Advanced Cancer Imaging: Research and Clinical Practice
Advanced image methods at the Lombardi
Comprehensive Cancer Center are used to accomplish five tasks. Each is a focus of advanced clinical
care and research:
- Detect lesions that could be cancer
- Evaluate lesions that could be cancer to determine
whether they are benign or due to cancer
- Monitor the treatment of cancer
- Directly guide the removal of cancer
- Directly guide the treatment of cancer
Imaging Methods
Research Programs
The detection of cancer
Different types of cancer are detected in different ways:
Breast cancer is detected by clinical breast examination and mammography. We use state of the art mammography equipment at Lombardi. The center is
fully accredited by the American College of Radiology (ACR) for mammography, digital
mammography, breast ultrasound and stereotactic needle core biopsy. The research activities in
breast cancer detection include:
- Digital mammography and digitized film mammography
- Computer aided breast cancer detection
- Image enhancement by image processing
- A comparative trial of clinical breast examination, mammography, screening ultrasound and breast MRI for early cancer detection in women who carry BRCA-1 and BRCA-2 mutations
Prostate cancer is detected by PSA blood test, digital rectal examination and
prostate biopsy. The research activities in prostate cancer detection provide a model for
the testing of new strategies to improve the cancer detection rate in prostate biopsy.
Lung cancer is detected by two main methods: The chest radiograph and sputum
cytological examination. Research activities related to lung cancer detection are based on
the use of new technologies to improve chest radiograph quality, in the use of computer
analysis of chest radiographs for cancer detection, and in the use of chest computed
tomography for the early detection of lung cancer.
Lombardi is a participating site for the PLCO study, a multi-center trial of several cancer
detection methods, and for the National Lung Screening Trial, a multi-center trial of lung cancer detection methods.
Back to Top
The evaluation
of detected lesions that could be cancer
To detect early disease, doctors use tests that have a very high
likelihood of finding cancer. Most of these screening tests also find lesions that are not
cancer. Additional tests are then needed to determine whether or not cancer is present.
Advanced imaging is one of the ways doctors assess these suspicious lesions to determine whether or not
they truly are due to cancer.
Breast lesion evaluation - Standard methods for assessment are diagnostic
mammography and ultrasound. In some cases, magnetic resonance imaging (MRI) and nuclear
imaging are used.
At Lombardi, we are researching ways to evaluate breast lesions using imaging techniques. We are studying special research imaging methods in women scheduled for breast biopsy, including T-Scan Electrical Impedance Spectroscopy or a special transmission ultrasound system under development by Imperium LLC. Electrical impedance imaging of the breast is a new FDA-approved method for the evaluation of indeterminate breast lesions. Lombardi is a lead site for research of this technology.
Chest lesion evaluation - The standard method of evaluation of lesions detected
by chest radiography and/or sputum cytology is chest computed tomography (CT). This
procedure is performed many times each day at Lombardi.
New methods for the computer assessment of very small lesions seen on chest CT are
under development. Such small lesions are usually not due to cancer, but on occasion may
prove to be very small cancers. It is hoped that new computer technology will improve the
radiologist's ability to identify the small cancers as being different than benign lung
scars.
Back to Top
Monitoring the
effectiveness of cancer treatment
Cancer that is being treated is often monitored with advanced imaging methods. Different types of tumors require different methods of monitoring. Breast cancer is usually monitored by mammography and chest radiographs. Lung cancer is usually monitored by chest radiographs or CT. Liver cancer can be monitored with CT or MRI.
Cancer monitoring is used to document that the treatment is effective and to look for possible recurrence of the tumor.
There are two special research studies at Lombardi evaluating the response of locally advanced breast cancer. In these research studies, we are attempting to learn if we can predict the overall effectiveness of therapy after only the first course of therapy has been given. In the first study, we are investigating the changes in gene expression in the cancer as it is treated with chemotherapy. The tumors are monitored using ultrasound and by studies of the gene expression in tumor tissue removed with ultrasound guidance.
Second, we are part of a national study to determine whether breast MRI obtained after one course of chemotherapy is predictive of tumor response at the completion of therapy. The underlying goals of both of these studies is to, in the future, allow prediction of lack of response so that the types of chemotherapy used may be changed to ones that may be more effective.
Back to Top
Directly
guiding the removal of cancer
Advanced imaging methods are used to guide instruments to the site of cancer, allowing
for its removal with the smallest possible scar. Sometimes the tumor is removed in the
Radiology Department. Other times, advanced imaging is used to guide the surgeon in the
removal of the tumor.
