Chest: Other Thoracic Applications Including 3D Airway Imaging and Aortic Dissection

Leo P. Lawler, MD, FRCR1, Elliot K. Fishman MD, FACR1.

3D Lung CT-Introduction

The potential for superior image quality and alternate methods of data display and analysis continues to evolve and provide us with novel approaches to non-invasive thorax evaluation which may improve on the already high diagnostic accuracy and yield of helical CT. Multidetector row CT (MDCT) technology has improved many of the parameters of helical scanning and imaging acquisition with rapid progress towards isotropic (cubic) voxels of data [Berland, 1998 #11; Rubin, 2000 #113; Klingenbeck-Regn, 1999 #74; Fuchs, 2000 #45; Rydberg, 2000 #116; Rydberg, 2000 #117]. Newer methodologies in post-processing techniques seek to realize the full potential of this MDCT data and to create new avenues for the understanding and interpretion of disease as well as influencing the design of new management algorithms and therapeutic strategies [Kirchgeorg, 1998 #73]. The first section of this chapter will look at these developments as they relate the airways, lung parenchyma, chest wall and diaphragm. The second part will address the spectrum of disease that is aortic dissection as seen through newer CT imaging techniques. In both section there will be an initial discussion of MDCT acquisition and three-dimensional (3D) post-processing followed by a review of clinical conditions.



Part I-3D Chest-Airway, Lung,Chest Wall and Diaphgram

Multidetector row CT-Airway and Lung Imaging Technique

Patients are scanned supine in a cranial to caudal direction and the clinical history will dictate the coverage required. If the problem concerns the intrathoracic airways coverage will be from the thoracic inlet to below the diaphragm whereas if the area of concern is above the thoracic inlet coverage is usually from the glottis (C5 vertebral body) or supraglottic airway [Naidich, 1996 #206; Naidich, 1997 #207]. For the majority of 3D airway and diffuse lung parenchyma studies non-contrast examinations will suffice. Contrast enhanced studies are employed when there are questions regarding airway bronchovascular relationships (e.g. rings and slings), mediastinal involvment or focal lung disease[Remy-Jardin, 1999 #225]. Dynamic imaging of the airway with MDCT [Gilkeson, 2001 #172] is possible but in an effort to reduce radiation exposure inspiration and expiration imaging is reserved for those patients in whom there is a high index of suspicion for tracheobronchomalacia or air trapping and a clear management benefit can be suggested [Goldin, 1997 #174; Gluecker, 2001 #173]. 100 mAs and 140kV are standard scanner settings but are adjusted for body size. Though low dose techniques using 50-80mA may be employed without significant loss of anatomic detail due to high inherent contrast of the lungs but must be balanced against the need for narrower slice widths with greater quantum mottle [Naidich, 1996 #206; Naidich, 1997 #207].

Volumetric CT [Kalender, 1990 #63; Kalender, 1990 #64; Kalender, 1991 #65; Kalender, 1994 #66] as applied to the airway represents significant progress over sequential scanning [Newmark, 1994 #210; Naidich, 1996 #206; Naidich, 1997 #207; Zeiberg, 1996 #245; Vock, 1990 #240] and MDCT breath-hold imaging allows for scan parameters [Klingenbeck-Regn, 1999 #74; Berland, 1998 #11; Fuchs, 2000 #45] better suited for both optimal 3D airway and lung parenchyma reconstruction [Neumann, 2000 #209]. From the thorax MDCT raw data various reconstruction slice positions, thicknesses, intervals and kernels (e.g. edge enhancing or soft tissue) may be generated from a single study acquisition. With the beam collimated to four 1mm detectors MDCT, 1.25mm slice thicknesses approximate and can depict the airways to their 6th order branches [Neumann, 2000 #209] and the secondary pulmonary lobules using a pitch of 6 (adaptive array design) without significant broadening of the slice sensitivity profile or effective slice width. Rapid MDCT acquisition [Hu, 2000 #54] such as with 2.5 mm detectors providing 3mm slice widths will give faster coverage with less noisy images adequate for airway interpretation with coverage of the entire chest in ten seconds or less ,which is especially relevant to patients with respiratory disease in whom breath-holding presents major challenges, and in whom dynamic alterations of airways might otherwise introduce interpretative difficulties through misregistration [Doi, 1999 #165].Though a dedicated airway protocol is employed with 20-30% overlapping reconstruction intervals to generate optimal 3D reconstructions, subsequently larger slice widths or alternate kernels can be used for routine parenchyma or mediastinal images or whole lung high-resolution images. MDCT improves temporal resolution [Taguchi, 2000 #131], reduces or eliminates breathing mis-registration and cardiac pulsation in the medial lung while overlapping thin sections reduce volume averaging [Zeiberg, 1996 #245; Kauczor, 1996 #183].



