MPScope: A versatile software suite for multiphoton microscopy

https://doi.org/10.1016/j.jneumeth.2006.03.001Get rights and content

Abstract

MPScope is a software suite to control and analyze data from custom-built multiphoton laser scanning fluorescence microscopes. The acquisition program MPScan acquires, displays and stores movies, linescans, image stacks or arbitrary regions from up to four imaging channels and up to two analog inputs, while plotting the intensity of regions of interest in real-time. Bidirectional linescans allow 256 × 256 pixel frames to be acquired at up to 10 fps with typical galvanometric scanners. A fast stack mode combines movie acquisition with continuous z-focus motion and adjustment of laser intensity for constant image brightness. Fast stacks can be automated by custom programs running in an integrated scripting environment, allowing a 1 mm3 cortical volume to be sampled in 1 billion voxels in approximately 1 h. The analysis program MPView allows viewing of stored frames, projections, automatic detection of cells and plotting of their average intensity across frames, direct frame transfer to Matlab, AVI movie creation and file export to ImageJ. The combination of optimized code, multithreading and COM (Common Object Model) technologies enables MPScope to fully take advantage of custom-built two-photon microscopes and to simplify their realization.

Introduction

Multiphoton microscopy (Denk et al., 1990) was first applied to neurobiology more than a decade ago (Denk et al., 1994) and has since acquired a prominent status among neuroscience methods (for a review, see Nguyen et al., in press). The advantages of multiphoton laser scanning microscopes (MPLSMs) include deep penetration (100–1000 μm) into highly scattering neural tissues (Helmchen and Denk, 2005), micron to sub-micron resolution and optical z-sectioning, as well as reduction of fluorophore bleaching and/or phototoxic damages.

Custom-built MPLSMs (e.g. Fan et al., 1999, Mainen et al., 1999, Majewska et al., 2000, Nguyen et al., 2001, Nikolenko et al., 2003, Roorda et al., 2004, Tan et al., 1999, Tsai et al., 2002) are not only considerably cheaper than commercial microscopes (by almost a factor of two) but can also be specifically designed for a particular purpose such as in vivo recordings (Tsai et al., 2002). One limitation of this approach is that programs controlling custom MPLSMs (e.g. Nguyen et al., 2001, Pologruto et al., 2003, Tsai et al., 2002) incorporate only a certain number of features from an ideal “wish list” of requirements for neuroscience.

The most important feature is a high frame rate that will enable the capture of fast functional events such as variations in intracellular calcium concentrations as well as the possibility to provide a near real-time assessment of cellular responses as the experiment unfolds. This can be achieved by first drawing areas of interest encompassing select neurons and then by having the program plot during the acquisition session the average pixel intensity of these areas for each frame as a time series in an oscilloscope-like display. Also, simultaneous frame acquisition, display and disk streaming is of considerable importance and thus precludes the use of a “blind” mode where frames are captured and stored without any form of visual feedback.

Further, the need to coordinate electrophysiological recordings and image acquisition is important for many functional experiments. One solution is to centralize the control of most electrophysiological operations in the scanning program to ensure proper synchronization of the analog and video streams.

Another requirement of MPLSM systems is the ability to generate, magnify and rotate laser scan patterns in order to bring the scanning field into a favorable orientation to image particular structures or processes such as dendrites or blood vessels. This feature is critical when measuring cerebral blood flow (Kleinfeld et al., 1998), since the method depends on aligning linescans on top of linear portions of blood vessels. Some multiphoton microscopes (e.g. Tsai et al., 2002) depend on a custom-designed electronic circuit based on an embedded DSP or PIC processor to rotate and magnify the scan pattern. These devices increase the overall complexity of the microscope and are difficult to replicate, maintain and upgrade. The phenomenal computing power now available in virtually all current personal computers should, in principle, be largely sufficient to allow the MPLSM program to undertake the functions of these devices.

