- Email: [email protected]
Po-topic IV-12
Academic Radiology, Vol 10, No 8, August 2003
PO-Topic IV-10 MIXTURE-BASED COMPUTER AIDED DETECTION SYSTEM FOR VIRTUAL CYSTOSCOPY L. Li,* Z. Wang, D. Harrington, W. Huang, Z. Liang, State University of New York at Stony Brook, Stony Brook, NY Bladder cancer is the fifth leading cause of cancer deaths in the United States. While currently available cyctoscopy is an accurate diagnostic procedure for detecting abnormalities, it is invasive, with limited field of view, and lack of an objective scale. Recently, virtual cystoscopy (VC) has been developed as an alternative means for early detection of bladder cancer. Several CT-based VC studies have been reported recently. However, they can only detect the tumor based on geometry information because bladder muscle has almost no contrast from other soft tissues. This limits their ability for early detection of small tumors. In this study, we proposed a novel mixture-based computer aided detection (CAD) system using multispectral (T1- and T2-weighted) MR images, where a better contrast is given as compared with CT images. As T1- and T2-weighted images are spatially registered over 3D space, information extracted from multispectral images is obviously more valuable than that extracted from each image individually. In addition, because bladder tumors tend to develop gradually and migrate slowly from the mucosa into the bladder muscle, we focus on analyzing the mucosa layer for tumor detection. Our mixture-based segmented scheme quantifies the mucosa layer as percentages of tissues inside each voxel. It minimizes theoretically partial volume effects (PVE) on the tissue boundaries and provides tissue growth tendency in addition to the anatomic structure of the tissues. MR images were acquired by a Picker 1.5T Edge (Philips Medical Systems, Andover, MA) whole-body scanner. After the patient voided and then drank a cup of water, images of the nearly empty state were acquired. Waiting for a while, when the patient’s bladder feels full, full-state images were acquired. We reconstructed the inner and outer bladder surfaces from the mixture-based segmented data. The system provides physicians four windows for viewing, through which we can view the bladder at different states simultaneously. At each state, the system also provides two-view direction (from both inside and outside). When physicians rotate the bladder at any window, the system will calculate the relative parameters and rotate the bladder at the other three windows. We tested our detection scheme further on several patient and volunteer studies. The scheme searches abnormalities first from the full state dataset, and then the empty state, based on geometrical information. If an abnormality is detected at the same location for both states, then the detected one is a candidate for tumor. If the detection misses one in one state, then texture information is examined at the location as indicated by the other state with abnormality. Perfect results were obtained from volunteer studies, as expected. For the patient study, one abnormal 12-mm section was detected at both states. A small 4-mm one was detected in the full state and missed in the empty state. When the scheme examined the texture information at the corresponding location in the empty state dataset, this small one was detected and became a candidate. All candidates were finally examined by the physician using the display system. Results are promising.