Post-processing-3D Airway and Lung CT

Though CT is an established standard for airway imaging the value of an individual chest CT and its relevance to the specific needs of bronchoscopists and the individual patient may be increased through the use of an increasing number of post-processing applications [Kirchgeorg, 1998 #73; Remy, 1998 #109; Zeiberg, 1996 #245; Fleiter, 1997 #171]. The potentially isotropic voxel (volume elements equal in size in all dimensions) of data produced by MDCT scanners mean that the choice of image display perspectives may be independent of the original acquisition plane and the variable anatomic airway orientation i.e. that the volume acquired may also be interpreted as a volume. Earlier attempts at 3D airway CT were not widely accepted because the images were not of sufficient quality but the quality of three-dimensional images obtained with MDCT can now provide a credible correlation for flexible bronchoscopy and one which can be generated in routine radiological practice with minimal labor intensity [Lawler, 2001 #78].

Airway and lung CT reconstructions have been largely performed by straight and curved multiplanar reconstruction (MPR and CMPR), maximum and minimum intensity projections (MIP and MinIP), shaded surface display (SSD) and volume rendering techniques (VRT) [LoCicero, 1996 #198; Lee, 1997 #192; Naidich, 1996 #206; Naidich, 1997 #207; Rubin, 1996 #227; Rubin, 1996 #228; Silverman, 1995 #234]. MPR and CMPR are not computer intensive and produce 1-voxel thick tomographic sections which can provide cross-sectional or longitudinal images [Remy-Jardin, 1996 #220; Summers, 1998 #235; Lacrosse, 1995 #191] useful for airway stenoses [Newmark, 1994 #210; Quint, 1995 #219]. MIP and MinIP with sliding thin slabs will provide specific information through a projection of attenuation values at the extremes of the range imaged at the cost of much of the data in between and though they have a role in interstitial lung disease are not applicable for airway stenoses [Remy-Jardin, 1996 #220]. SSD depicts the airway by simulating the boundary between air and soft tissue and segmenting it as a series of polygons with lighting models. SSD is good for central stenoses but since it is less good in the lung periphery, it may obscure high-density structures such as stents or broncholiths so that it has little advantage over multiplanar approaches [Naidich, 1997 #207; Murray, 1997 #91; Kauczor, 1996 #183; Lacrosse, 1995 #191].