Perhaps the most difficult requirement for MPLSM software is the possibility to use the same program in widely different experiments ranging from time-lapse imaging (Ruthazer and Cline, 2002) to histology (Tsai et al., 2003). MPLSM programs can, in theory, be customized by modifying their source code. However, this solution can challenging even when the programs are written in the Matlab (MathWorks, Natick, MA) or Labview (National Instruments, Austin, TX) programming languages. As an alternative, customization and automation of repetitive tasks can be achieved by providing users with the possibility to write small programs called scripts, usually in an easy to learn programming language like Microsoft VBA (Visual Basic for Application) or VBScript (Visual Basic Scripting Edition). This approach is adopted in the commercial programs LaserSharp and LSM that control, respectively, the Radiance2100 MP from Bio-Rad (now Carl Zeiss CellScience) and the LSM 510 NLO (Zeiss) multiphoton systems, and also in the custom-build system described by Nikolenko et al. (2003). Scripting based on the Microsoft ActiveX standard has the advantage of allowing users to include in their scripts some of the many ActiveX software components already written or to tap into functionalities present in existing programs, thus creating script-based “meta-applications” (Nguyen and Miledi, 2003).

In this paper we present MPScope, a suite of turnkey applications that fulfills the requirements outlined above. MPScan, the frame acquisition program, is flexible enough to control any MPLSM based on generic scanning and data acquisition hardware. MPView is designed to analyze the data generated by MPScan. We describe the design principles we adopted for MPScope and present benchmark data based on a novel imaging scheme allowing deep z-stacks to be taken almost up to an order of magnitude faster than with a previous program written in LabView.

Section snippets

Software development

MPScope was developed in Object Pascal using the Delphi 5 programming environment (Borland Corp., Scotts Valley, CA) on a Dell Inspiron 8000 laptop (Dell Corp., Austin, TX). MPScan ran on a dual 1 GHz processor Dell workstation with 1 GB RAM running Microsoft Windows 2000 Professional Edition (Microsoft Corp., Redmond, WA).

Software and hardware requirements

MPScope is designed for IBM-compatible personal computers running the Microsoft Windows 2000 or XP operating systems. As stand-alone compiled executables, MPScan and MPView do

MPScan: principle of operations

In most imaging modes, MPScan drives the fast mirror using a symmetrical waveform in order to record pixels during the forward and backward linear motion of the mirror. The dual direction linescan pattern, which resembles a sine wave, consists of two linear (e.g. constant speed) segments connected with parabolic turnabout sections. The shape of the linescan was designed to avoid sharp discontinuities that could have resulted in distortions of the fast mirror position due the limited bandwidth

MPScan: design goal

The main design objective of MPScan was to maximize the frame rate of our custom multiphoton system, which was not limited by the scanning hardware but the performance of the LabView program that originally controlled the microscope. This program and also the ScanImage application developed in Matlab by the Svoboda lab (Pologruto et al., 2003) can achieve only two frames per second with 256 × 256 frames (Timothy O’Connor, Svoboda lab, Personal communication). In theory, the scanners allow frames

Conclusion

All MPLSMs, including those custom-built, are very expensive instruments that require extensive care in the design of their optics and electronics to fully exploit the potential of the microscopes. Similarly, programs controlling custom-built MPLSMs should be written to maximize the capabilities of the microscope. To achieve the highest performances, advanced features of modern operating systems can be used to develop powerful control and analysis software comparable to much more expensive

Acknowledgements

We would like to thank Ms. Nozomi Nishimura and Dr. Chris B. Schaffer (Cornell University) for suggestions to enhance MPScope. Q.-T.N. is funded by the National Institute of Mental Health (MH071566). P.S.T. and D.K. are supported by the National Institute of Health (RR021907 and EB003832) and the National Science Foundation (DBI0455027).

Glossary

ActiveX Automation
A COM standard to programmatically control local or remote applications or libraries.
API
Application Programming Interface. A set of programmatic procedures or functions specific to a particular task.
AVI
Audio Video Interleaved. A multimedia file format for movies or sound clips.
COM
Common Object Model. A set of specifications to make objects and applications interoperable in the Microsoft Windows operating system.
Demand Mode
An enhancement to DMA circuits allowing them to

Cited by (0)

View full text