POSTER SESSIONS AND ABSTRACTS
The increasing complexity and length of percutaneous minimally invasive procedures provides the impetus to develop procedure guidance technology that does not rely on ionizing radiation. Our research has focused on demonstrating the feasibility of magnetic tracking guidance in the interventional suite. Our goals are (1) to demonstrate magnetic tracking technology for following internal organ respiratory motion; and (2) to enhance the accuracy of needle placement in a variety of interventional abdominal procedures. Ultimately, we envision integrating anatomic and functional imaging information with real-time magnetic tracking guidance to accurately target and deliver local antineoplastic therapies. We designed and constructed a unique liver phantom that features translational movement simulating respiratory motion. To simplify the design, we assumed that hepatic respiratory motion occurs in the craniocaudal direction only and that the liver itself is not deformed by diaphragmatic motion. We incorporated the AURORA magnetic tracking system (Northern Digital Inc, Waterloo, Canada), which is based on cylindrically shaped sensors that are extremely small (0.9 mm in diameter, 8 mm in length). The small size enables the sensors to be embedded into standard size needles and catheters. For procedure guidance, we created a graphical user interface including a path planning capability and an active tracking algorithm. The system displays the target organ motion simulating resting respirations, allowing for accurate needle movement during the end-expiratory phase of the respiratory cycle. A single magnetically tracked needle was placed within the liver phantom to serve as an internal fiducial. An initial accuracy evaluation of magnetic-tracking guided-needle placement within the phantom was completed in a standard interventional radiology suite. The targets consisted of two elliptical liver tumors of maximum dimension 3.1 and 2.2 cm embedded in the synthetic liver. 16 liver tumor punctures were attempted. Successful puncture of liver tumors was achieved in 14 of 16 attempts (87.5%) by two users. The average time of each procedure was short (163⫾57 seconds.) In a second series of experiments, single or sequential two-vessel punctures simulating simultaneous hepatic and portal vein punctures in a TIPS procedure were attempted during simulated resting respiratory motion. A catheter-based magnetically tracked fiducial was used as an internal reference. Targeted vessels measured 4 –5 mm in diameter, and both vessels were successfully punctured in a single attempt in 65% of the trials. The mean error calculated according to the selected target points in magnetic space was 4.8 mm for successful punctures and 8.2 mm for unsuccessful punctures. The mean registration error was 2.8 mm. Our current work is focused on determining the accuracy limitations of our system. Specifically, we are investigating the relationship between target vessel diameter and puncture success. Our long-term goal is to develop the algorithms to guide accurate needle placement within tumors for the delivery of specific therapy (eg, ablative, chemotherapeutic, or gene vector delivery), targeted for specific tumor morphology or vulnerable physiology.
PO-Topic IV-12 PO-Topic IV-11 MAGNETIC TRACKING OF INSTRUMENTS FOR INTERVENTIONAL PROCEDURES E. Levy,1 K. Cleary,1 F. Banovac,1 D. Lindisch,1 J. Tang,1 S. Xu,2 N. Glossop,3 (1) Georgetown University, Washington, DC, (2) Johns Hopkins University, Baltimore, MD, (3) Traxtal Technologies LLC, Bellaire, TX Magnetic tracking of instruments during image-guided procedures is an emerging area of research interest as the technology becomes miniaturized.
ROBOTICALLY ASSISTED INTERVENTIONS IN THE SPINE AND LUNG K. Cleary,1 V. Watson,1 D. Lindisch,1 D. Stoianovici,2 D. Mazilu,2 A. Patriciu,2 C. White,3, (1) Georgetown University, Washington, DC, (2) Johns Hopkins Medical Institutions, Baltimore, MD, (3) University of Maryland Medical Center, Baltimore, MD Percutaneous interventions are performed by freehand passages of instruments, such as needles, from the skin surface to the anatomy of interest. The main problem with this approach is that the physician can be inaccu-
957
POSTER SESSIONS AND ABSTRACTS
rate in aligning the instrument and staying on course. A joystick-controlled robotic needle driver may allow the physician to more precisely target the anatomy. For the past 2 years, we have been adapting an existing robotic system to place a 22-gauge needle for nerve and facet blocks in the spine. A cadaver study has been completed in which the robot was used under joystick control to place 12 needles into the lumbar paraspinal region of an embalmed female cadaver. Small metal BB nipple markers (1 mm in diameter) were inserted percutaneously to serve as targets. All needles were placed within 3 mm of the target BB, and the average distance was 1.44 mm with a standard deviation of 0.66 mm. Following this study, IRB and FDA approval was received to conduct a randomized clinical trial with an initial sample size of 20 patients (10 with the robot, 10 without). To date, 10 patients have been completed with no complications. The results of this trial should be available by November 2003, but preliminary results indicate that it is feasible to use a joystick-controlled robotic needle driver to accurately place needles for nerve and facet blocks. The second clinical application we are investigating is lung biopsy. The goal is to use the robotic system to assist the physician in accurate computed tomography (CT) fluoroscopy-guided needle biopsy of lung nodules. The use of CT for lung cancer screening is rapidly expanding. Percutaneous image-guided biopsy of the lung is a moderately difficult procedure with the potential for morbidity from pneumothorax and hemorrhage. For those nodules less than 1 cm in size, biopsy is more difficult, and there are a limited number of trained personnel who can perform them. We expect the increasing use of screening CT will result in a rapid growth in demand for image-guided percutaneous biopsy of these nodules. This project will proceed in two phases. The goal of the first phase is to demonstrate the feasibility of using a joystick-controlled robotic system to accurately hit simulated lesions in a phantom under CT fluoroscopy guidance. A prototype gripper will be developed and tested on a custom-built respiring lung phantom model. The goal of the second phase is to develop an enhanced gripper along with a path planning capability and demonstrate this approach in phantom and animal studies. Financial disclosure: A company (ImageGuide) has been formed to commercialize the robot described here and several of the authors have a financial interest in this company.
