Site-wide links

Bio-X

Bio-X builds on RIT's core technical strengths to address biological, health-care, and medical challenges of the 21st century through interdisciplinary research.

Imaging to Enhance Diagnosis

Imaging to Enhance Diagnosis

RIT researchers are developing and evaluating imaging methods that enhance the accuracy and efficiency of medical diagnoses.

Biomedical Imaging at RIT

While the field of biomedical imaging has been around since the early 1900s, new advances in technology and computer science have greatly enhanced the diagnostic results of this discipline. RIT's Chester F. Carlson Center for Imaging Science's current research in the field includes efforts to enhance the development and application of multimodal image fusion and digital simulation. Additional research seeks to improve the capabilities of magnetic resonance imaging (MRI) systems.

With significant computer facilities and software capabilities, combined with a depth of experience, RIT provides an ideal location for conducting biomedical imaging research. "One of the Carlson Center's key goals is to promote initiatives that have real-world impact," notes Dr. Stefi Baum, professor of imaging science and center director. "Nowhere is this truer than in our work in biomedical imaging, where our research results could literally help save lives."

Multimodal Breast Imaging

MRI data

Benefits of Image Fusion:

The three orthogonal images on the left represent MRI data, while the three on the right are PET images that include the application of a red color table. The center images are the result of fusing the MR and PET images using the weighted average fusion plug-in. The fused images allow analysts to view both the anatomical structure and metabolic activity of the body part being imaged.

Click the image for a detailed view.

Currently, researchers in the Biomedical and Materials Multimodal Imaging Lab (BMMIL) are collaborating with Dr. Andrzej Krol of the State University of New York Upstate Medical University to advance positron emission tomography (PET), X-ray computed tomography (CT), and MRI capabilities. Dr. Maria Helguera, assistant professor of imaging science and director of BMMIL, is working to advance fusion techniques to enhance accuracy and efficiency for reading registered MR and PET breast images. While this research examines breast cancer, the technique could be applied to other ailments.

X-ray mammography is the primary tool for detecting and diagnosing breast cancer. Additional information can be provided by PET to show metabolic activity of the tissue, while MRI provides high-resolution structural information. Currently, these different imaging modalities are generated from separate equipment. To maximize the benefits both PET and MRI provide, the images need to be registered (brought into spatial alignment) and fused for evaluation by the radiologist.

The PET and MR images are registered using a finite element method technique that uses fiducial skin markers to provide common information visible in both modalities. However, even when viewing the registered images side by side, spatial relationships may be difficult to ascertain. By fusing the PET and MR images into a single viewable image, it is expected that radiologists will be able to diagnose with greater specificity and accuracy.

A number of fusion techniques have been developed that allow the radiologist to see the anatomical structure and metabolic activity in a single image volume. Helguera, in partnership with SUNY Upstate Medical University, tested the validity of fusion for visualization techniques; including a genetic algorithm-based technique to generate 2-D color tables, developed by Helguera's research team.

PET and MR images from study

Testing Fusion for Visualization Techniques:

Dr. Maria Helguera, in partnership with SUNY Upstate Medical School, completed a study to test the validity of fusion for visualization techniques. Click on the image for a detailed view.

The initial study demonstrates the need and benefit of a joint display because of the inaccuracy when using a side-by-side display. In addition, the study indicates that the color tables generated by the genetic algorithm are good choices for fusing PET and MR images. The advantage of the proposed method over other color table generation techniques is its ability to consider a much larger set of possible color tables, ensuring that the best technique is found.

"Multimodal breast imaging holds promise for better diagnosis; however, it is important to thoroughly test the fusion techniques in the clinical setting," Helguera says. "We were pleased to see our technique showed favorable results, but we continue to work on additional assessment and modification."

Medical Image Simulation

PET, CT, and MRI systems provide valuable diagnostic image data. However, the associated costs, resources, and clinical administration they require limit the ability to gather clinical data to conduct research and improve their diagnostic capabilities. By creating digital models of the human anatomy and modeling the medical physics, RIT researchers are able to study system design, acquisition protocols, and reconstruction techniques, and develop new image processing techniques with precise ground-truth.

Synthetic medical data sets can be used for training purposes or for quantitative evaluation in image processing and analysis algorithms. "Obtaining 'ground-truth' data in medical imaging is an almost impossible quest when pathology reports are not available," says Karl Baum, a post-doctoral fellow in imaging science and researcher with BMMIL. "One way to circumvent this limitation is by leveraging digital simulators that model the tissues being imaged, the imaging systems, and the image acquisition physics in order to create realistic synthetic images." Image simulation is not a new idea; however, until recently, practicality and accuracy has been limited due to the complexity of the procedures and long computation times. Improvements in computational systems and advancements in processor architecture have enabled high-resolution realistic 3-D synthetic medical data sets to be generated in less than five hours.

Synthetic multimodal PET/MR images

Synthetic Data Sets:

The image on the left is a synthetic multimodal PET/MR image of the brain showing a metabolically active lesion present in the PET image, created by fusing and registering the MR (middle) and PET (right) images and adding in color tables created by the research team. Synthetic data sets enable scientists to conduct research without the clinical adminstration and associated costs of obtaining actual PET/MR images.

Click on the image for a detailed view.