VRT is the latest approach to 3D tracheobronchial imaging [Fishman, 1991 #38; Calhoun, 1999 #16; Ney, 1990 #211; Rubin, 1996 #227; Rubin, 1996 #228] and is a technique, which maintains high fidelity to the originally acquired data [Calhoun, 1999 #16; Remy-Jardin, 1998 #223; Remy-Jardin, 1998 #224; Johnson, 1996 #59; Johnson, 1998 #60]. All the density values within the chest may be manipulated through the application of various VRT tools to a histogram representation (i.e. a trapezoid) of all the Hounsfield values contained within the MDCT image [Calhoun, 1999 #16; Johnson, 1996 #59; Johnson, 1998 #60]. Unlike previous methods of image display, with this approach the final volume rendered image shows tissues according to their original representation in the data. The spatial relationship of structures is preserved with depth cues and opacity settings and infinite permutations of airway image editing, perspective and trapezoid application can be performed and tailored to individual cases. Clip plane editing can rapidly remove slabs of data such as overlying chest wall or lung and used with appropriate perspective settings individual bronchi may be isolated. Both internal and external perspectives may be depicted with a range of opacity settings from solid to nearly transparent and can be segmented in isolation or illustrated in relation to the soft tissues within the thorax. For example, we choose from solid airway, air bronchography, and soft tissue or luminal views among others depending on the case in question. ‘Virtual bronchoscopy’ (VB) depicts the airways from and endoluminal perspective using shaded surface display or volume rendering [Hopper, 2000 #180] with increased opacity settings [Becker, 1999 #155; Kalender, 1990 #63; Naidich, 1997 #207; Naidich, 1996 #206; Vining, 1996 #239; Rubin, 1996 #227; Rubin, 1996 #228; Summers, 2002 #236; Summers, 1998 #235; Remy-Jardin, 1998 #223; Remy-Jardin, 1998 #224; Ferretti, 2001 #37; Ferretti, 1996 #168; Ferretti, 1997 #169; Ferretti, 2000 #170]. Fly-through tools that automatically find the center-line facilitate endobronchial navigation. Combined color-coded SSD with volume rendering has been suggested to depict complex anatomy in virtual endoscopic renderings [Seemann, 2001 #233] and attempts have been made to register real and CT derived bronchoscopic data to optimize the yield of both [Bricault, 1998 #157]. Contemporary workstations also allow real time image processing and display, which is vital for a viable clinical tool for radiology interpretation and clinician consultation.



3D Airway CT-Role in Clinical Practice

The natural contrast of the air-airway interface is well suited to the properties of CT, which include high contrast resolution. Axial planar CT and bronchoscopy have largely replaced conventional bronchography for whole lung central and peripheral airway imaging. In many cases the conventional 2D axial CT will suffice, providing information on the status of the lumen, the airway wall and the extraluminal structures [Seemann, 2001 #233; Naidich, 1996 #206; Naidich, 1997 #207]. Airways that travel perpendicular to the imaging plane, though well seen for the most part, may however have subtle changes in caliber that are not appreciated without three-dimensional reconstruction [White, 1995 #146; Remy-Jardin, 1998 #223; Remy-Jardin, 1998 #224]. Other complex airway branching patterns oblique to the imaging plane also have limitations with 2D planar axial CT alone [White, 1995 #146; Remy-Jardin, 1998 #223; Remy-Jardin, 1998 #224; Naidich, 1996 #206]. 3D airway imaging reformats the data in novel ways more suited to individual tracheobronchial anatomy and a range of pathologies [Remy-Jardin, 1996 #220; Silverman, 1995 #234]. Moreover, many bronchoscopists find the 3D perspective to be more intuitive for endoscopic findings than the conventional cross-sectional axial images. When derived from an MDCT acquisition protocol it is a simple matter to supplement 2D studies with additional 3D reconstructions when indicated in routine practice.