PO-Topic IV-13 DOSE OPTIMIZATION IN HDR BRACHYTHERAPY FOR CERVICAL CANCER USING FDG-PET S. Wahab,* R. Malyapa, S. Mutic, P. Grigsby, I. Zoberi, T. Miller, D. Low, Washington University Department of Radiation Oncology, St Louis, MO The purpose of this study was to investigate the possibility of dose optimization in high-dose rate (HDR) intracavitary brachytherapy for the treatment of cervical cancer with tumor volumes delineated using positron emission tomography (PET). 18F-fluorodeoxyglucose (FDG)-PET scans were acquired for 10 patients with implanted tandem and ovoid (T&O) applicators containing tubes of FDG. The FDG-PET images were transferred to a commercial 3D treatment planning system where the tumor volume was defined using a binary threshold technique. Tumor tissue was identified on FDG-PET images as any voxel in the 3D data set with counts greater than a fixed threshold fraction (40%) of the peak intensity of the tumor. This threshold level was selected from information gained by correlation of FDG-PET volumes with volume of cervical tumor clearly defined on CT scan images in a series of patients with cervical cancer. The bladder and rectum were contoured by the physician who treated the patients. The tandem and ovoid applicators, which contained catheters filled with FDG, had a characteristic appearance on the FDG-PET images and were uniquely identified by their expected positions. FDG-PET based 3D treatment plans were first simulated for conventional HDR delivery.
958
Academic Radiology, Vol 10, No 8, August 2003
The source distribution, based on our clinical protocols, was designed to deliver 6.5 Gy to Point A under ideal conditions. Point A was identified according to the classical definition relative to the reconstructed applicators. There were nine active dwell positions in the tandem and three active positions in each of the ovoids. Maintaining identical contours on the FDG-PET images, investigational treatment plans were then created my manipulating dwell positions and dwell times. The new plans used an integrated reference air kerma that was equivalent to the conventional plans. The source arrangement that best maximized tumor coverage was found to be a tandem hypothetically extended by 2 cm superiorly, continuing along its original trajectory. The extended tandem included 15 active dwell positions. Dwell times were customized to optimize dose distributions to the individual GTV versus dose distributions for the two sets of plans were compared by percent tumor coverage and by minimum dose within tumor. The optimized plans demonstrate improvement in both tumor coverage (mean 33.0%, std 15.1%, median 31.4%) and peripheral dose (mean 32.3%, std 43.9%, median 36.0%). The optimized plans demonstrate bladder reference point doses that are increased (mean 14.1%, std 32.5%, median 13.6%) and rectal dose that are decreased (mean ⫺31.3%, std 22.2%, median ⫺35.2%). This study demonstrates the potential for improving HDR brachytherapy dose distributions for FDG-PETdefined cervical cancer. Superior tumor coverage and peripheral dose can be achieved by utilizing a single extended tandem applicator compared with conventional T&O dose distributions, and by optimizing source weighting to conform to the geometry of the tumor. Acceptable levels of toxicity based on bladder and rectal dose volume histograms were observed in the investigational plans of most patients.
