Daniel C. Sullivan, M.D., Associate Director

The Cancer Imaging Program (CIP) promotes and supports cancer-related basic, translational, and clinical research in imaging sciences and technology. The goal is to assure the integration and application of these imaging discoveries and developments to understanding cancer biology and the clinical management of cancer and cancer risk.

GRANTS

Magnetic Resonance Spectroscopy Imaging for Prostate Cancer Evaluation
NCI-supported principal investigators at the University of California-San Francisco (UCSF), Drs. John Kurhanewicz, Sarah Nelson, and Daniel Vigneron are at the forefront of development of magnetic resonance spectroscopy imaging (MRSI). MRSI uses the same device used by the more widely known magnetic resonance imaging (MRI). MRI gives physicians information about a lesion's size and location but cannot determine whether the lesion is benign or malignant. The UCSF groups developed methods for extracting more information, called chemical spectra, during an MR scan, allowing them to more accurately determine which lesions are malignant, discriminate cancer from surrounding healthy tissue, and predict response to therapy.

This work led to the clinical implementation of MRSI at UCSF, primarily to diagnose prostate and brain cancers, and plan and monitor treatment in those patients. About 3600 patients have undergone combined MRI/MRSI staging exams for prostate cancer, a procedure that takes less than one hour per patient and greatly improves their clinical management. For example, in some patients who normally would be recommended for watchful waiting, MRSI has found aggressive-appearing cancer. In patients with rising prostate-specific antigen and negative transrectal ultrasound biopsies, MRSI has been used to direct biopsies to obtain positive results. In patients with suspected recurrent prostate cancer, MRSI can discriminate residual cancer or recurrence from normal and necrotic tissue. At least 500 MRI/MRSI examinations have aided diagnosis and treatment of patients with suspected and confirmed brain cancer.

Because the software required for the combined MRSI/MRI examinations can be implemented on most standard MRI scanners, the UCSF investigators have worked with industrial and academic collaborators to make the technology widely available. This technique is now ready to move from its development site into multi-institutional trials. Continued technical advances and more clinical implementations of the combined MRSI/MRI examinations will lead to better cancer diagnosis and treatment. The American College of Radiology Imaging Network is planning a multi-institutional trial of MRSI in prostate cancer management.

Computer-Aided Diagnosis for Lung Cancer Screening
Lung cancer is the leading cause of cancer-related deaths in the United States. Evidence that early detection and therapeutic intervention produces a more favorable patient prognosis has prompted the creation of lung cancer screening programs. Radiologists have shown that the sensitivity of helical (spiral) computed tomography (CT) to detect potential lung cancers (nodules) surpasses that of chest radiography. The efficacy of low-dose spiral CT protocols has created interest in and demand for lung cancer screening. The potential mortality benefit, however, of any lung-cancer screening program is a topic of debate in the medical community.

Identifying small lung nodules in CT scans is difficult because lung blood vessels are prominent on CT images. The large amount of image data acquired during a single CT examination may quickly lead to information overload for interpreting radiologists. Computer-aided diagnosis (CAD) for nodule detection is increasingly used as a second reader to reduce the variability of observers and increase sensitivity. Various investigators have developed methods for automating lung-nodule detection in CT scans, including geometric modeling, fuzzy clustering, spatial filtering, and gray-level thresholding.

NCI has funded CAD work by Drs. Maryellen Giger and Kunio Doi at the University of Chicago for many years. In a recent Radiology article, they reported results on automated lung nodule cancer detection missed in CT screening. A computer detection method that used gray-level thresholding techniques to identify 3-dimensionally contiguous structures in the lungs was applied to the CT data. Computer-extracted volume data were used to determine whether a structure was a nodule candidate. A rule-based scheme and a cascaded automated classifier were applied to the nodule candidates to distinguish nodules from areas of normal anatomy. Overall performance of the computer detection method was evaluated with free-response receiver operating characteristic analysis. The computer method correctly detected 84% of cancers missed during visual interpretation.

