Improved technology changes radiologists’ approach
Technology advancements?such as the 64-slice multidetector CT?increase detection capabilities by providing radiologists with more information than ever before. To give radiologists a clear picture of technology expectations and best practices for thoracic imaging, four physicians presented the CME course “New Thoracic Imaging Technologies: 64-slice Multidetector CT, Postprocessing, PET/CT, and Enhanced Radiography Methods” at the 2006 annual meeting of the Radiological Society of North America (RSNA) on November 26, 2006, in Chicago.
The presenters outlined the advantages of new thoracic imaging methods, noted which patients make the best candidates for examination using these modalities, and predicted advances in the field. The course was moderated by 2007 RSNA president-elect Theresa C. McLoud, MD, associate radiologist-in-chief and director of education for the Department of Radiology at Massachusetts General Hospital, Boston, and professor of radiology at Harvard University, Cambridge, Mass.
Multidetector CT: Clear Results
Geoffrey D. Rubin, MD, of the Department of Radiology, Stanford University School of Medicine, Stanford, Calif, spoke on 64-slice multidetector CT and its place in imaging examinations of the chest. He said that a phenomenal amount of information is now available using this modality, pushing radiologists beyond the standard ways they have been accustomed to viewing CT data. Resolution is tremendous when slices 1 mm thick are obtained using 64-row CT scanners, and features that are very difficult to detect on axial images become quite clear following volume rendering. Volumetric data create a striking increase in the ability to understand both disease and anatomy, Rubin said.
Current workstations, he added, allow fly-throughs to be put together in minutes. With the availability of 64-row data, however, comes a responsibility to glean all possible information from the examination.
Routine reconstruction of sections of the lung that are thinner than 1 mm has not been subjected to formal study, at least in the peer-reviewed literature; Rubin found only subtle differences between reconstructions that were 1.5 mm and 0.75 mm thick, so the use of very thin sections is probably not going to become routine soon.
The impact of 64-row CT technology on imaging of the lung, Rubin said, is especially important in reducing the motion artifacts that radiologists have dealt with for so long in the left lower lobe, where the motion of the heart has often caused blurring, ghosting, or double images. The enhanced speed of a 64-row CT scanner makes it practical to use cardiac gating routinely in imaging the chest during a single breath hold. Half scans, using only the data acquired during 180? of detector motion instead of the full rotation, help reduce cardiac motion artifacts. Existing data can be reinterpolated using a half scan to make many motion artifacts disappear; no additional image acquisition is needed, as the half scan is just a different use of the data already captured.
Retrospective gating is associated with its own set of artifacts, but prospective gating, in which the scanner is triggered repeatedly at the same part of the cardiac cycle (also using a half-scan interpolation without the helical aspect), seems to be subject to fewer artifacts.
The capabilities of 64-row CT in imaging of the thorax extend beyond the heart and lungs to the mediastinum and, especially, the vascular system, Rubin said. Routine use of gating allows the dynamic physiology of the vessels to be seen, and dissections and calcifications that were not detected using earlier CT methods are being viewed now. This can have a substantial impact on patient management. With routine gating, the clots that create acute coronary syndrome can be seen, and the source of a leak leading to pseudoaneurysm can be pinpointed.
Faster rotation is the hallmark of 64-row CT, and this provides better temporal resolution that makes high-quality gated scanning possible. Whether it is used with or without prospective gating, 64-row CT also increases table speed, making it easier to perform retrospective gating for full-chest imaging. Better cardiac and aortic imaging are promoted by 64-row CT, even though pulmonary imaging itself may not be improved significantly, Rubin concluded.
Postprocessing: Interactive Interpretation
James F. Gruden, MD, of the Mayo Clinic, Scottsdale, Ariz, covered postprocessing techniques used in thoracic imaging studies. Multidetector CT has reached a point at which the chest can be imaged, with submillimeter collimation, within 4 to 6 seconds. This has eliminated artifacts such as pseudofractures, he explained, as well as other problems associated with respiratory motion. This has been tremendously helpful in chest imaging overall, but its real advantage in viewing the lung has been the ability to conduct isotropic imaging. Radiologists can now interact with CT data in ways that increase diagnostic capabilities.