Back to Top
Directly
guiding the treatment of cancer
Georgetown is performing advanced research in the use of image guidance for the
treatment of cancer metastases to the spine and in the non-surgical treatment of uterine
fibroids, a benign type of uterine tumor.
Back to Top
Imaging methods
In standard clinical practice the following methods are used:
Radiographs: Radiographs are images obtained using x-rays. Common types of
radiographs are chest radiographs, radiographs of the skeleton and mammography.
CT: CT is a method for obtaining images of thin sections of the body using
x-rays and computer analysis. CT is commonly used for the diagnosis of lung cancer, liver
cancer and lymphoma.
MRI: Magnetic resonance imaging is a method for obtaining images of thin
sections of the body. It is based on recording the electromagnetic signal.
Nuclear Imaging: In nuclear imaging, a very small amount of radioactive medicine
is injected - or, in some cases, inhaled - to image specific organs in the body. Several types
of devices can be used to record the location of the radioactive material, and diagnoses
can be based on the rate of accumulation of radioactivity and the anatomic location of the
uptake of the radioactivity. Nuclear imaging is commonly used for the detection or
diagnosis of thyroid cancer, liver cancer, breast cancer and bone cancer.
Positron emission tomography is a special form of nuclear imaging in which radioactive-
labeled metabolic chemicals are injected to look for areas of increased metabolism. The most commonly used agent is similar to glucose, a sugar, and shows where tissues are actively using glucose.
Ultrasound: In ultrasound (also called sonography), high frequency sound waves
are transmitted into the body. The reflected waves are recorded and reconstructed into
images of the body. Ultrasound in cancer evaluation is used most commonly for evaluating
breast lesions that might be cancer, in guiding prostate biopsies for suspected prostate
cancer and, less often, for liver and pancreatic cancer evaluation.
Back to Top
Research
The Lombardi Comprehensive Cancer Center and the Imaging Sciences and Information Systems Center of
the Department of Radiology have the following active research programs related to cancer.
Breast cancer
Breast Cancer Detection
- Digital Mammography/Digitized Film Mammography/Image Processing of Breast Images:
Research in this field started in 1991 and continues. FDA-approved digital mammography is available clinically in Lombardi's Ourisman Breast Health Center.
Digital mammography has undergone rapid development. Our tests with
a third-generation system indicate that, at this time, digital mammography is equivalent
to, but not better than, conventional mammography. Research in this field continues at
Lombardi, but we are not now accumulating new patient data. The following
articles and scientific report summarize much information in
this field:
Freedman M, Artz DS, Hogge J, Zuurbier RA, Mun SK. Digital Mammography. In
Siegel EL, Kolodner RM Filmless Radiography. Springer-Verlag. New York. 1999.
Lewin, JM, Hendrick RE, D'Orsi CJ, Moss LJ, Sisney GA, Karellas A
Clinical evaluation of a full field digital mammography prototype for
cancer detection in a screening setting. Radiology
209(P):238, 1998.
Freedman M, Steller Artz D, Hogge J, Zuurbier RA, Jafroudi H, Lo S-C Benedict,
Mun SK. An ROC study of screen film mammography and storage phosphor digital mammography:
Analysis of non-concordant classifications. Implications for the approval of digital
mammography systems. SPIE Medical Imaging 1997. 3036. 281-291.
- Computer-Aided Detection: Computer-aided detection for breast cancer has been an
area of intense investigation since 1991. This work has received continued funding from
the US Army breast cancer program since 1992. The Georgetown work is closely affiliated
with our research partners at the University of Michigan, the University of Iowa and
Brooke Army Medical Center.
In computer-aided diagnosis, a digitized conventional
mammogram is evaluated by specially designed computer neural network-based algorithms that
have been trained to detect micro-calcifications (a common finding of breast
cancer) and breast masses. Laboratory results have shown that radiologists using these
systems detect more suspicious areas than radiologists not using computer
assistance. We are in the process of completing for publication the results of our multicenter study of computer aided breast cancer detection. A technical publication in this
field by Lombardi researchers is:
Lo SCB, Freedman MT, Mun SK. Detection of mammographic masses using sorted
boundary features through an organized neural network. SPIE Medical Imaging 1998.
3338-124.
