Worm Breeder's Gazette 12(2): 19 (January 1, 1992)

These abstracts should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.

Use of digital image archiving for lineaging and 'retrospective embryology'

Andrew Fire

Carnegie Institution of Washington

I have assembled an image archive system based in principle on John White's '4D microscope' (now marketed by BioRad) but differing in that images are stored digitally as opposed to the analog LaserDisk storage used in the White/BioRad system.

Our archive system is based on a relatively inexpensive computer (a 'clone' of an IBM type PC). The computer controls a focusing motor, which cycles the microscope through different focal planes. For each focal plane, an image is collected from a video camera, digitized, and stored (in a somewhat compressed format) on a high capacity magnetic disk drive. In a typical experiment lineaging a single embryo, images would be taken in each of 25 consecutive focal planes in a cycle, with each cycle lasting 30 seconds. The current system (with 2.5 gigabytes of disk storage) can accommodate approximately 18,000 full frame images: six hours of continuous recording at maximum resolution.

Once recorded, images stored on the disk can be 'played back' in any order. Generally a series from a single focal plane is played consecutively in either forward or reverse time. The focal plane can be controlled during playback (using keyboard or joystick), allowing individual cells to be followed as they move between focal planes. Theoretical playback rates are as fast as 60 full frame images per second, but currently the system can only run at a fraction of this rate (3-5 full frame images per second). Although rapid playback is useful for general impressions of developmental events, precise lineaging generally involves using the keypad to step through time and focal planes to follow single cells.

Image series can be archived by copying information from the magnetic disks onto digital tape. An image series stored on tape can be restored at any time by recopying the data onto the hard disk of the system. Each digital tape holds about 40,000 full frame images, costs $5 and can be rewritten multiple times. A disadvantage of digital tape archiving is a waiting time of 1-3 hours for an archived image series to be copied from tape back to disk. When compared to the magnetic tapes that we are using, the analog LaserDisks used to store images in the White/Biorad system have some advantages and disadvantages: LaserDisks can be readily interchanged, provide rapid access to 43,500 images per side, and provide faster playback rates. Unfortunately the LaserDisks cannot be erased or rewritten and are costly (around $200 each; the laserdisk recorder is also considerably more expensive than magnetic disk and tape drives). Unprocessed image quality should be comparable for analog and digital storage systems. The digital system allows for more sophisticated image enhancement and analysis techniques, but at present we are using only very simple image averaging and contrast enhancement. This gives screen images almost equivalent to viewing specimens directly through the microscope.

While early lineages can easily be followed with the system as it is now working, later lineages (i.e. the last two cell divisions) are more difficult to follow reliably. Current efforts aimed at improving image resolution and data compression should facilitate later lineaging.

A second type of experiment for which the setup has been used is to follow large groups (10-30) of embryos at lower magnification (nomarski observation with a 20 or 40x objective; generally 8-12 focal planes recorded for 12-14 hr series). At this resolution one can observe early cleavages, cell movements, morphogenesis and embryonic muscle function. Being able to follow embryos in retrospect (e.g. only those embryos that fail to hatch) facilitates analyses of the earliest defects and course of development of zygotic lethal mutations and deficiencies.

A parts list and the software driving the system are available. Because our system is pieced together and includes some older equipment from the Carnegie attic, duplication would require some mechanical tinkering and likely some programming.

(Thanks to John White, Joe Vokroy and Jim Priess for their help)