Multidetector CT, after the scan, is the subject of sophisticated manipulation, some of it performed by radiologists. Technologists also perform some postprocessing, typically as part of batch protocols for PACS that enhance the ability of radiologists to make diagnoses based on axial images.
At this point, data management becomes an issue because of the massive size of the files involved. Archiving strategies and workflow management, for example, may need to be adjusted to cope with the data that multidetector CT generates, Gruden said, and the network and PACS also may require changes to maintain practice efficiency.
“Interactive interpretation” may be a more accurate term than “postprocessing” for working with volume data, and multiplanary formats are the cornerstone of isotropic interactive interpretation. In his clinical experience, Gruden said, this is most commonly used to look at bones. In axial imaging, it is very time-consuming to look at every rib and vertebral body in cases involving rib trauma or neoplasms, but multiplanar imaging makes it easy to see fractures and other bone problems on a workstation simply by dropping the cursor on the area of interest. Diagnosis of focal lung disease also is aided, since what appear to be ground-glass opacities on axial images often turn out, when seen in multiplanar views, to be areas of atelectasis that require no follow-up action. Even in cases where the findings are present in axial views, multiplanar imaging increases the ease and speed of diagnosis, Gruden added.
Maximum-intensity projection and minimum-intensity projection imaging improve accuracy and decrease interpretation time, he continued. In maximum-intensity projection, the brightest objects can be used to create a slab showing only the structures of highest density. This is primarily used for vascular imaging (in fact, it was developed for MR angiography) and nodule detection, and it can be used in conjunction with multiplanar imaging to determine whether a structure is within or outside a vessel. Minimum-intensity projection is particularly useful in delineating the extent of emphysema.
Virtual bronchoscopy and other types of surface shading are not in routine use, but can be desirable in certain cases. They are performed by the radiologist (not the technologist) when a stricture is seen, mainly to clarify the degree of narrowing present.
Likewise, volume rendering is helpful in specific instances, although it is rarely used in chest imaging. It can help the radiologist assess spatial relationships quickly when these are unusual or unclear, because these images are truly 3D; but they can obscure some diagnostically important information because they are constructed using density data. Some aneurysms or thrombi can be hidden, for example, so volume rendering should not be used in isolation.
PET/CT: Approved Coverage
Reginald F. Munden, DMD, MD, of the University of Texas MD Anderson Cancer Center, Houston, presented the course section on PET/CT imaging of the thorax. As he reported, approved indications for using PET/CT include characterization of the solitary pulmonary nodule (which was one of the first uses of the modality) and the initial staging and restaging of lung cancer.
PET/CT use in cases of small-cell lung cancer is not approved by the Centers for Medicare and Medicaid Services (CMS) for standard reimbursement, but there is a program that permits it to be covered if CMS is provided with both the information necessary for preapproval and the follow-up data. This policy allows CMS to build a database to support future coverage decisions and lets Medicare and Medicaid patients who would not otherwise qualify receive PET/CT studies.
In cancer cases, PET/CT is approved for diagnostic use only when it has the potential to allow the patient to avoid undergoing another procedure, such as a biopsy. For financial reasons, it is not routinely used for screening purposes. PET usually is used for staging after the lung-cancer diagnosis has been made. Restaging is also performed, usually after the patient has completed a cycle of chemotherapy. Assessment of the patient’s response to therapy before that cycle has ended typically is not covered.
Lung-cancer staging involves the assignment of T numbers based on the size of the tumor and whether it is invading a structure that cannot be resected, with the line between T3 and T4 indicating whether the tumor is resectable or not. PET does not have sufficient inherent resolution to make it useful for assigning T numbers, but CT and MRI can be used to perform this task. PET indicates activity in an area, but does not precisely indicate a tumor’s borders.
New Techniques: Radiology’s Future
Heber MacMahon, MD, of the University of Chicago summarized recent advances in the use of new detection methods with chest radiography. Most of the forms of thoracic imaging discussed by the other presenters, MacMahon noted, are brought to bear only after disease has been seen on a chest radiograph.