The computer can also be trained to detect changes in sequential mammograms and in
sequential breast MRI examinations. The methods for this are being developed at our
affiliated biomedical engineering research site at the Catholic University of America. Lombardi
researchers are providing clinical data, graduate student supervision and technical advice
for this project. Our recent technical publications on this study include:
Huang K, Xuan J, Varga J, Freedman MT, Szabo Z, Hayes D, Stearns V, Wang Y. MRI Image-Based Tissue Analysis and its Clinical Applications. Proc. of SPIE: Image Processing 2002. Vol. 4684: 424-432
Srikanchana R, Xuan J, Huang K, Freedman MT, Wang Y. Mixture of Principal
Axes Registration: A Neural Computation Approach. Proc. of SPIE:
Image Processing 2002. Vol. 4684: 923-932
- Image Process for Improved Mammographic Diagnosis: Image processing of mammograms
(both conventional and digital) has the ability to enhance the visibility of breast
cancer. This has been a research focus of our group since 1991. To date, the enhanced
images appear very good and interesting, but in our view are not yet sufficiently improved
to make detecting breast cancer easier. Work is continuing in this direction, both at Lombardi
and with an industrial partner. A publication summarizing the potential of this
field is:
Freedman M, Artz DS, Mun SK: Image processing in digital
mammography: the optimum image for each woman's breasts. pp. 1/1-1/3. IEEE Colloquium
on Digital Mammography. 1996. The Institution of Electrical Engineers, London.
Breast Lesion Diagnosis: Is it benign or malignant?
- Electrical Impedance Imaging of the Breast: The electrical
characteristics of breast cancer differ from the electrical characteristics
of normal breast glandular tissue and breast fatty tissue. It is
now possible to make recordings of these electrical signals, and
available data indicates that this method probably enhances the
ability to tell benign and malignant lesions apart in women younger
than 50. Additional research in this subject will begin shortly.
- Computer-Aided Classification of Breast Lesions: Once the computer has detected a
lesion on a mammogram, the computer neural network algorithm can be trained to provide
information about the likelihood that a detected lesion is cancerous. Preliminary
work on this started in 1993 at Georgetown. A recent technical publication on our work in this field is:
Kinnard L, Lo SCB, Wang P, Freedman MT, Chouikha M. Separation
of Malignant and Benign Masses using Maximum-Likelihood Modeling
and Neural Networks. Proc. of SPIE: Image Processing 2002. Vol.
4684: 733-741
- Assessment of Breast Tissue Response to Breast Cancer Preventive Agents: Several
medications are currently available and others are under development for use in the
prevention of cancer. For several years the Lombardi imaging group has been interested in
techniques that could be used to monitor the response of tissue to these agents. These
could serve as methods of showing that a selected medicine was having a desired effect on the
tissue, and preventing the development of cancer.
In one project in this area, the sequential mammograms of women who took a drug
with the potential to prevent breast cancer were converted into
digital form and then measured for changes in breast density. We found that women who received estrogen showed an increase
in breast density, while those who received Raloxifene decreased their breast density during follow-up to a degree similar to that of a placebo.
A related publication is:
Freedman MT, San Martin J, O'Gorman J, Eckert S, Lippman ME, Lo SCB, Walls EL. Digitized mammography in postmenopausal women receiving raloxifene or estrogen in a two-year placebo-controlled, randomized clinical trial. Journal National Cancer Institute 2001; 93:51-56.
In a second study, which is about to start, imaging methods will be compared to tissue biopsy measurements of biomarkers in women who are receiving a breast cancer prevention medication.
Lung Cancer
Researchers currently associated with the Lombardi have been working in the early detection
of lung cancer by radiographs since 1974. The application of computer methods to lung
cancer detection research started in 1991. Current research focuses on improved lesion
detection through digital image processing and computer aided diagnosis.
- Digital Radiography and Image Processing: In 1991, Lombardi researchers began work
with Agfa corporate and US Army funding to improve digital chest radiography through image
processing enhancements. The group has published many articles and chapters, including:
Freedman M. Digital Chest Radiography. pp 315-348. In Boiselle PM and White CS. New Techniques in Thoracic Imaging. New York: Marcel Dekker, Inc. 2001
Freedman M and Strickland NH. Digital Radiography and PACS: Image Processing in Computed Radiography. In Grainger and Allison's Diagnostic Radiology: A Textbook of Medical Imaging, Third Edition. RG Grainger and D Allison, eds. Churchill Livingstone, New York. 1997:27-34.
Freedman M, Artz DS. Digital Radiography. In Siegel EL, Kolodner RM. Filmless Radiography. Springer-Verlag, New York. 1999
Freedman M, Artz DS: Digital Radiography of the Chest. Seminars
in Roentgenology 1997. Guest Ed. Matthew Freedman. XXXII, 1:38-44.