PO-Topic IV-14 IMAGING & LOCALIZING DEVICE USING THE BARKHAUSEN EFFECT J.F. Dicello,* S. McAllister, R.J. von Gutfeld, J.F. Ziegler, M. Ziegler, Johns Hopkins University School of Medicine, Baltimore, MD, IBM Microprocessor Design Division, East Fishkill, NY, IBM T.J. Watson Research Center, Yorktown Heights, NY, BioMed Innovations, Timonium, MD We have been investigating the use of amorphous magnetic materials as locator tags for medical applications. Heinrich Barkhausen discovered in 1919 that a slow, smooth increase of magnetic field applied to a piece of ferromagnetic material, such as iron, causes the material to become magnetized not continuously, but in small steps. A large Barkhausen effect (large step change) was observed in 1931 because of domain reversal at a rapid velocity along the wire. In the late 1980s, an even larger Barkhausen effect was produced in amorphous wires giving rise to an even sharper magnetic pulse because of domain or flux reversal in the core of the wire. In an AC magnetic field, the reversal of the domains emits large, sharp magnetic pulses of the opposite sign. We are taking advantage of this pulsed response to tag and image the location and orientation of implanted devices. Several factors affect the feasibility of imaging and reconstruction of embedded tags. These include the strength of the detected signals; the ability to reconstruct locations and orientations of the tags; and signal sensitivity to tissue composition, inhomogeneities, and the presence of neighboring metals. We have obtained voltage signals from wires of several diameters with an iron-core pickup coil along a distance from the center of a wire with the coil as a function of angle for a 60 – 80 Hz driving field. At a distance of about 8 cm., we have obtained a strong signal of typically 2V (depending on length), more than 50 times the background noise (about 20 mV). At 30 cm separation, we detected a signal of 0.5 V, 25:1 above background noise. When the wire direction is mis-oriented relative to the driving AC magnetic field, the amplitude and phase of the resultant output signal changes. This is because only the portion of the magnetic field that is parallel to the wire axis contributes to the Barkhausen
PO-Topic IV-10 MIXTURE-BASED COMPUTER AIDED DETECTION SYSTEM FOR VIRTUAL CYSTOSCOPY L. Li,* Z. Wang, D. Harrington, W. Huang, Z. Liang, State University of New York at Stony Brook, Stony Brook, NY Bladder cancer is the fifth leading cause of cancer deaths in the United States. While currently available cyctoscopy is an accurate diagnostic procedure for detecting abnormalities, it is invasive, with limited field of view, and lack of an objective scale. Recently, virtual cystoscopy (VC) has been developed as an alternative means for early detection of bladder cancer. Several CT-based VC studies have been reported recently. However, they can only detect the tumor based on geometry information because bladder muscle has almost no contrast from other soft tissues. This limits their ability for early detection of small tumors. In this study, we proposed a novel mixture-based computer aided detection (CAD) system using multispectral (T1- and T2-weighted) MR images, where a better contrast is given as compared with CT images. As T1- and T2-weighted images are spatially registered over 3D space, information extracted from multispectral images is obviously more valuable than that extracted from each image individually. In addition, because bladder tumors tend to develop gradually and migrate slowly from the mucosa into the bladder muscle, we focus on analyzing the mucosa layer for tumor detection. Our mixture-based segmented scheme quantifies the mucosa layer as percentages of tissues inside each voxel. It minimizes theoretically partial volume effects (PVE) on the tissue boundaries and provides tissue growth tendency in addition to the anatomic structure of the tissues. MR images were acquired by a Picker 1.5T Edge (Philips Medical Systems, Andover, MA) whole-body scanner. After the patient voided and then drank a cup of water, images of the nearly empty state were acquired. Waiting for a while, when the patient’s bladder feels full, full-state images were acquired. We reconstructed the inner and outer bladder surfaces from the mixture-based segmented data. The system provides physicians four windows for viewing, through which we can view the bladder at different states simultaneously. At each state, the system also provides two-view direction (from both inside and outside). When physicians rotate the bladder at any window, the system will calculate the relative parameters and rotate the bladder at the other three windows. We tested our detection scheme further on several patient and volunteer studies. The scheme searches abnormalities first from the full state dataset, and then the empty state, based on geometrical information. If an abnormality is detected at the same location for both states, then the detected one is a candidate for tumor. If the detection misses one in one state, then texture information is examined at the location as indicated by the other state with abnormality. Perfect results were obtained from volunteer studies, as expected. For the patient study, one abnormal 12-mm section was detected at both states. A small 4-mm one was detected in the full state and missed in the empty state. When the scheme examined the texture information at the corresponding location in the empty state dataset, this small one was detected and became a candidate. All candidates were finally examined by the physician using the display system. Results are promising.
