Berkeley Lab Science Beat Magazine banner Berkeley Lab logo
November 26, 2003
Robot Revs Up Electron Microscopy Film Scanning

Film scanners don't often make headlines. But a scanner that automatically digitizes 750 microscopy images without reloading may turn some heads.

The scanning robot was developed by a team of Berkeley Lab engineers and biophysicists to keep pace with the deluge of high-resolution images produced by today's electron microscopy research. An inquiry into the molecular structure of a cell membrane, for example, may yield hundreds of images, each of which must be digitized into a database before it's analyzed.

From film to pixels without the drudgery: Robert Nordmeyer prepares the scanner to run through a stack of images.

Unfortunately, currently the only way to load microscopy film into a scanner is by hand, a slow approach in which one image is scanned roughly every ten minutes. It takes a person 120 hours — or fifteen days of full-time work — to scan 750 images.

"Loading the film into a scanner by hand is a very tiresome job," says Robert Nordmeyer of Berkeley Lab's Engineering Division, who partnered with Robert Glaeser of the Life Sciences Division to invent the system.

As Nordmeyer explains, even the fastest commercial scanner requires six minutes to digitize the data in a high-resolution microscopy image. This means the person tasked with loading and unloading the scanner faces a dull mix of downtime and repetitive work: load, wait, unload, reload, wait.

"But we can load the automated scanner with film and let it run night and day, even through the weekend," Nordmeyer says. "This alleviates the need for a full-time person dedicated to scanning."

Glaeser, a biophysicist who uses electron microscopy to explore the biochemistry of proteins found in cell membranes, adds "the instrument promises to make high-resolution projects, which previously no one would attempt, a fast and routine part of electron cryomicroscopy."

The scanning robot resembles one stop in a miniaturized mass production line. A suction-cupped arm grabs the top image from a stack of 750 microscopy films. It places the film into a custom-designed plate that holds the film in a flat plane without creating optical fringes. The plate is then fed into the scanner. Once the scan is complete, the film emerges and the motorized arm transports it to the "done" stack. The system, which runs on Visual Basic software, is designed to work with an off-the-shelf Nikon Super Coolscan 8000 ED scanner.

Nordmeyer also explained why digital cameras haven't yet replaced scanners when it comes to digitizing microscopy images. Although they capture images in a flash, poor scintillator performance means digital cameras provide far less than the 4,000 x 4,000 pixels built into high-end cameras. A scanner, however, can digitize photos at 10,000 x 13,000 pixels. To cover the same area at the same resolution, a digital camera must shoot roughly 30 images.

"To obtain the same resolution as a scanner, we'd have to take a mosaic of squares and form a single image, and this takes time," Nordmeyer says.

In addition, when a microscopy sample is illuminated, say from a digital camera's electron beam, the sample partially deteriorates and valuable information is lost. And in high-resolution work like electron microscopy, losing a few pixels may mean losing potentially rewarding scientific insights.

"There is a lot of information on a microscopy image, and you don't want to lose any of it," Nordmeyers says.

In the future, Nordmeyer and colleagues hope to boost their scanning rate. One possibility involves adding another scanner to the system, a cheap, simple solution that could double its speed from one scan every six minutes to one scan every three minutes.