When
a woman undergoes breast cancer screening, she first must get a
mammogram and an ultrasound of her breasts. If the results from these
tests are inconclusive, which is often the case for women with dense
or scarred breast tissue, she then is told to get an MRI. MRIs are
expensive tests that insurance companies do not want to pay for
unless absolutely necessary. Other radiographic imaging tests are not
viable alternatives to MRI because increased exposure to radiation is
not desirable in patients who may have cancer. Ideally, the initial
steps of breast cancer screening should be improved so that a woman
can be definitively told that she does not have breast cancer at the
time of her screening or that she absolutely needs an MRI scan.
The ultimate goal of this research project is the fusion of mammogram and ultrasound imaging data to improve the screening process and detection rate for breast cancer. While both imaging modalities provide images of breast tissue, they differ in several ways. In mammography, the breast is compressed perpendicularly to the torso between two plates. Low energy X-rays pass through the breast and are recorded on a film plate, resulting in a 2-dimensional image. Breast ultrasounds are recorded with the breast compressed into the torso by a single screen. The ultrasound data is stored as a 3-dimensional stack of 2-dimensional images. Each modality also records different information about the tissue. By combining the data from both imaging types, a better information about the breast tissue can be obtained without the need for any extra tests.
Mammogram and ultrasound images have already been taken of a breast phantom. An algorithm is being implemented to generate a 2D projection of a 3D ultrasound volume. Once this is complete, optimization techniques will be used to map the projected ultrasound images to the mammogram images. This will take into account the different types of deformations in each of the imaging modalities. As mentioned above, the resulting registered images will provide more information about the tissue than either of the imaging modalities could provide separately.
The ultimate goal of this research project is the fusion of mammogram and ultrasound imaging data to improve the screening process and detection rate for breast cancer. While both imaging modalities provide images of breast tissue, they differ in several ways. In mammography, the breast is compressed perpendicularly to the torso between two plates. Low energy X-rays pass through the breast and are recorded on a film plate, resulting in a 2-dimensional image. Breast ultrasounds are recorded with the breast compressed into the torso by a single screen. The ultrasound data is stored as a 3-dimensional stack of 2-dimensional images. Each modality also records different information about the tissue. By combining the data from both imaging types, a better information about the breast tissue can be obtained without the need for any extra tests.
Mammogram and ultrasound images have already been taken of a breast phantom. An algorithm is being implemented to generate a 2D projection of a 3D ultrasound volume. Once this is complete, optimization techniques will be used to map the projected ultrasound images to the mammogram images. This will take into account the different types of deformations in each of the imaging modalities. As mentioned above, the resulting registered images will provide more information about the tissue than either of the imaging modalities could provide separately.