire such that it typically takes approximately 1-4 seconds per image. In contrast, a resonant scanner is always oscillating to create the scan pattern and you can choose how many scans to average to increase signal-noise. The version that we utilize results in 400x480 pixel scans being generated 30 times per second. In this setting, frame rates of 30fps can be achieved although most applications require some amount of frame-averaging (typically 5-10 frames averaged) resulting in effect rates of 3-6 frames per second. When these scanners are coupled with piezoelectric z-drives (with settling times of approximately 10msec) the result is effectively 3-6 z-sections per second, a dramatic improvement over conventional scanning modes. A downside of this technology is simply that subregions of scanning cannot be defined in order to facilitate FRAP-type bleaching or faster scanning of smaller regions. In practice, this limitation is not significant for many current in vivo applications where FRAP has not yet seen great utility and where large fields are desirable.

Simplified Optics:
2 photon excitation relies high energy pulsed lasers—these generate high photon fluxes for very brief periods of times. Pulsewidth is a measure of the shape of the high energy photon fluxes and sharp peaks mean higher peak power resulting in more robust excitation. As the pulses propagate through glass or quartz optics, longer wavelengths are slowed down relative to shorter wavelengths and this dispersion spreads the peak out resulting in lower overall peak power. Since a dedicated 2-photon instrument does not need a lot of optics (the required lightpath consists of approximately 3 cm of glass, dispersion can be kept low and excitation maximal. As an added mechanism to boost power, we also use a prism based compression system that 'chirps' the pulses going into the microscope and effectively cancels the limited effects of dispersion within the system.

References:
1. Nguyen, Q.T., N. Callamaras, C. Hsieh, and I. Parker. 2001. Construction of a two-photon microscope for video-rate Ca(2+) imaging. Cell Calcium 30:383-393.
2. Sanderson, M., and I. Parker. 2003. Video-rate confocal microscopy. Meth.Enzym. 360:447-481.
3. Krummel, M.F., M.D. Sjaastad, C. Wülfing, and M.M. Davis. 2000. Differential Assembly of CD3z and CD4 During T cell Activation. Science 289:1349-1352.
4. Sanderson, M.J. (2004) Acquisition of multiple real-time images for laser scanning microscopy. Microscopy and analysis 18, 17-23.