Ontario Centre of Excellence in Breast Cancer Imaging Research
Vision
This project brings together a strong, interdisciplinary group of scientists dedicated to solving key problems associated with breast cancer. Our research will result in new methods for earlier detection, more accurate diagnosis, improved accuracy of biopsy, determination of the true extent of disease to avoid repeat surgery. In addition, we will develop algorithms to predict breast cancer risk by computer analysis of digital images. The program will train leaders in the science and technology of medical imaging and produce valuable intellectual property that will be exploited by existing and newly-created companies in Ontario or licensed to others.
Project Description
This ORDCF project is for the purpose of strengthening the pre-eminent position of Ontario researchers based at Sunnybrook and Women's College Health Sciences Centre (S&W), University of Western Ontario (UWO) and Carleton University (CU) in multidisciplinary breast cancer research, with imaging-related projects as a central theme. The projects address important problems in detection, diagnosis and therapy for breast cancer as well as approaches to prognosis. The projects described provide a mechanism for training young clinical and basic scientists in the high technology imaging areas that will be important in the future of medicine. Our vision is that the partnership between S&W, ORDCF and our industrial and academic colleagues at collaborating institutions will allow us to hire and train young scientists, while at the same time helping to realise economic benefits from our ideas and inventions.
One of our main motivations for seeking support from ORDCF is to accelerate our progress in the fast-moving field of breast cancer imaging, particularly in those areas where our competitive edge is currently threatened by lack of adequate resources - primarily human resources. While the Ontario Centre of Excellence in Breast Cancer Imaging Research (OCEBCIR) is being formed to strengthen an existing nucleus of research and development, we will also provide leadership and resources in collaborations with other researchers in Ontario who will benefit in their work from access to techniques and equipment available at OCEBCIR, or who simply lack an adequate "critical mass" of faculty. At the same time, this will strengthen our ability to conduct further studies of greater scope and will allow migration of techniques to other sites.
As our comprehensive breast cancer centre grows to incorporate screening and novel therapies, we plan to work closely with our strong group of clinicians and cancer biologists to exploit imaging in many different aspects including detection, diagnosis, staging, therapy, as well as research to understand the etiology of the disease and ultimately provide strategies for its prevention. At S&W, there are excellent programs in angiogenesis related to breast cancer (Dr. Robert Kerbel) and markers for malignancy (cyclin dependent kinases, Dr. Joyce Slingerland) where cross fertilisation will be valuable, and, through OCEBCIR, we will exploit these linkages.
The projects are organised along the themes of breast cancer detection, diagnosis, therapy and prognosis. In many cases, however, a particular project will have impact upon more than one area.
- Design of optimal image display systems for digital mammography
Our group has contributed to the development of digital mammography through basic laboratory research followed by interaction with industry through the International Digital Mammography Development Group (IDMDG), initially led by Dr. Donald Plewes. There are still significant problems to be solved in digital mammography. Many of these relate to how the images should best be displayed and manipulated to improve the accuracy of mammography, while allowing the physician to work efficiently. Our unique experience will allow us to design the optimal workstation for digital mammography by using artificial intelligence to analyze digital mammograms and optimize parameters for image display. We have recently formed working relationships with GE Medical Systems to investigate optimal display strategies using their state-of-the art digital mammography platform. Similarly, we are working with CADx Medical Systems to exploit the potential of computer-aided detection for optimizing image display. Our ability to have close interaction between radiologists and software design engineers will facilitate relevant and efficient solutions. This work will allow us to attract and train both engineers and computer scientists in the important area of computer vision while training radiologists in the use of digital mammography for detection and diagnosis.
- Next-generation x-ray detector technology for digital mammography
Our x-ray detector group has been active since the early 1970s, investigating electrostatic imaging methods in radiology. This group is an international leader in research on detectors for digital imaging and members consult widely to industry in North America, Europe and Japan. We have established one of the strongest laboratories in the world for design, integration and evaluation of digital x-ray detectors. Although initial systems for digital mammography employed detectors based on light-emitting x-ray phosphors, it is more efficient to convert absorbed x-ray energy directly into electric charge, which can be digitized. We will continue our development of these direct conversion detectors in collaboration with our industrial partners, working on leading-edge detector materials such as amorphous selenium, lead iodide and cadmium zinc telluride and readout methods based on active matrix arrays and liquid crystal light valves. We will explore the possibility of manufacturing the sophisticated detector readouts in Ontario with our industrial collaborators DALSA and Westaim. To guide detector developments, we will work closely with Dr. Ian Cunningham of The University of Western Ontario, an international expert in signal/noise transfer analysis in digital imaging systems. He will extend the development of "stochastic" linear-systems theory, so that it can be used to accurately describe image noise in the new generation of flat-panel and scanning digital-imaging systems. The result of our research will be a generalized resource "tool-kit" of transfer relationships that can be used by system designers to ensure all new digital imaging systems provide the best possible image quality.