Improving Magnetic Resonance Imaging

MRI Coil

MRI Coil for the Extremities:

RIT Professor Joseph Hornak was part of a team that developed an asymmetric single-turn solenoid style MRI coil, as shown on the right. The device is especially suitable for MRI of the human wrist.

The creation of an MRI system with enhanced diagnostic power is also the goal of a Carlson Center research team lead by Dr. Joseph Hornak, director of the Magnetic Resonance Laboratory and RIT professor of imaging science and professor of chemistry. The Magnetic Resonance Laboratory at RIT is a research and development laboratory dedicated to solving real-world problems with magnetic resonance. The laboratory's efforts have focused on a variety of areas over the last two decades.

The lab develops novel resonators for MRI of extremities. A resonator is the antenna of an MRI system, sending out radio waves into the body and detecting the weak signal coming back. Hornak and his team patented a class of resonators called single-turn solenoids, which produce higher signal-to-noise ratio magnetic resonance images with less power. Discussions are underway with manufacturers to integrate this technology into the next generation of small, dedicated function MRI scanners. By incorporating their technology into unilateral MRI systems, the patient will no longer be surrounded by a magnet and a resonator.

In collaboration with MRI manufacturers, the laboratory develops specialty MRI phantoms, or anthropogenic objects within an MRI signal, to test the performance of an MRI system. Currently, Hornak's team is creating a resolution phantom for quantitative MRI of volume, which will enable the measurement of linearity and the point spread function at thousands of locations within its volume. This design is advantageous because the phantom does not need to be repositioned to collect measurements that cover the volume of interest or the three orthogonal imaging planes.

The laboratory is also recognized as a pioneer in the field of multispectral tissue classification and segmentation. Hornak's group developed an unsupervised MRI-based approach capable of segmenting the six healthy tissues in a magnetic resonance image of the brain. Such an approach has the potential to identify pathology as these tissues also have unique signals identifiable by the algorithm. In collaboration with the Institute for Biostructure and Bioimaging, National Research Council, in Naples, Italy, the team is developing a digital brain phantom to assess brain tissue segmentation algorithms.

Dr. Joseph Hornak

Enhancing MRI:

Dr. Joseph Hornak, director of the Magnetic Resonance Imaging Lab and professor of imaging science and chemistry at RIT, is working to develop MRI systems with improved diagnostic power. His efforts over the last two decades at RIT include the development of novel resonators for MRI of the extremities and investigating the impact of certain MRI contrast agents on the body.

Currently, Hornak is leading an effort to better understand the impact of certain MRI contrast agents on the body. Contrast agents are paramagnetic substances introduced into the body during some magnetic resonance scans to enhance the contrast between tissues and pathology. Recently, some of these contrast agents have been associated with a fibrosing disorder called nephrogenic systemic fibrosis (NSF), a painful and sometimes life-threatening condition typified by hardened sections of skin and stiffening of joints.

Hornak's team has discovered that non-cyclic contrast agents preferentially interact with two copper ions, while not interacting with other transition metal ions. This discovery may contribute to the understanding of the mechanism of NSF as it shows that beneficial copper ions can be removed from a tissue through this interaction and that gadolinium may be released. The team also discovered that the interaction in essence creates a new contrast agent with an enhanced ability to alter magnetic resonance contrast. A potential spin-off from this discovery is a targeted contrast agent for copper. Such a contrast agent could be used to study disease pathways associated with a high concentration of copper.

Through his work in the field, Hornak realized the need for basic MRI training and capitalized on the opportunity of delivering educational material over the Internet. His hypertext book, The Basics of MRI, is helping scientists around the world gain quicker entry into the field of MRI. The Basics of MRI is currently being accessed by an astonishing 15,000 people per month and has been translated into three languages, most recently Italian.

A Shared Benefit

The combination of education and research in the area of biomedical imaging has assisted the Carlson Center in gaining national notice for their efforts. However, center director Stefi Baum notes that "publicity is not the driving force behind our research efforts. It's about the development of better methods to diagnose disease, the creation of better and more cost-effective medical equipment, and the training of future generations of biomedical scientists to assist in making our communities healthier—and this is the true reward."

More information about this work and other related research is available online at www.cis.rit.edu/research/biomedical and at www.cis.rit.edu/people/faculty/hornak/mrl/.

Benefits of Image Fusion:

The three orthogonal images on the left represent MRI data, while the three on the right are PET images that include the application of a red color table. The center images are the result of fusing the MR and PET images using the weighted average fusion plug-in. The fused images allow analysts to view both the anatomical structure and metabolic activity of the body part being imaged.

Testing Fusion for Visualization Techniques:

Dr. Maria Helguera, in partnership with SUNY Upstate Medical School, completed a study to test the validity of fusion for visualization techniques. Color table examples produced by the genetic algorithms are shown in images (a) and (b) above. Joint PET/MR images created using the new color tables are shown in (c) and (d). The MR and PET images that were fused are shown in (e) and (f). Initial results demonstrate the need and benefit of a joint display and emphasize the benefits of the genetic algorithm fused images over other fusion techniques currently available.

Synthetic Data Sets:

The image on the left is a synthetic multimodal PET/MR image of the brain showing a metabolically active lesion present in the PET image, created by fusing and registering the MR (middle) and PET (right) images and adding in color tables created by the research team. Synthetic data sets enable scientists to conduct research without the clinical adminstration and associated costs of obtaining actual PET/MR images.