There have been many studies positively reflecting the diagnostic accuracy of bronchoscopy or laryngoscopy compared with 3D CT [Vining, 1996 #239; Lee, 1997 #192; Fleiter, 1997 #171; Kauczor, 1996 #183; Gluecker, 2001 #173; Liewald, 1998 #197] but perhaps this comparative approach alone underestimates some of the intrinsic properties and advantages of both modalities. Though it can produce compelling endoscopic simulations [Becker, 1999 #155; Rubin, 1996 #227; Rubin, 1996 #228; Ferretti, 2001 #37; Ferretti, 1997 #169; Ferretti, 1996 #168] three-dimensional CT bronchoscopy will not replace the direct visualization, diagnostic sampling, and intervention capabilities of bronchoscopy, which directly visualizes and faithfully represents the mucosa both in hue and induration. Virtual bronchoscopic imaging is uniquely valuable when used to display both the internal airway findings and those beyond the airway together. This can be achieved by applying increased transparency or through the simultaneous display of endoluminal and multiplanar views [Naidich, 1997 #207] [Becker, 1999 #155] and software developments have matched VB images to flexible bronchoscopy images [Bricault, 1998 #157]. The added value of 3D airway CT lies in its unique potential to non-invasively produce whole lung tracheobronchial images together with precision of measurement, functional assessment and extraluminal information on the lung and mediastinum [Aquino, 1999 #152; Haponik, 1999 #177; Ferretti, 1996 #168].

The current clinical role of 3D CT is in those patients who require additional non-invasive airway and lung evaluation that cannot be provided by axial planar 2D CT alone [Remy, 1998 #109]. Though it may increase diagnostic sensitivity in select cases, in the majority its role is to provide more sophisticated interpretation, measurement accuracy and improved appreciation of the subtleties of airway or lung disease for radiologists and non-radiologists alike. 3D airway CT aids patient triage for bronchoscopy, helps plan the bronchoscopy procedure, can be of value when endoscopy is inconclusive and may obviate the need for bronchoscopy in certain cases [Kauczor, 1996 #183; Vining, 1996 #239; Ferretti, 1996 #168; Ferretti, 2001 #37] though the true impact of virtual bronchoscopy on endoscopy management has not yet been evaluated. 3D airway CT is of particular value in measuring the length of airway abnormalities and in delineating their location in relation to endoluminal and extraluminal landmarks and pathology. A pre-procedure review of 3D volume rendered images may provide a road-map for bronchoscopy with or without fluoroscopic guidance, offering preparation, which potentially enhances the efficiency or duration of the procedur and VB tools have been proposed as a means to give novice trainees some of the skills required for performing bronchoscopy[Colt, 2001 #162].



3D Airway CT-Pediatric Airway

Many of the principles of airway imaging apply equally to both the adult and pediatric population alike, with appropriate protocol dose modification for body size. Helical scanning improved our ability to get adequate airway images in younger infants who were less able to cooperate with breathholding [Gustafson, 2000 #175] and the brevity of scanning time with multidetector row CT[Hu, 2000 #54] has also decreased the need for pediatric sedation [Pappas, 2000 #96; Donnelly, 2000 #29]. In at risk patients such as children with small or malacic airways where the bronchoscope may occlude the airway, 3D CT can provide a safer noninvasive alternative [Konen, 1998 #187; Contencin, 1997 #163; Salvolini, 2000 #230; Dunham, 1996 #167] with objective evidence of the extent of expiratory collapse [Goldin, 1997 #174; Gilkeson, 2001 #172] and MDCT has recently been shown to be of some value in this area[Gilkeson, 2001 #172]. The ability to generate 3D CT angiograms and image relations beyond the airway wall is beneficial for pulmonary vascular anomalies that lead to airway compromise [Hopkins, 1996 #179] and have facilitated surgical planning.

Though there is some overlap in airway pathology there are distinct conditions for the pediatric population that deserve special mention. Congenital airway anomalies are a more common reason for referral in children [Carpenter, 1991 #161; Mahboubi, 1995 #199], for whom 2D and 3D CT have been used effectively for delineating pathology [Gustafson, 2000 #175; Kirks, 1983 #186] [Kornreich, 1993 #188; Sagy, 1996 #229; Nicotra, 1997 #213; Konen, 1998 #187]. Older forms of 3D spiral CT have been applied successfully to cases of anomalous branching patterns, stenosis [Toki, 1997 #238; Sagy, 1996 #229; Manson, 1994 #200], bronchiectasis [Kornreich, 1993 #188] pulmonary rings and slings [Katz, 1995 #182; Hopkins, 1996 #179; Manson, 1994 #200; Dunham, 1996 #167; Nicotra, 1997 #213]and extrinsic tracheal compression from mediastinal masses [Kirks, 1983 #186]. Anomalous branching patterns which may be unclear with a limited endoscopic luminal view or which may be volume averaged on axial reconstructions may be starkly obvious on whole lung 3D renderings, which clearly depict the orientation and relation of bronchi. Subtle abnormal bronchi origins or angulations such as bridging bronchus are quite difficult to appreciate with bronchoscopy or 2D CT but the relationships and angles are easily measured with volume rendering.