POSTER SESSIONS AND ABSTRACTS
The increasing complexity and length of percutaneous minimally invasive procedures provides the impetus to develop procedure guidance technology that does not rely on ionizing radiation. Our research has focused on demonstrating the feasibility of magnetic tracking guidance in the interventional suite. Our goals are (1) to demonstrate magnetic tracking technology for following internal organ respiratory motion; and (2) to enhance the accuracy of needle placement in a variety of interventional abdominal procedures. Ultimately, we envision integrating anatomic and functional imaging information with real-time magnetic tracking guidance to accurately target and deliver local antineoplastic therapies. We designed and constructed a unique liver phantom that features translational movement simulating respiratory motion. To simplify the design, we assumed that hepatic respiratory motion occurs in the craniocaudal direction only and that the liver itself is not deformed by diaphragmatic motion. We incorporated the AURORA magnetic tracking system (Northern Digital Inc, Waterloo, Canada), which is based on cylindrically shaped sensors that are extremely small (0.9 mm in diameter, 8 mm in length). The small size enables the sensors to be embedded into standard size needles and catheters. For procedure guidance, we created a graphical user interface including a path planning capability and an active tracking algorithm. The system displays the target organ motion simulating resting respirations, allowing for accurate needle movement during the end-expiratory phase of the respiratory cycle. A single magnetically tracked needle was placed within the liver phantom to serve as an internal fiducial. An initial accuracy evaluation of magnetic-tracking guided-needle placement within the phantom was completed in a standard interventional radiology suite. The targets consisted of two elliptical liver tumors of maximum dimension 3.1 and 2.2 cm embedded in the synthetic liver. 16 liver tumor punctures were attempted. Successful puncture of liver tumors was achieved in 14 of 16 attempts (87.5%) by two users. The average time of each procedure was short (163⫾57 seconds.) In a second series of experiments, single or sequential two-vessel punctures simulating simultaneous hepatic and portal vein punctures in a TIPS procedure were attempted during simulated resting respiratory motion. A catheter-based magnetically tracked fiducial was used as an internal reference. Targeted vessels measured 4 –5 mm in diameter, and both vessels were successfully punctured in a single attempt in 65% of the trials. The mean error calculated according to the selected target points in magnetic space was 4.8 mm for successful punctures and 8.2 mm for unsuccessful punctures. The mean registration error was 2.8 mm. Our current work is focused on determining the accuracy limitations of our system. Specifically, we are investigating the relationship between target vessel diameter and puncture success. Our long-term goal is to develop the algorithms to guide accurate needle placement within tumors for the delivery of specific therapy (eg, ablative, chemotherapeutic, or gene vector delivery), targeted for specific tumor morphology or vulnerable physiology.
PO-Topic IV-12 PO-Topic IV-11 MAGNETIC TRACKING OF INSTRUMENTS FOR INTERVENTIONAL PROCEDURES E. Levy,1 K. Cleary,1 F. Banovac,1 D. Lindisch,1 J. Tang,1 S. Xu,2 N. Glossop,3 (1) Georgetown University, Washington, DC, (2) Johns Hopkins University, Baltimore, MD, (3) Traxtal Technologies LLC, Bellaire, TX Magnetic tracking of instruments during image-guided procedures is an emerging area of research interest as the technology becomes miniaturized.