References 1. Armato SG 3rd, Altman MB, La Riviere PJ. Automated detection of lung nodules in CT scans: effect of image reconstruction algorithm. Med Phys 2003 Mar;30(3):461-72 2. Armato SG 3rd, Li F, Giger ML, MacMahon H, Sone S, Doi K. Lung cancer: performance of automated lung nodule detection applied to cancers missed in a CT screening program. Radiology 2002 Dec;225(3):685-92

Optical Tomography and Optical Contrast Agents
Much functional imaging of cancer using MRI or positron emission tomography uses contrast agents that provide information about tumor vasculature or metabolism. Similar contrast agents are being developed for optical imaging based on photon reflection, absorption, or fluorescence, that will expand potential applications. The ease of use and non-invasive nature of optical imaging make it a potentially attractive, low-cost method for functional diagnosis. Over the past few years, significant progress has been made in understanding the fundamental nature of light propagation in tissue. These discoveries have led to advanced techniques for tracking light photons through tissue over distances relevant to breast imaging. These advances seem to provide a basis for a new class of optical imaging methods that will substantially improve use of optical data to determine breast cancer risk, and to detect and diagnose breast cancer.

Dr. Keith Paulsen and colleagues at Dartmouth University are developing a new tomographic imaging modality using near-infrared (NIR) light to create images of hemoglobin, deoxyhemoglobin, water, and lipids in the breast. Their prototype 3-D multiwavelength NIR tomographic imaging system images tumor-like breast heterogeneities. It has the potential to noninvasively provide new physiological information, such as concentrations of hemoglobin, deoxyhemoglobin, water, and lipids that can identify and characterize abnormalities deep within breast tissue.

In a February 2003 Applied Optics paper, they reported that calculated photon absorption allowed them to find higher oxy-deoxyhemoglobin concentrations and higher oxygen saturation levels in a breast cancer patient, indicating ductal carcinoma. They are working to take those methods to clinical trials. The potential of these methods is the ability to noninvasively measure time-related changes in breast tissue. Such measurements may lead to the identification of patients with high cancer risk.

Multiwavelength optical methods permit data to be obtained simultaneously from endogenous substances and exogenously administered passive or activatable molecular probes. This combination of data may reflect the molecular signatures of cancers and provide a means for improved early detection and cancer diagnosis. Optical methods are now being combined with MRI as a multimodality approach for cancer diagnosis and intervention. Dr. Paulson's work provides an important underlying platform for cancer-related developments that are expected to be commercially available in two to three years.

Optical Screening for Cervical Cancer
Researchers led by Rebecca Richards-Kortum, Ph.D., carried out large clinical trials to explore the information content in fluorescence emission spectra from cervical tissue using different excitation wavelengths. This large body of data enabled them to demonstrate that, using 2-3 excitation wavelengths, they could retain the majority of diagnostic information [1]. This has simplified the design of inexpensive imaging systems that can be used for screening trials in developed and developing countries.

They have also made major progress in understanding the biological basis of cervical tissue fluorescence and used this knowledge to develop mathematical models of cervical fluorescence. These important studies revealed the biological basis for changes in normal cervical fluorescence as women age (stromal collagen fluorescence increases with age, epithelial nicotinamide adenine dinucleotide and flavin adenine dinucleotide fluorescence decrease with age). As dysplasia develops, there is an increase associated with the cytoplasmic fluorescence of dysplastic epithelial cells and a decrease in the stromal fluorescence of collagen [2,3]. They incorporated these optical properties into a numerical model of cervical tissue fluorescence [4] and can accurately separate normal and neoplastic tissue.

In one important clinical trial, Dr. Richards-Kortum's group demonstrated that the menstrual cycle did not account for a significant amount of variability in measurements; interpatient variability was more important. This important and labor-intensive clinical trial allows clinicians to use fluorescence and reflectance spectroscopy at any time during the cycle except during menstruation, as is the Papanicolaou (Pap) smear. Patients' distress and satisfaction with spectroscopy compared with the Pap smear and colposcopically directed biopsy were measured in the screening study. Participants reported significantly less pain and anxiety and were more satisfied with spectroscopy than with the usual care procedures.

References 1. Drezek R, Sokolov K, Utzinger U, et al. Understanding the contributions of NADH and collagen to cervical tissue fluorescence spectra: Modeling, measurements, and implications. Journal of Biomedical Optics 2001;6:385-96. 2. Drezek R, Guillaud M, Collier T, et al. Light scattering from cervical cells throughout neoplastic progression: Influence of nuclear morphology, DNA content, and chromatin texture. Journal of Biomedical Optics 2002; (in press).