Digital imaging is clearly past the point at which argument over its benefits, in comparison with analog systems, can reasonably continue. Both image quality and image access are improved using digital imaging, and radiologists are beginning to take advantage of diagnostic capabilities that were not present in film-based systems. Because of the complex anatomy of the chest, for example, it is easy to miss lung cancers; MacMahon explained that several series have shown that the average size of missed tumors, in plain-film examinations, is 1.5 cm. Of these overlooked lesions, about 80% are partially obscured by bone.
Dual-energy imaging is used to overcome this problem. Two approaches are common: In the first, two phosphor plates are separated by a copper filter. The first plate captures the full energy spectrum, but the second captures only the higher energy spectrum. The combination is then used to generate dual-energy soft-tissue and bone images. One standard exposure thus creates a conventional image and the two dual-energy images, with no additional radiation. The second approach to dual-energy imaging is sequential exposure, used with newer flat-panel detectors. Not enough radiation passes through to perform two-plate image capture, so two images are obtained in rapid sequence; one at a low kV level, and the second at a high kV level. Very good bone and soft-tissue images result, with the slight disadvantage that the 200-millisecond separation between the two shots can create a small motion artifact. Both techniques work well, MacMahon reported, and either can (and should) be used routinely.
Temporal subtraction, used in Japan, is not widely used in the United States and is not yet commercially available. It can be applied to a series of radiographs taken at different times (even years apart) to make the differences between them more obvious. Diagnostically, these differences are of high importance. The system used for temporal subtraction automatically warps the position of the previous examination to match the current study and then performs a subtraction between the two. If there has been no change, everything cancels out and is seen as a medium gray. If there have been changes, the new areas stand out clearly because they are darker. In observer studies conducted in Chicago and Japan, MacMahon said, radiologists also read images more quickly when temporal subtraction was available, perhaps because it increased their confidence levels.
Tomosynthesis, developed at Duke University, Durham, NC, allows 3D information to be extracted from a radiograph. The device used for this technique obtains 71 images during a single breath-hold lasting 11 seconds, using continuous movement of the x-ray tube over an arc of 20?. The patient’s exposure to radiation is comparable to that created by a lateral screening chest radiograph. Tomosynthesis is used instead of standard chest radiography to narrow the gap between the amount of diagnostic information provided by radiography and that provided by CT. This technology is not yet readily available, but is expected to become a replacement first-line imaging method.
Dynamic chest radiography uses the flat-panel detector’s speed to produce a series of images that are, in essence, fluoroscopic. This is not of clear diagnostic value in itself, but investigators have used very rapid temporal subtraction techniques to amplify an attenuated image obtained during expiration. This yields a color-coded map that shows changes in opacity during respiration due to the amount of air in the lungs. In healthy subjects, the opacity changes are uniform; in the presence of pathology, they are not. The method allows regional ventilation and even abnormal regional perfusion to be seen. Ongoing work should make this technology available in the future.
Computer-aided detection (CAD), however, has been in wide use for some time. It involves automated analysis of the chest radiograph, primarily for the detection of lung nodules. Candidate nodules are identified first by the software; then, a number of complex checking methods are used to eliminate false positives. Marks that indicate suspicious areas are superimposed on the radiograph. In observer studies, MacMahon said, radiologists performed better using CAD, but false positives due to overlapping anatomy can be a problem. This is becoming less true as CAD software improves, but complete integration of CAD and PACS is of great importance, MacMahon said.
All four course presenters believed that thoracic-imaging technologies of all kinds have reached an important stage of development: While most of them are clearly mature enough for broad clinical use, innovations in their application and refinements in the devices and their software are continuing at a pace that should yield better diagnostic results. Since the demand for thoracic imaging continues to rise, and procedural volumes grow in response to that demand, there should be no shortage of opportunities to use new thoracic-imaging methods clinically for the benefit of patients and their physicians.
Kris Kyes is technical editor of Medical Imaging. For more information, contact .