Indiana University      Research & Creative Activity      April 1998 Volume XXI Number 2

Three-Dimensional Visualization and Cellular Imaging:

Technologies in Interaction

by William Rozycki

The hum of computers is the only noise in a darkened, climate-controlled room as Kenneth Dunn adjusts controls, methodically capturing images ever deeper in a kidney cell. A laser illuminates a focal area inside the cell, a microscope scans the fluorescent emissions, and a photomultiplier tube records the images. Dunn, an assistant professor of medicine at the Indiana University School of Medicine, examines cell processes of protein transport that may be key to understand of several diseases.

Kenneth Dunn, Assistant Professor of Medicine, Indiana University School of Medicine, and Director of the Nephrology Division's optical imaging facility, inspects digital images of kidney tubules generated with a laser scanning confocal microscope. The facility includes three other microscopes- for high resolution epifluorescence, differential interference, and phase-contrast microscopy--and is equipped with micro-incubators for live cell studies and for micro-injection. The microscopes are housed in three separate rooms adjoining a central image processing center equipped with Pentium and Silicon Graphics image processing computers and a publication-quality dye-sublimation printer. Software developed in collaboration with the IUPUI Department of Computer Sciences provides users with the latest in 3D image processing, analysis, and visualization capabilities. --credit

A specialist in cell biology, Dunn studies endocytosis, the mechanism by which various materials are incorporated into cells. Endocytosis is exploited by some viruses, including HIV, to infect cells and to spread from cell to cell; endocytosis is also the means by which cholesterol is internalized and degraded at the cellular level. Dunn's research also addresses how molecular transport proteins are regulated by endocytosis, for example how the genetically defective transporter responsible for cystic fibrosis impairs cellular functions, or how cancer cells use molecular transporters to resist therapeutic drugs.

To carry out his research, Dunn introduces a variety of fluorescently labeled proteins into living cells. After exciting the fluorescence (a process in which a molecule absorbs and then re-emits light) by illumination with a laser, he tracks the path of these chemical markers as the cell carries out its regular functions. Quantifying the chemical markers as they make their way through the cell, and thereby mapping the pathways, is a task made possible by confocal microscopy and computer imaging.

"Confocal microscopes have a very narrow focus, so that unlike a wide field microscope, they can discriminate depth," Dunn explains. Wide field microscopes are the type familiar to those who have taken a biology class. They provide a good overall view of most cells. However, an epithelial cell--a type typically found on surface structures like skin, as well as in the linings of internal organs--has a tall, polar shape. To observe its structure and working at all depths is only possible with a confocal microscope. The confocal microscope, as used by Dunn and other researchers, probes the full depth of a cell by multiple scans. When these scans are combined, they present a three-dimensional view of the cell. The acquisition, manipulation, and analysis of this data are only possible through a remarkable confluence of instrumentation and measurement technologies. Dunn's current research techniques stand upon a tripod of advances: in fluorescent probes, in sensitive microscope imaging equipment, and in efficient and affordable image processing systems.

Obtaining images of a cell's interior first requires fluorescently labeling the cellular structures of interest. For fixed samples, this generally involves adding fluorescently labeled antibodies, which bind to particular proteins, thus making particular cellular structures fluoresce. For living samples, which most fluorescent molecules won't permeate, more ingenious methods are required. For example, cells may be microinjected with fluorescent molecules that will bind to particular cellular structures, or cells may even be genetically engineered to produce fluorescent forms of particular molecules. The fluorophores are excited into fluorescence by a controlled light source, usually a laser operated in conjunction with the microscope. The amount of fluorescence then detected by the microscope depends on several parameters, including the amount of fluorescent probe, the amount of background fluorescence, the spectrum of the illumination used for the microscopy, and the optical characteristics of the microscope itself. Many studies require that the fluorescence signal be quantified, a process that is simplified by the use of ratio calculation, which can control for many of these extraneous factors.

After digital images of intracellular fluorescence emissions have been obtained, the next step in imaging cells is the manipulation and visualization of the data. To obtain clear distinctions among separate objects within a cell, algorithms are applied to digitally adjust the brightness and contrast. Most of the software for this process is designed for two-dimensional applications. The need to adapt it to three-dimensional image processing ties in with the final step, visualization.

This actin cytoskeleton of a kidney tubule consists of about twenty cells in cross-section. The image is reconstructed from an original series of eighty images collected with a laser scanning confocal microscope. In the reconstructed image, one can see the actin-rich microvilli, which form finger-like projections at the surface of tubular cells, extending into the center of the tubule. The actin stress fibers support the base of each cell, forming a thin ring around the entire tubule. The actin cytoskeleton is seriously disrupted during renal injury, resulting in loss of kidney function, and various members of the nephrology group are studying why that disruption occurs and how it might be prevented.--credit

Microscopy has been an immensely powerful tool in medicine. Much of what we know about how normal and diseased cells behave has come from understanding the structure and organization of different kinds of cells. Yet, unlike textbook representations, cells are really three-dimensional, and a full understanding of cell organization only comes through three-dimensional microscopy. Advances during the past few decades in optical technology and, above all, in computer processing, gave Dunn and other researchers the tools to image three-dimensionally. But the technology for three-dimensional microscopy is in its infancy, and plenty of challenges remain.

