Faculty Experiences - Bill Grisham
What matters most to you in your teaching?
How are you using technology as a tool to achieve your teaching goals?
How have your students responded to your use of technology?
What new goals do you have for using technology in teaching?
How could the University better facilitate the use of technology in instruction?
Class web site
NIH Image (software)
Replicating Experiments with Image Data
The most important skill neuroscience students need to acquire is how to analyze data and interpret it correctly. From that, they learn how to understand other people's data; throughout their lives, they will be confronted with data trying to convince them of one thing or another, and they need to be able to sort out that which is valid from that which is not. I think that's the biggest take-home value of Psychology 116, the psychobiology lab course. Beyond that, they gain experience in neuroscience techniques. We can't cover the whole gamut in a quarter, but we like to give a reasonable sampling from the cellular to the molecular and behavioral levels of data gathering and analysis.
Among the techniques they learn is how to analyze images of tissue at the microscopic level. Once upon a time, we had to use a light microscope hooked up to a video camera, which was expensive: the microscope cost about $500-$1000, the camera and the board that you slap in your computer's another $1500, and we would need that set-up for each lab station. By digitizing all these images and archiving them for repeated use, we've managed to provide this experience to everyone at the same time, at the same moment in the lab, and we have cut lab costs by about $2000 per station.
The images in the Rats Spinal Cord Image Archive (SCIA) are all from an experiment I published previously in which I was studying how hormones affect the development of a particular set of neurons in 36 rats. At that time we took pictures of all the relevant neurons, sectioned at 50 microns. There are images of spinal cords from male and female rats, with two treatments: some were injected in utero with an anti-androgen drug, and some got a control injection. It's what we call a "two by two design." Each student gets assigned four sets of images from four rats, and they don't know which rat is which. They analyze the presence and size of the neurons, and after they've gone through their four sets of images, I drop the blind and tell them which rats were control males, control females, anti-androgen-treated males, and anti-androgen-treated females. We put all the students' data together and analyze it. Not only should they find sex differences--the neurons are more numerous and larger in males than in females--but they should see a precipitous reduction in the number of neurons in the males that have had the anti-androgen treatment. They are also required to write an APA-style report on the experiment. I give them the background in lectures and reading assignments, and talk about the methods of how this tissue was originally prepared.
The analysis is done using NIH image, a software package from the National Institutes of Health. It's a freebie that anyone can download from the NIH web site, and features a graphics-based interface that lets you quantify the image. The student loads the image into the software, and once they've identified a neuron, they draw around the outside of it. A panel in the corner takes the input from the outline and displays the size in square microns. The student also counts the neurons in the image.
The original inspiration for this project came from a conversation at a conference on how to teach neuroscience to undergraduates. In the course of trading ideas with another attendee, I began describing the live version of this rat spinal cord lab. I told her, "this is a great lab, even if you just want to get male versus female differences. You snip the spinal cords out, you cut them, you stain them, you count the cells, it works every time." She responded, "I couldn't possibly do that. I don't have tools to cut tissue. I don't have microscopes." I once taught at an institution like that--my lab was four bare walls, and there was no budget. So I began thinking about how I could make this something that was accessible to anyone with a computer, and it occurred to me that I could digitize the images.
The process was actually pretty cheap. I started doing it myself, because I thought I could do it in a weekend. When I realized I couldn't, I had a work-study student working on it for a while. She got interested in it, and ended up doing a student research project in which she tried to replicate the original experiment. I didn't want to put it out there for everyone if it didn't work. As we did a little bit of sampling, it turned out that to digitize each and every last section was going to take a huge amount of storage space, so we selected every fourth. In the end she analyzed the pictures in a blind test, and demonstrated that the experiment would still work using the selection of images in the archive. I'm currently working with the same student to develop a module on the bird song system, which also has big sex differences and is affected by hormones early in development. Other faculty have created experimental modules for use with NIH Image--students are examining images of mouse brains in the Quantitative Trait Loci (QTL) module developed by Jack Beatty, for example. In addition, there is a project in the works to develop a lab in which students analyze sex differences in human brains by examining structural MRI images.
Rats SCIA contains 700 images in total. My ultimate goal is for the archive to reside permanently somewhere on the web. At the moment, all these images are available online from the Journal of Undergraduate Neuroscience Education. You can download all the sets, along with an article that I wrote about how to use them, and run this lab anywhere in the world, as long as you have a computer.
I think students like using Rats SCIA a little bit better than prepared slides. When they're working with real tissue, they have to find the relevant neurons, and there's a lot of stuff that is irrelevant that they have to learn to ignore. There's a much larger sample on each slide than in each image, and they have to poke around in it to find the stuff that they need, and figure out where the relevant neurons start and stop, and then count them up. With Rats SCIA, it's a lot faster. I still think there's some value--especially since most of my students are going to grow up to be doctors and dentists and that sort of thing--in their learning to do this with a real microscope. But they like working with this a lot better because it's just so much more efficient and they don't get distracted and confused.
The UCLA library has become really interested in this project, because they want to be an accessible information resource for students. Right now, these images are all in the UCLA Digital Library. They want to see if can they can take a series of such collections--including materials from other disciplines such as Art History--and then make them accessible through the California Digital Library. I've also had interest from people at Washington College in Maryland and at Northern Illinois University, who want copies of the archive to use in their own course labs.
The UCLA Digital Library is also working with us to make the archive accessible via a course web site. If you could do that, you could use it in blended instruction--in fact, I'm going to try that this summer on a limited scale. I applied for a mini-grant from the Office of Instructional Development (OID) to enable us to pay University Extension to host a class web site for us, because they have the necessary software. The older version of Blackboard Life Sciences Computing has will not accommodate the file format used by the archive. I'm only going to use Rats SCIA on a minimum basis to start, so if the system completely collapses, we'll still have other material to cover.
Blended instruction is a first step toward thinking about doing this lab as a complete distance course. Students currently have to spend three hours a week in the lab. This summer I'm going to try to have them do three of the labs on site, and the fourth lab on their own computers. If it became a complete distance-learning course, they'd be assigned their four animals, they'd download the tool, calibrate the tool, get the data, and then they could go back to the web site and enter their data. Ultimately I would give them back a spreadsheet saying, OK, the blinds are dropped; now here is all the data from all the different groups. I don't think that's in anybody's thinking right now, but it opens the door to that possibility.