- Quality control for digital mammography
The critical importance of quality control (QC) procedures for breast imaging is now widely recognized, e.g. in the Mammography Quality Standards legislation in the U.S. As digital mammography enters clinical use, it will be necessary to have in place a viable system of QC procedures. Working closely with GE Medical Systems (GEMS), we plan to continue our initial work with the IDMDG on the development and evaluation of test objects and software tools for QC in digital mammography, particularly for an automatic system. We will refine methods for QC on soft copy display devices and, working with Kodak, will create a similar system for high-resolution laser hard copy displays. The intellectual property arising from this work will include software packages as well as test tools that can be licensed to vendors of digital imaging systems.
- Imaging of women at high-risk for breast cancer
MRI offers new capabilities to enhance our ability to diagnose and treat breast cancers with potential impact of cost-effective breast cancer management, reduced morbidity and improved outcome. We are developing MRI imaging software and instrumentation to provide rapid imaging of breast tumours based on the uptake of Gd DTPA contrast medium. We will work with new faculty to develop new MRI pulse sequences, magnet modifications, receiver coil designs and image reconstruction techniques. Of prime interest is screening women at high-risk for breast cancer based on genetic factors. Our preliminary studies have shown this to be feasible, accurate and cost-effective and we believe that breast MRI will become the standard of clinical practice for these women. Application to other breast imaging problems such as tumour staging and response to therapy is also possible and will be explored.
- Novel breast MR contrast agents based on hyperpolarized xenon
We believe that increased tumour vascularity can be imaged with very high sensitivity using hyperpolarized xenon (H-Xe). H-Xe, produced using optical pumping methods, will generate an MR signal up to four orders of magnitude larger than that obtained conventionally. This permits imaging of very small quantities of dissolved xenon with high spatial and/or temporal resolution. To our knowledge, H-Xe has not yet been used to investigate breast cancer. Enhanced accumulation of H-Xe in breast lesions compared to surrounding tissue will enable direct measurement of blood flow uncomplicated by compartmental effects, exchange, extraction fraction or complications due to a background of non-vascular signal. H-Xe may therefore be more sensitive to subtle increases in blood flow and capillary permeability allowing earlier detection of malignancy, providing improved discrimination between benign and malignant lesions. Unlike conventional MR imaging, H-Xe image SNR is independent of the magnitude and homogeneity of the static magnetic field permitting ultra-low field MR imaging. This suggests the exciting possibility of a low cost MR imaging system based on H-Xe, which would be particularly welcome for breast cancer screening. The commercial potential of H-Xe has already been recognized by Nycomed-Amersham, with whom we are currently negotiating a research agreement (please see letter). Using measurements of relaxation times and chemical shifts we will develop novel injectable contrast agents containing H-Xe based on lipid and liposome suspensions and perfluorocarbon emulsions. We will establish the feasibility of H-Xe imaging of breast cancer using an implanted tumour model in rats. This work will also be extending to ultra-low field imaging using a prototype electromagnet operating at 80 gauss.
- Optimization of contrast-enhanced digital angiography
As an alternative to MRI, it is possible to perform rapid imaging of contrast medium using digital mammography. We believe that spatial and temporal signal processing techniques can be developed to improve the signal-to-noise ratio of the images and therefore the sensitivity to low concentrations of iodine that will identify the location and extent of disease. Preliminary experiments with our technique have indicated selective uptake in the breast; however, the results clearly indicate that it is necessary to optimize the digital mammography technique further with respect to the injection protocol, the x-ray spectrum employed, image registration and image processing. We propose to perform such optimization, leading to a technique potentially providing the excellent results that we have obtained with MRI, but at lower cost and with full access by patients in centres where MRI is not available. We will compare the relative performance of the two techniques in phantoms and patient studies.
- MRI, ultrasound, and optical techniques to improve the accuracy of lesion characterization
To provide fast and accurate diagnosis of suspicious tumours immediately after detection, we are exploring the use of optical, ultrasound, and MRI viscoelastic (stiffness) assessment. Tissue viscoelastic properties have been shown to be related to tumour grade; therefore we are developing MRI motion techniques for their measurement. We are exploring in-situ optical and ultrasound assessment of detected lesions. The optical approaches will use fibre optic sensors based on Raman and reflectance spectroscopy techniques. Previous studies have shown these techniques have considerable capability to differentiate tumour grade and malignant potential. The ultrasound approaches will use miniature, needle-based ultrasound imaging techniques developed at S&W. Approaches will be designed to permit detailed simultaneous real-time visualization of structure and blood flow at the needle tip. The recent development of high frequency Doppler will allow us to test the hypothesis that the hemodynamic and morphological properties of tumour vasculature can be linked to tumour grade and ultimately to prognosis. We will work with VisualSonics Inc., a spin-off company based on technology developed at S&W.