Magnetic-Resonance-Guided Minimally Invasive Therapy
Magnetic-resonance-guided therapy (MRT) refers to the use of MR image guidance to achieve minimally invasive surgery or therapy of various localized cancers, including brain, breast, liver, and prostate. Investigators at Harvard's Brigham and Women's Hospital in Boston, led by Ferenc Jolesz, M.D., are conducting research on the scientific, technical, and medical aspects of MRT. This includes research on key aspects of MRT, from image acquisition to development of MR-compatible minimally invasive surgical and therapeutic methods and their clinical translation through early-phase clinical trials. This group led the field in all these areas. Dr. Joelsz received an NCI grant award for MRT, and a curriculum development award for multidisciplinary training in image-guided therapy. Other team members are principal investigators on MRT-related NIH grants.

Key accomplishments include development of an integrated system for intra-operative MR image guidance in neurosurgery. The team has conducted more than 150 neurosurgical procedures, including minimally invasive craniotomies for brain tumor resection and brain biopsies. The procedures have incorporated advances in MR image acquisition and processing provided by other team members. MRI innovations include implementation of broadband MRI for real-time dynamic and multiresolution image acquisition. The innovations have been extended to adaptive MRI techniques for guidance in probe-tracking and thermal-ablation therapies. Advances in image processing for guiding surgery and therapy include development of a high-quality open-source 3-D image registration and display software package (3D-Slicer), which is now being integrated with a library of image-processing toolkits (ITK) provided by the National Library of Medicine. Neurosurgeons use the 3D-Slicer for co-registration of anatomical and functional MR data for surgical planning and intra-operative image guidance.

The group has also made significant advances in developing MR-compatible high-intensity focused ultrasound (HIFU) for applications in image-guided therapy, in particular thermal ablation of tumors. This is achieved by focusing a collection of ultrasonic beams to increase sonic beam intensity at a point deep within tissue to cause thermal coagulation. MR guidance localizes HIFU foci and temperature maps to guide the therapy. Clinical collaborators of the scientific team, led by Dr. Kullervo Hynenen, are conducting clinical trials of MR-guided HIFU therapy of breast cancers and uterine fibroids. Dr. Hynenen has also investigated the utility of HIFU in focal opening of the blood-brain-barrier for targeted image-guided delivery of drug or gene therapy.

Another significant group accomplishment is integrating MRT in a radiation oncology procedure to more accurately localize prostate brachytherapy. Radiologist Dr. Clare Tempany is leading clinical trials in this research. For the past decade this group has advanced MRT by developing innovative imaging techniques and integrating them into surgical and therapeutic cancer management. Recent group publications are cited below.

References 1. Jolesz FA. Future perspectives in intraoperative imaging. Acta Neurochirurgica Supplement 2003;85:7-13. 1. Jolesz FA, Talos IF, Schwartz RB, et al. Intraoperative MR imaging and MR-guided therapy for brain tumors. Neuroimaging Clinics of North America 2002; (in press). 3. Ferrant M, Nabavi A, Macq B, Balck PM, Jolesz FA, Kikinis R, Warfield SK. Serial registration of intraoperative MR images of the brain. Medical Image Analysis 2002;6:337-59. 4. McDonald N, King RL, Jolesz FA, Hynenen K. The use of quantitative temperature images to predict the optimal power for focused ultrasound surgery: in-vivo verification in rabbit muscle and brain. Medical Physics 2002;29:356-65. 5. Walker DG, Talos IF, Bromfield E, Balck PM. Intraoperative magnetic resonance imaging for surgical treatment of lesions producing seizures. Journal of Clinical Neuroscience 2002;9:515-20. 6. Hirose MA, Bharatha A, Hata N, et al. Quantitative MR imaging assessment of prostate gland deformation before and during MR imaging-guided brachytherapy. Academic Radiology 2002;9:906-12. 7. Vosburg KG, Jolesz FA. The concept of image-guided therapy. Academic Radiology 2003;10:176-9.