ROBOTICALLY ASSISTED INTERVENTIONS IN THE SPINE AND LUNG K. Cleary,1 V. Watson,1 D. Lindisch,1 D. Stoianovici,2 D. Mazilu,2 A. Patriciu,2 C. White,3, (1) Georgetown University, Washington, DC, (2) Johns Hopkins Medical Institutions, Baltimore, MD, (3) University of Maryland Medical Center, Baltimore, MD Percutaneous interventions are performed by freehand passages of instruments, such as needles, from the skin surface to the anatomy of interest. The main problem with this approach is that the physician can be inaccu-
957
POSTER SESSIONS AND ABSTRACTS
rate in aligning the instrument and staying on course. A joystick-controlled robotic needle driver may allow the physician to more precisely target the anatomy. For the past 2 years, we have been adapting an existing robotic system to place a 22-gauge needle for nerve and facet blocks in the spine. A cadaver study has been completed in which the robot was used under joystick control to place 12 needles into the lumbar paraspinal region of an embalmed female cadaver. Small metal BB nipple markers (1 mm in diameter) were inserted percutaneously to serve as targets. All needles were placed within 3 mm of the target BB, and the average distance was 1.44 mm with a standard deviation of 0.66 mm. Following this study, IRB and FDA approval was received to conduct a randomized clinical trial with an initial sample size of 20 patients (10 with the robot, 10 without). To date, 10 patients have been completed with no complications. The results of this trial should be available by November 2003, but preliminary results indicate that it is feasible to use a joystick-controlled robotic needle driver to accurately place needles for nerve and facet blocks. The second clinical application we are investigating is lung biopsy. The goal is to use the robotic system to assist the physician in accurate computed tomography (CT) fluoroscopy-guided needle biopsy of lung nodules. The use of CT for lung cancer screening is rapidly expanding. Percutaneous image-guided biopsy of the lung is a moderately difficult procedure with the potential for morbidity from pneumothorax and hemorrhage. For those nodules less than 1 cm in size, biopsy is more difficult, and there are a limited number of trained personnel who can perform them. We expect the increasing use of screening CT will result in a rapid growth in demand for image-guided percutaneous biopsy of these nodules. This project will proceed in two phases. The goal of the first phase is to demonstrate the feasibility of using a joystick-controlled robotic system to accurately hit simulated lesions in a phantom under CT fluoroscopy guidance. A prototype gripper will be developed and tested on a custom-built respiring lung phantom model. The goal of the second phase is to develop an enhanced gripper along with a path planning capability and demonstrate this approach in phantom and animal studies. Financial disclosure: A company (ImageGuide) has been formed to commercialize the robot described here and several of the authors have a financial interest in this company.
PO-Topic IV-13 DOSE OPTIMIZATION IN HDR BRACHYTHERAPY FOR CERVICAL CANCER USING FDG-PET S. Wahab,* R. Malyapa, S. Mutic, P. Grigsby, I. Zoberi, T. Miller, D. Low, Washington University Department of Radiation Oncology, St Louis, MO The purpose of this study was to investigate the possibility of dose optimization in high-dose rate (HDR) intracavitary brachytherapy for the treatment of cervical cancer with tumor volumes delineated using positron emission tomography (PET). 18F-fluorodeoxyglucose (FDG)-PET scans were acquired for 10 patients with implanted tandem and ovoid (T&O) applicators containing tubes of FDG. The FDG-PET images were transferred to a commercial 3D treatment planning system where the tumor volume was defined using a binary threshold technique. Tumor tissue was identified on FDG-PET images as any voxel in the 3D data set with counts greater than a fixed threshold fraction (40%) of the peak intensity of the tumor. This threshold level was selected from information gained by correlation of FDG-PET volumes with volume of cervical tumor clearly defined on CT scan images in a series of patients with cervical cancer. The bladder and rectum were contoured by the physician who treated the patients. The tandem and ovoid applicators, which contained catheters filled with FDG, had a characteristic appearance on the FDG-PET images and were uniquely identified by their expected positions. FDG-PET based 3D treatment plans were first simulated for conventional HDR delivery.
