Computer-assisted image analysis helps UConn researchers learn how bone heals

University of Connecticut researchers in the School of Dental Medicine are studying how bones heal when fractured. As the lead researchers for a large international study called Knockout Mouse Project (KOMP), the team needed a way to image the bones obtained from animals in a consistent and highly automated manner to facilitate complex digital analyses.

 Using the new Axio Scan.Z1 whole slide scanner developed by Carl Zeiss Microscopy, the team is now able to deliver on its goal of gathering huge amounts of visual data that can be analyzed faster and with a greater amount of accuracy. The team considers the Axio Scan.Z1 to be a revolutionary machine for imaging complex multi stained samples. The ability to have nine or more fluorescence channels as well as a brightfield channel enables them to quickly and easily image multiple stains and sample preparation without complex external filter wheels.

Skeletal research requires faster and more accurate image analysis

Dr. David Rowe, professor of reconstructive sciences at the UConn Health Center’s School of Dental Medicine, proudly refers to himself as a bonehead. His lab conducts skeletal research, focusing on how bones heal when fractured and how bones renew themselves. The method of assessing how rapidly a bone is being remade is known as dynamic histomorphometry, the quantitative study of the microscopic organization and structure of bone, especially by computer-assisted analysis of images formed by a microscope. ^[1]

The work used to be done manually in a tedious and time consuming process, in which researchers used a microscope to image multiple tiles of a whole slide and then used custom build stitching software to combine the times into one large image.   Over the last ten years, researchers have been making a serious attempt to further automate the process specifically for histomorphometry.

They had developed a new way of studying histomorphometry on frozen sections of bone, rather than using the classical technique of plastic or paraffin embedding of bones. The new technique made it possible to see sections in florescence in a way not possible with other methods. Showing florescence in such clear and distinct ways lends itself very well to computer automation and interpretation of slides, so the team needed better ways for the computer and microscope to work together to be more automated. “Once you make the transition to florescence, you never go back,” says Dr. Rowe. “Colorimetric dumbs the data down so much – if you can use florescence you can generate four to five times more information than color.”

They began working with ZEISS instruments, including the MIRAX Midi Slide Scanner in 2008, which provided automatic capturing of whole slide images more easily than standard research microscopes. The MIRAX Midi was recently replaced with the new ZEISS Axio Scan.Z1, a powerful whole slide scanner that enables researchers to digitize specimens and create high-quality digital slides of a consistently high quality, even when capturing fluorescence images at very high speeds.

The UConn researchers, in collaboration with The Jackson Laboratory in Bar Harbor, ME, are serving as lead skeletal researchers for KOMP, a massive National Institutes of Health project to “knock out” (delete) each gene in the mouse genome to systematically examine the function of each particular gene on the animal’s development and health. ^[2]

Participation in KOMP significantly increased the UConn researchers needs for a high quality, high capacity whole slide scanner, as the team hoped to look at the hundreds of mice that come through the study every year in a very consistent way. Researchers jumped at the chance to jettison prior manual methods and use the new ZEISS Axio Scan.Z1 to scan the huge number of slide images and submit them to a computer for image analysis.

“With prior instruments there was a wealth of information on slides that was not being captured. The information is so valuable – it is tied in with genomic and molecular data collected from the same tissues, which was not being integrated,” said Dr. Rowe.

The availability of the ZEISS Axio Scan.Z1 microscope removed major impediments to the team’s goal of capturing and automating data. Researchers can now make measurements in minutes that would have taken multiple days in the past. What may have taken a full 40-hour week can now be done in the equivalent of one day. Not only can researchers process a larger number of samples, they are also able to avoid all personal bias, since all the information is machine-captured and analyzed by computer. In the past, the likelihood of subjective bias was so high that histology could not be used as a cross-laboratory quantitative tool.  The new imaging platform and associated image analysis program provide an opportunity of consistent and uniformly performed histological evaluation of skeletal structures that are affected a genetic locus 

Automation saves time and improves data collection

The new techniques mean that for any one section, one imaging set will yield a vast amount of information. Scientists prepare bone sections on a glass microscope slide and the computer provides a precise location; researchers then provide an identifier to the imaging system. Once found by the imaging system, the section will be imaged under high power and repeated for as many different florescent colors as investigators are interested in seeing, typically four to five. 

The team looks at the same section under one set of conditions, then another set, and then another. Each reveals one parameter; layering one image on top of another enables researchers to see how they relate, revealing the complexity found within the tissue.

The technique allows the team to image sections from many mice and also many sections from one mouse. It allows them to cut all the way through the tissue, taking sections, scanning each one and using the computer to line them up to reconstruct the tissue in two dimensions from which it is possible to understand the tissue in three dimensions.

 “We can do this procedure on a minimum of 16 different mice, each with three sections per mouse,” explains Dr. Rowe. “It all happens overnight while we’re away, so it takes very little of a technician’s time. The optics are great, the images are in focus, the software finds the areas of interest easily, and stiches the images together without a flaw. And the images we get are gorgeous.”

The instrument empowers the researchers to look at the large volume of sections to spot variations. The numbers of samples they get takes care of biologic variations, and the consistency achieved takes care of technical variations. Researchers think they can use this as a tool to measure genotype versus phenotype in a far more sensitive way, and in the volume needed for such a large scale big science study of bones and joints.

Dr. Rowe concludes, “ZEISS has a winner with the Axio Scan.Z1. The impact it has on being able to generate large volumes of high resolution imaging has met expectations, and we are very pleased. The instrument is an essential tool for gathering visual data that can be systematically deconvoluted to make measurements.”

[1] http://en.wiktionary.org/wiki/histomorphometry, retrieved 1/29/14 

[2] JAX, UConn Collaborators Awarded $3.2 Million Grant, http://today.uconn.edu/blog/2013/08/jax-uconn-collaborators-awarded-3-2-million-grant/, retrieved 1/29/14

Source: Carl Zeiss Microscopy