Sometimes scientists are interested in rare cellular events, such as mitotic recombination, an error in cell division that happens perhaps once in 100,000 divisions. Even with labeling, spotting the condition just once is difficult, but collecting statistically meaningful information for analysis requires many examples. However, a system that combines automated multifocal, multiphoton microscopy with robotics and a microtome may offer a solution to this problem.Timothy Ragan and Karsten Bahlmann, co-founders of TissueVision Inc., and their colleagues at MIT and at Harvard Medical School, all in Cambridge, developed the system, which uses a microlens array to split a beam from a Spectra-Physics Ti:sapphire laser into multiple excitation points for multiphoton microscopy. The array of foci is swept across the surface of a sample with a galvanometric scanning mirror from Cambridge Technology Inc. of Lexington, Mass. A dichroic mirror separates the returning emission signal from the excitation signal and sends it to a Hamamatsu multianode photomultiplier tube. Together, the microlens array and parallel optics help increase the overall scanning speed of the system.Because the system is two-photon-based, it can image deeper (between 200 and about 500 μm) into tissue than single-photon microscopy. However, it can’t reach the centimeter or so depth needed to image entire organs. To overcome this limitation, it has a robotic stage that moves the sample from the optics and onto an automated milling machine. After a layer of tissue is removed, the stage moves the sample back to the optics. In this way, the instrument steps down through the tissue, with enough overlap between slices for images from different layers to be co-registered.A cluster of GFP-labeled cancer cells (green) were located within a bulk mouse liver (a).As in flow cytometry, rare events can be characterized but with the added advantage of maintaining crucial information about native tissue morphology and vascularization. Scale bars are 1 mm for a and 50 μm for b.Several enabling technologies made the system possible, Ragan said. The advent of multicore microprocessors allowed additional computing power to be given to the image analysis that the system performs. Multicore processors also help because the processes used in the system lend themselves to being split up onto multiple cores running in tandem.Another key technology advance is the continued decline in the cost of magnetic storage. The system can generate terabytes of data, all of which must be stored and be readily accessible — requirements best satisfied with a hard disk drive.Fast imagingHowever, Ragan noted that what really helped the instrument fly was the combination of multifoci and the galvanometric scanner, which allows scanning of relatively large areas and quick acquisition of data sets. Quickly scanning relatively large areas pushed the galvanometric technology because this required moving the scanner through a large angle and then stopping very quickly where desired. In response to the needs of TissueVision and other customers, Cambridge Technology has been attacking a fundamental limiting factor to such motion: As scanners move, they generate heat that must be dissipated.The solution, as seen in the 6215H scanner used in the TissueVision system, is to improve the torque produced per watt, thereby boosting efficiency and reducing heat. This also allows it to fit into a smaller volume, said Cambridge Technology senior engineer Devin Maxey-Billings. A smaller galvanometric device will perform better in general, he continued. In the design and construction, however, a balance must be struck with the stiffness of the rotor, another fundamental limitation. A very heavy rotor will be stiff and will move precisely but will consume a lot of energy. On the other hand, a very light rotor will move easily but will be too flexible and may coil up like a spring. The result would be a very inexact positioning.An entire mouse heart measuring 5 × 5 × 6 mm was imaged at micron resolution to reveal the details of the nuclei (blue), vasculature (green) and 3-D tissue (red) morphologies. Multiscale imaging allows correlation of subcellular features, mesoscale tissue morphology and macroscale organ anatomy, all on the same data set. The scale bars in the figure are 1 mm, 200 μm and 20 μm for b, c and d, respectively.Thanks to the speed, TissueVision’s prototype has demonstrated the capability necessary to image an entire mouse heart in a day, significantly faster than the 60 days needed for the same task using standard equipment. Future refinements should reduce that time to about 4 h.Besides eventual instrument sales, TissueVision plans to offer a tissue imaging service. The company will image a customer’s sample and provide high-resolution scans of cells in their natural setting, thus allowing contextual information to be gathered such as on vasculature and on the relationship of the cells to the larger organ.Contact: Timothy Ragan, TissueVision Inc., Cambridge, Mass.; e-mail: tragan@tissuevision.com and Devin Maxey-Billings, Cambridge Technology Inc., Lexington, Mass; e-mail: scanners@camtech.com.