- Computer-Aided Diagnosis: Computer-aided diagnosis is defined as the use of the
computer to process digitized chest radiographs or chest computed tomography to focus the
radiologist's attention to regions that may contain lung cancer. Lombardi has been a major
research site for this effort. This work has been supported
by Agfa Corporation, the American Cancer Society and Deus Technologies Inc.
One computer-aided diagnosis device, based on digitized chest radiographs, has recently received FDA Pre-Market Approval based in part on a randomized, retrospective clinical trial managed by Lombardi.
For this trial, 80 selected cancer cases that met the inclusion criteria for the study were evaluated in several statistical tests using fifteen radiologists. The study showed that, on average, the radiologists detected 14% more of the smallest cancers (9-14 mm in size) when using the system than when not using the system.
Radiologists with computer support and a high level of suspicion that cancer was present detected 43% of cancers that had been missed originally by two radiologists.
Chest computed tomography is a more expensive, but also a more sensitive method for the
detection of early lung cancer. The major problem with chest CT, apart from its expense, is the large number of small
nodules that it detects that are not cancer. Lombardi has an active program in computer aided detection of lung cancer on chest CT and participates in the National Cancer Institute funded National Lung Screening Study: a comparison of the chest radiograph with the chest CT.
We are working to develop computer methods that might better tell apart small nodules that are cancer from those that are not cancer.
Energy Subtraction Radiography: Energy subtraction radiography is a special type
of digital radiography in which a radiography of the chest can be separated by computer
processing into three different images: one that looks like a standard chest radiograph, a
second image that shows the lungs with the bones subtracted from the image, and a third
that shows mainly calcified structures. The system improves the ability to seen small lung
nodules and early pneumonia. In its current design, however, this approach significantly
increases the cost of obtaining chest radiographs and the added expense is not reimbursed
in clinical operations. Research at Lombardi has focused on improving the image processing for
this system. This work has been supported by the manufacturer, Fuji Photo Film
Corporation.
A project on computer aided detection of lung cancer on energy
subtraction radiographs is expected to start in the spring of
2003.
- Image guided robotic lung biopsy: We recently were awarded a grant from the National Cancer Institute
to extend our current work in image guided robotic therapy of spinal disease toward
the development of image guided robotic lung biopsy. The placement of a needle into a lung nodule,
when the nodule is small, is quite difficult. As we expand the use of chest radiograph and CT screening for lung cancer,
there will be an increasing number of very small nodules found. This project is to develop robotic techniques
so that the biopsy of these small lung nodules becomes easier and will have less morbidity.
Prostate cancer
Advanced computer methods have been developed for special research analyses of prostate
cancer. This work has been performed in conjunction with researchers at the Catholic
University of America, the Armed Forces Institute of Pathology, the Department of Urology
at Walter Reed Army Medical Center and the Department of Urology at Georgetown University.
Cancer treatment
Research in advanced imaging methods for improving cancer treatment is currently
underway. The current focus is the treatment of cancer metastatic to the spine. The
project is supported by the US Army. Currently, once cancer has metastasized to the spine,
the degree of patient disability increases substantially.
There are two methods under active study at Georgetown. One of these is the use of
advanced imaging methods to improve the open surgical treatment of tumors by improving the
ability to define, intra-operatively, the extent of tumor that still remains in bone and
which requires further surgery. In this project, intra-operative CT is used in the
Neurosurgical Operating Room. Eventually, 3D image fusion methods will be used to register
the preoperative MRI with the intra-operative CT. Since the location of the tumor is best
seen on the MRI and the bony landmarks are best seen on CT, this method will provide
improved information to the operating surgeon. We are testing whether or not this method
improves long-term outcome. This work is being done in conjunction with the Department of
Neurosurgery at Georgetown.
The second method is percutaneous vertebroplasty. In this method, image guidance is
used to place needles into vertebral bodies that have started to fracture or are at great
risk of fracture. A plastic material (Polymethylmethacrylate) is then injected into the
vertebral body to reinforce it to prevent further fracture. By increasing the strength of
the vertebra, it is expected that major complications of vertebral fractures may be
avoided. Currently, the method is being used in selected patients with metastatic cancer,
myeloma and for patients with unremitting pain following osteoporotic spine compression
fractures.
Finally, methods for the use of radio frequency probes for the ablation of metastatic cancer are
currently under development. Treatment for selected benign tumors of the spine and
skeleton has recently been introduced by the Georgetown Radiology Department, and is
available for routine clinical care.
Back to Top
