The second, more difficult aspect of the MRI was the noise. MRIs are loud. When performing a functional scan, the MRI sounds like a machine gun. At nearly 100 decibels, being in an MRI is like standing next to a leaf blower. Although not terribly painful for humans, dogs have more sensitive hearing, and we worried about damaging their ears. Also, many dogs are just plain afraid of loud noises. It would do no good to scan the brain of an anxious, frightened dog. Not only would we need to find dogs with a calm temperament, we would still need to get them used to the noise. The simulator had to mimic this key aspect of the MRI procedure.
It is commonly believed that dogs have more sensitive hearing than humans. But how much more? At the low end, humans can hear frequencies of about 8 Hz, which sounds like a very deep vibration. The high end is limited to about 20,000 Hz. This range tends to shrink with age, with the high frequencies being lost to ear damage from loud noises over the years. Most human speech is in the 300 to 3,000 Hz range. The first investigation of dogs’ hearing was done in the 1940s. But because sound-generation technology was limited at the time, the scientists couldn’t generate very high frequencies, and they couldn’t determine the dogs’ upper frequency range. It wasn’t until the 1980s that technology could reliably generate high-frequency tones. In the 1990s, an even more sophisticated technique was developed. This technique measures electrical responses in the part of the dog’s brain that responds to sound. These brainstem auditory evoked responses, or BAERs, are also used in humans. Today, most scientists agree that dogs can hear tones up to about 60,000 Hz, well beyond the human range.
The actual MRI scanner makes a wide variety of sounds. These noises originate from what are called the gradient magnets. There are two types of magnetic fields in an MRI. The main field is produced by the miles of superconducting wire that are wrapped around the bore. The main field never changes and is always on. The gradients are much smaller magnetic fields that are constantly changing during a scan. By switching the gradients on and off, we can select specific locations in the brain. A gradient can be switched on by running electrical current through it, which activates the magnetic field. The sudden inrush of electricity causes the magnet to expand slightly, and this rapid expansion causes a pressure wave inside the MRI, which we hear as a loud banging. The exact noise that it makes depends on the type of scan being performed and whether it is a structural or a functional MRI.
We still needed a physical mockup of the scanner to train the dogs in. Many universities that house MRIs for brain research have mock scanners. There are many situations where it makes sense to train people in the scanner before actually scanning their brains. Brain-imaging studies of children, for example, must first teach kids to lie still in the scanner by using a simulator. Because the MRI can be frightening, it is very helpful to allow children to get used to the environment before they go into the real scanner.
It isn’t surprising that a few companies have cropped up to sell mock scanners. The price tag, however, is steep. When we embarked on the Dog Project, the going rate for a mock scanner was about $40,000. Since we had no funding, this wouldn’t be practical. Besides, I couldn’t see spending so much money for what amounted to an empty tube with a few speakers inside to simulate the noise.
But how much of the real MRI would we need to simulate? Did we need something that could fill an entire room, like the real scanner? Or could we get away with a simple tube? After all, the dog was going to be inside only the scanner.
Mark came to the hospital to check out the MRI facility and determine how much of the actual MRI we would need to simulate for dog training. He hopped up on the patient table and lay down, putting his head in the birdcage. With the press of a button, the table glided into the center of the magnet. A quick thumbs-up indicated he was good to go. We went through a few quick brain scans so that Mark could get an idea of the types of noises and how loud they were.
Mark came out of the scanner with a big grin on his face and proclaimed, “This is completely doable.”
“Do you think we need a mockup of the entire scanner?” I asked.
“No, just the patient table and the tube will be fine.”
This was good news, because we were going to have to build the simulator ourselves. The mock scanner would have three elements: a tube to simulate the inside of the scanner; an exact model of the birdcage, which the dog would have to shimmy into; and a sound system to play recordings of the scanner noise at the appropriate loudness.
I was looking forward to this. Constructing a simulator would let me dust off some woodworking equipment that had been lying dormant in the garage. It’s fun to build stuff.
To simulate the inside of the scanner, we needed a tube of the right diameter. The MRI bore measured sixty centimeters, or two feet, in diameter—larger than any tubing you would find at your local hardware store. That is, however, a common diameter of concrete pillars used in the construction of buildings. These pillars are made by pouring concrete into molds sold under the trade name Sonotube. Andrew made a few calls to construction supply houses around Atlanta and soon located a twelve-foot-long, two-foot-wide Sonotube.
“Do they sell it by the foot?” I asked him.
“Nope,” Andrew replied. “We have to buy the whole piece.”
“How much?”
“About a hundred bucks.”
“How long is the MRI bore again?”
“Six feet.”
“This is great,” I said. “We can do this NASA style.”
Andrew looked puzzled.
“In the old days,” I explained, “NASA always launched two spacecraft for their missions. The reason was that most of the cost of a project was in the design and development. Once those were achieved, the added cost of a second spacecraft was minimal. Plus, it added a level of protection if one craft failed. If we need to buy twelve feet of Sonotube, we might as well just cut it in half and build two simulators. We’ll give one to Mark to use at CPT, and I can keep one at home to test on Callie.”
Building the main tube didn’t require much beyond cutting the Sonotube in half. Andrew and I found everything else we needed at the local Home Depot. We bought two folding tables to mount the tubes on. A sheet of plywood and some lumber would serve as the patient table inside the tube. We did the construction in my garage on a Saturday.
Even though the result looked nothing like an actual MRI scanner on the outside, the important part for the dog was the interior of the tube. Andrew had obtained all the measurements from the real scanner. All we had to do was duplicate the height and width of the patient table when fully inserted into the MRI bore. To test it, we took turns crawling into the simulator. We both agreed it felt just as claustrophobic as the real thing.
Andrew and I building the MRI simulator.