- Use of new ultrasound imaging methods for characterizing breast lesions
Ultrasound is widely used in the diagnosis of soft tissues because of relatively low cost and ability to give real-time information non-invasively in the absence of ionizing radiation. This project seeks to provide functional information on cancer progression by locating and imaging the microscopic "angiogenic" blood vessels that accompany a developing tumour. Knowledge of these new, abnormal vessels should allow us to differentiate malignant from benign disease, monitor a number of anti-angiogenic therapies, including thermal therapy, and follow the effects of tumour therapy to ensure that vascularization has been suppressed.
The ultrasound method to be used,"pulse inversion Doppler imaging" was developed and patented at S&W. It detects microvessels by means of the unique scattering properties of very small, encapsulated gas bubbles injected into a peripheral vein. The bubbles are stimulated into nonlinear oscillation by the sound field of the imaging transducer, and the harmonic components of their echoes allow them to be identified in the midst of echoes from surrounding tissue. Drugs that comprise such microbubbles, developed as ultrasound contrast agents, have recently been approved for use in Canada. In collaboration with imaging clinicians at the University of Toronto hospitals, we have already demonstrated the method's ability to detect tumour blood flow in the liver. The present project will extend this method to the breast by using proprietary software developed with our industrial partners GEMS and ATL Ultrasound, together with microbubble agents developed by DuPont Pharmaceuticals Inc.
- Methods for more accurate establishment of tumour margins (DCIS)
In a small but growing proportion of breast cancer involving ductal carcinoma in-situ (DCIS), tumours do not present clear evidence of their margins during surgery so that resected tissue may not encompass the entire tumour bed. Re-excision is required in 20-30% of these cases. Fortunately, these tumours are well resolved by MRI. We are developing surgical guidance and planning methods based on MRI data. With colleagues at University Health Network/Ontario Cancer Institute, we will carry out pre-clinical exploration of the use of optical tracers linked to monoclonal antibodies as tumour-specific probes. These agents have exciting potential to aid in tumour detection and as markers for tumour margins and represent commercial potential in the biotechnology area. In conjunction with MRI, surgical techniques can be substantially improved, thereby reducing both the frequency of repeated surgery and breast cancer recurrence rates. If successful, we will train surgeons and pathologists in these new methods.
- Development of techniques in computer-assisted 3-D pathology
To achieve accuracy in diagnosis and planning of therapy, it is essential to develop a system for the precise 3-D correlation of imaging data and pathology, where the correlation encompasses the entire lesion and its margins. Most pathology processing is geared to tissue fragments no greater than 2.5 cm in maximum dimension, often precluding the assessment of the entire lesion in breast specimens. Sections of whole tissue slices are ideal but these are usually between 3 and 7 cm in maximum dimension and up to 15 cm in whole breast slices. This requires a specific processor and the use of a giant, motorized microtome to cut the sections. Furthermore, because of the sheer amount of information to be viewed and analyzed, this task is formidable using manual approaches. It is our plan to create a system for digital acquisition of the enormous amount of data emanating from these large sections and to employ image display techniques to allow visualization of the 3-D histological properties of excised breast tissue. In addition, we will employ computer image feature analysis and segmentation techniques to facilitate or automate the diagnosis of DCIS over these large volumes. This project offers an excellent opportunity for training of pathologists in these leading edge methods and also for transferring relevant artificial intelligence techniques developed for digital mammography, while familiarizing imaging researchers with issues faced in pathological diagnosis, thereby stimulating further needed developments.
- Quantitative assessment of breast cancer risk from digital x-ray, ultrasound and MR images
Our research is based on previous work in which the radiological feature of extensive radio-dense tissue in the breast on mammography ("mammographic densities"), has been found to be strongly associated with risk of breast cancer. We have demonstrated that there is a dose-response relationship, i.e. risk increases with increasing density. This suggests that factors responsible for mammographic densities could also be responsible for causing a substantial fraction of breast cancer. Furthermore, in contrast to other risk factors, mammographic densities can be changed by hormonal or dietary interventions. The identification of factors that change density may lead to a mechanism for reducing the incidence of breast cancer.
Our research will increase the understanding of this important risk factor using a multi-disciplinary approach to develop improved methods of measuring breast tissue characteristics, through analysis of mammograms, ultrasound and MRI images of the breast. The techniques that we have developed thus far are promising, but major improvements are needed. These include: 1) extension of the mammographic analysis from 2-D to 3-D; 2) development of the hardware and software for ultrasound analysis; and 3) for the MR technique, semi-automated measurement of the complete breast volume, better accuracy in determining water content, and robustness for different types of MR imagers, sizes of subjects, and designs of breast coils.
Our studies will include comparisons of breast tissue in women born in China who have migrated to Canada with that of Caucasian women, an examination of the breast tissue characteristics of young women to illuminate the natural history of breast density, and research into the identification of the genes associated with high-risk breast tissue.