958
Academic Radiology, Vol 10, No 8, August 2003
The source distribution, based on our clinical protocols, was designed to deliver 6.5 Gy to Point A under ideal conditions. Point A was identified according to the classical definition relative to the reconstructed applicators. There were nine active dwell positions in the tandem and three active positions in each of the ovoids. Maintaining identical contours on the FDG-PET images, investigational treatment plans were then created my manipulating dwell positions and dwell times. The new plans used an integrated reference air kerma that was equivalent to the conventional plans. The source arrangement that best maximized tumor coverage was found to be a tandem hypothetically extended by 2 cm superiorly, continuing along its original trajectory. The extended tandem included 15 active dwell positions. Dwell times were customized to optimize dose distributions to the individual GTV versus dose distributions for the two sets of plans were compared by percent tumor coverage and by minimum dose within tumor. The optimized plans demonstrate improvement in both tumor coverage (mean 33.0%, std 15.1%, median 31.4%) and peripheral dose (mean 32.3%, std 43.9%, median 36.0%). The optimized plans demonstrate bladder reference point doses that are increased (mean 14.1%, std 32.5%, median 13.6%) and rectal dose that are decreased (mean ⫺31.3%, std 22.2%, median ⫺35.2%). This study demonstrates the potential for improving HDR brachytherapy dose distributions for FDG-PETdefined cervical cancer. Superior tumor coverage and peripheral dose can be achieved by utilizing a single extended tandem applicator compared with conventional T&O dose distributions, and by optimizing source weighting to conform to the geometry of the tumor. Acceptable levels of toxicity based on bladder and rectal dose volume histograms were observed in the investigational plans of most patients.
PO-Topic IV-14 IMAGING & LOCALIZING DEVICE USING THE BARKHAUSEN EFFECT J.F. Dicello,* S. McAllister, R.J. von Gutfeld, J.F. Ziegler, M. Ziegler, Johns Hopkins University School of Medicine, Baltimore, MD, IBM Microprocessor Design Division, East Fishkill, NY, IBM T.J. Watson Research Center, Yorktown Heights, NY, BioMed Innovations, Timonium, MD We have been investigating the use of amorphous magnetic materials as locator tags for medical applications. Heinrich Barkhausen discovered in 1919 that a slow, smooth increase of magnetic field applied to a piece of ferromagnetic material, such as iron, causes the material to become magnetized not continuously, but in small steps. A large Barkhausen effect (large step change) was observed in 1931 because of domain reversal at a rapid velocity along the wire. In the late 1980s, an even larger Barkhausen effect was produced in amorphous wires giving rise to an even sharper magnetic pulse because of domain or flux reversal in the core of the wire. In an AC magnetic field, the reversal of the domains emits large, sharp magnetic pulses of the opposite sign. We are taking advantage of this pulsed response to tag and image the location and orientation of implanted devices. Several factors affect the feasibility of imaging and reconstruction of embedded tags. These include the strength of the detected signals; the ability to reconstruct locations and orientations of the tags; and signal sensitivity to tissue composition, inhomogeneities, and the presence of neighboring metals. We have obtained voltage signals from wires of several diameters with an iron-core pickup coil along a distance from the center of a wire with the coil as a function of angle for a 60 – 80 Hz driving field. At a distance of about 8 cm., we have obtained a strong signal of typically 2V (depending on length), more than 50 times the background noise (about 20 mV). At 30 cm separation, we detected a signal of 0.5 V, 25:1 above background noise. When the wire direction is mis-oriented relative to the driving AC magnetic field, the amplitude and phase of the resultant output signal changes. This is because only the portion of the magnetic field that is parallel to the wire axis contributes to the Barkhausen