Great advances in imaging technology in other areas of medicine have made possible the three-dimensional visualization of internal organs of the human body. However, the visualization algorithms that work so well for CAT scan (computerized-axial tomography scanning) and MRI (magnetic resonance imaging) do not carry over to microscopy. The optics of microscopy, with inherent optical aberrations and poor signal-to-noise ratios, create unique problems for post-image analysis and visualization. Additionally, cell structures at microscopic scale typically show greater complexity than CAT scan or MRI images of human organs. All of this means that one of the greatest challenges for Dunn and other cellular researchers is in the final step, that of visual representation and manipulation of the images.

"The process of three-dimensional microscopy," Dunn remarks, "is like trying to read the pages of a transparent book, without opening the book. All the words may be visible from the outside, but they are not legible." The confocal microscope, says Dunn, was developed to address the problem. Providing a clear view at various depths in cells, the confocal microscope allows researchers to distinguish each page, as it were, of the book. Once the images are collected, the book must be reconstructed. That is the task of computer processing techniques--in essence, putting all the pages back together again in their proper order. It is the most difficult step of all because the software programs that do this are only now being developed and honed.

At the optical microscopy facility of the Department of Medicine's nephrology section, Dunn shows a visitor several colored, three-dimensional representations of columnar cells on a computer screen. It is difficult to see the image as more than two-dimensional because there are no clear background identifiers. Dunn moves the image on different axials, and eventually a sense of the dimensions comes across. "It's difficult," Dunn comments, "to evaluate the images after we've collected them. Since both computer and printed displays are two-dimensional, we have to take advantage of various optical illusions, like shading or rotating a projection of the image volume, to try to impart a sense of depth." These methods do not retain all the information of the original image, so the visualization is less than perfect. The approach is also subjective, depending as it does on the technique chosen for visualization. Does shading the polarity bring across the sense of depth better than a rotation? Or is a combination of the two more effective to the eye? These are decisions a human, manipulating the processing software, must make. "To objectively assess an image volume, you need to compare and optimize different approaches," Dunn explains. "That currently requires hours of computer processing."

Dunn, who is director of the microscope facility, has seen progress since beginning a collaboration with the Department of Computer and Information Science at Indiana University-Purdue University Indianapolis. The department is a national leader in work on virtual reality imaging and other three-dimensional representations. Tom Biddlecome, a graduate student in computer and information science, has been working with the optical microscopy facility during the past year to improve its three-dimensional image processing and visualization. Dunn shows a cellular structure on the computer screen and remarks, "Tom [Biddlecome] spent hours on this, to get it to do this." Dunn rotates the axial and the image suddenly has depth, has become three-dimensional.

This image of the microtubule cytoskeleton of nine cultured kidney cells is reconstructed from an original series of eighty images collected with a laser scanning confocal microscope. The microtubules, which provide the tracks on which vesicles are transported throughout the cell, can be seen forming a vertical network in the cell that arches over the top of the cell nucleus. Dunn studies how microtubules are used in intracellular transport, with the aim of understanding how endocytosis transports antibodies across epithelia to prevent infection, and how it is exploited by viruses like HIV to facilitate their invasion and spread.--credit

Dunn and a colleague, Dr. Robert Bacallao, an associate professor of medicine at the IU School of Medicine, have teamed with Shiaofen Fang, an assistant professor of computer and information science, and Mihran Tuceryan, an associate professor of computer and information science, both at IUPUI, on an ambitious collaborative project under the auspices of IU's Strategic Directions Initiative (SDI), an internal funding program emanating from the Office of the President. The project joins the expertise of computer specialists with the knowledge held by biomedical researchers to improve three-dimensional characterization of microscopic structures. Dunn and Bacallao will input the parameters needed for cellular microscopy, and Fang and Tuceryan will develop and apply image processing and visualization techniques. "One of the goals of our SDI project is to develop fast algorithms for visualization so that this process can occur in as close to real time as possible," Dunn says. The hope is that software development and new applications, such as the three-dimensional visualization of the IUPUI ImmersaDesk system can reduce the lengthy and laborious processing time now required to get suitable three-dimensional visualization.

Dunn plans to use a portion of the SDI funds to enhance the image collection capabilities of the nephrology imaging facility through the addition of a state-of-the art two-photon microscopy system. The two-photon system is an alternative to confocal microscopy that offers two significant advantages. First, the illumination of the two-photon system is much less harmful to cells, allowing more accurate characterizations of physiological processes for longer periods. Second, biological tissues are more transparent to the infrared illumination of the two-photon system, so images may be collected much further into tissues. This aids the ultimate goal of understanding the cellular-level mechanisms of HIV/AIDS, cancer, arteriosclerosis, and other diseases.

Summing up his hopes for the project, Dunn states, "I'm really excited by the combination of technologies and expertise we have at IUPUI. Three-dimensional microscopy and three-dimensional computer imaging have developed so rapidly in the past few years, but with almost no cross-fertilization. We have the opportunity to bring these technologies together in a way that can profoundly improve our understanding of how the cell functions." dingbat

Return to the Table of Contents