Breast cancer is not simple to diagnose or characterize. UCAIR has therefore put together a body of specialists, the Breast Cancer Imaging Research Committee, to aid in identification of the most effective imaging procedures:
|Saundra Buys, MD||Oncology|
|Kathryn Morton, MD||Radiology|
|Matthias Schabel, PhD||UCAIR|
|Leigh Neumayer, MD||Surgery|
|Ed Nelson, MD||Surgery|
|Phillip Bernard, MD||Pathology|
|Anne Kennedy, MD||Radiology|
|Glen Morrell, MD, PhD||Radiology|
Each contributes their expertise to our research projects, whether it is in the analysis of the data or making patients aware of opportunities to volunteer for research imaging procedures. Here are some of the projects currently in progress.
Improving MRI for Early Detection of Breast Cancer - Dynamic Contrast Enhancement
Matthias Schabel, Ph.D.
X-ray mammography is the current, clinical Gold Standard for the detection of breast cancer. It is a well understood and standardized procedure, it works fairly well in postmenopausal women and it is inexpensive. However, X-ray mammography works less well in premenopausal women, it is not very sensitive (i.e., a significant number of cancers are missed) and cumulative X-ray exposures can cause cancers.
MRI, or Magnetic Resonance Imaging, offers a better alternative. It is very sensitive (few cancers are missed), it works well in both pre- and postmenopausal women and no radiation exposure is involved, so there are no significant health risks. Yet MRI also has drawbacks: Specificity is too low (too many false positives that lead to unnecessary biopsies), interpretation is complex and not standardized and it is expensive.
Because MRI mammography has not been fully developed it is currently recommended only for screening of high-risk women - carriers of breast cancer susceptibility genes or those with a strong family history of breast and/or ovarian cancers.
The OBJECT of this project is to develop MRI imaging procedures that have both high sensitivity and specificity and thereby may be applicable to a wider patient population.
MRI is sometimes used in conjunction with a "contrast agent". This is, in effect, a dye that makes an object easier to see in the MRI image. The contrast agent is injected into a vein in the patient's arm and accumulates in a tissue as the blood circulates through it; as the contrast agent accumulates, the image of the tissue in question is "enhanced".
The rate at which a tissue enhances and the manner in which the contrast agent distributes in the tissue over time may be characteristic of a particular tissue. Breast tumors enhance more than the associated normal tissue. However, benign breast lesions also enhance.
So while this goes part way to identification of a breast tumor, additional characteristics must be considered.
In this project we aim to implement and validate efficient computer algorithms (computer programs that manipulate and interpret the MRI data) to integrate MRI data and give an estimate of the likelihood that a lesion is cancerous. The parameters we can currently measure with contrast agent enhanced MR imaging are:
- the rate of tissue enhancement
- blood vessel density
- tissue leakiness (the rate and extent to which contrast agent redistributes to surrounding tissues)
- tissue swelling
Malignant breast tumors have a characteristic blood vessel density, leakier blood vessels than benign lesions and exhibit more swelling. While none of these characteristics is diagnostic by itself, when measurements of these parameters are combined, areas in an image of the breast can be highlighted (in RED) which are highly likely to be malignant tissue.
Below are anatomical images of the breast (same as above). Move the cursor over the figure to see a computer generated image of the tissues classified (and colored) according to an analysis based on MRI data for blood vessel density, blood vessel leakage and tissue swelling
Note that combining the different forms of information provided by the analysis of contrast-enhancement data, the computer algorithm is now able to distinguish between the malignant tumor (RED - confirmed by biopsy in this instance) and the benign lesion (GREEN).
The image may also be represented in 3-D and rotated with the tumor seen in situ.
Further improvement is required before this technique can be applied in the clinic. In particular we require: Improved speed and resolution of the MRI scans, better data quality, and the development of a better understanding of how measurements can distinguish benign lesions from cancer. This work continues
Faster MRI Techniqes for Breast Imaging
Glen Morrell, MD, PhD
The MR imaging technique we describe above requires that the movement of contrast agent into breast tissue is followed over time. Multiple images of the breast must therefore be made, as if one were making a movie. In a movie, the more frames per second that can be projected, the clearer the action. In MR imaging the rate at which images can be made is currently limited by how long it takes to make each individual image. One object of this project is to increase the speed of image acquisition.
The technique being developed to increase image acquisition rate is called Selective Excitation. In a normal MRI scan the body is imaged in slices. To see a small object in the center of the body, the whole slice that includes that object must be imaged. That is, image data must be acquired for the entire slice, even though the majority of the data are not useful.
To speed up acquisition of the object in question, Selective Excitation uses a modified magnetic field to excite only the region of the body of interest. The amount of data that must be acquired is thus reduced and the time required to acquire the data is also reduced.
For example (see images below), instead of imaging the whole breast (left image), only the region of interest, perhaps a suspicious lesion, might be imaged (right image).
New Techniques for Even Faster MRI
Edward V. R. DiBella, Ph.D.
In the previous project of Dr. Morrell's, a technique is used to reduce the amount of data acquired. Another approach to faster imaging is to produce the same images from less data. The method described here is currently being developed for imaging the beating heart; however, it will be equally applicable to imaging the accumulation of contrast agent in the breast.
Model-based Reconstruction is a data processing technique that produces a high quality image of a dynamic (time-dependent) process with less data - as little as a quarter of the data normally required for image production. The figure shows an example of one time-frame from a cardiac study where a similar image is obtained with just a fraction of the data.
A Magnetic Resonance Imager acquires data in a form that is not directly interpretable by visual inspection. This data format, called k-space, must be processed ("reconstructed") to give an image we recognize. To reconstruct means to transform into an image. The processing is very fast. It is like following a cook book recipe and the image appears essentially right away on the MR console. This figure shows an example of what k-space looks like.
In Model-based Reconstruction the computer takes only a sample of the K-Space data (as little as a quarter).
Conventional reconstruction techniques could not produce an image from such sampled data, while Model-based Reconstruction produces an image almost indistinguishable from that derived with a complete data set.
To accomplish this, the reconstruction algorithm recognizes that events that occur over time, such as the accumulation of contrast agent in breast tissue, follow a predictable path and then reconstructs the images accordingly. This is a very exciting new method for improving dynamic MRI.