3D Airway CT-Airway filling defects

Airway filling defects may be classified as true or pseudolesions, benign or malignant (primary and secondary) and can be single or multiple (TABLES). 3D airway CT has been explored to assess endobronchial lesions and though secondary invasion is the most common indication it has been performed for a wide range of conditions such as primary neoplasms, metastatic disease and broncholith erosion [Summers, 1998 #235; Summers, 2002 #236; Vining, 1996 #239].

3D airway CT findings are quite non-specific for isolated filling defects and cannot reliably differentiate mucous (pseudomass), benign and malignant masses [Naidich, 1996 #206; Ferretti, 2000 #170]. Sensitivity to tumor is generally limited to detecting those lesions that distort the caliber or lumen of airways rather than submucosal or superficial spreading neoplasms [Ferretti, 2000 #170; Rodenwaldt, 1997 #226] though reformats parallel to an airway may be more sensitive to irregularity from early peribronchial tumor invasion and perspectives perpendicular to the airway better show thin web-like filling defects (e.g. tracheal web). Volume rendering is usually employed for accurate localization and enumeration of filling defects and to define any transmural extent. The 3D interpretation provides measurements from landmarks that can be cross referenced to plain film, fluoroscopy or endoscopy and may increase pulmonologist confidence in CT interpretation [Ferretti, 2000 #170; Ferretti, 2001 #37] and aid the radiation oncologist plan the therapeutic field [Armstrong, 1998 #153]. Though broncholiths may be seen and removed by endoscopy but it is important to document any extraluminal attachment to pulmonary vessels on 2D and 3D CT or their removal can cause catastrophic hemorrhage.



3D Airway CT-Airway Stenosis

Airway stenoses may be discrete or diffuse and may be mural or extramural in origin. 3D airway CT imaging has been applied to the full spectrum of congenital and acquired etiologies of stenosis both benign [Schafers, 1991 #231; Lee, 1999 #193][McAdams, 1998 #201; McAdams, 1998 #202; Summers, 1998 #235; Summers, 2002 #236; Ward, 2000 #241 and malignant [Fleiter, 1997 #171; Nicholson, 1998 #212; Toki, 1997 #238; Kauczor, 1996 #183]. Multidetector row thin section CT over longer areas of coverage facilitates depiction of more peripheral stenoses[Curtin, 1998 #164].

Etiologies resulting in airway wall thickening are usually well appreciated on 2D imaging (TABLE) but discrete and subtle airway tapering may not be appreciated on axial planar imaging alone and particularly in airways oblique to the imaging plane. In particular the length and cross-sectional area of the stenosis and its distance from anatomical landmarks are difficult with both 2D planar CT and endoluminal endoscopic views and are better served with customized clip planes along the individual airway and true orthogonal projections [Remy-Jardin, 1996 #220; Remy-Jardin, 1998 #224; Remy, 1998 #109; Quint, 1995 #219; Remy-Jardin, 1998 #223]. 3D CT interpretation involves comprehensive description of the number of lesions, their whole lung distribution, wall thickness and extraluminal relationships together with the length of lesions and their degree of lumen compromise [Remy-Jardin, 1998 #224; Rodenwaldt, 1997 #226; Fleiter, 1997 #171]. MPR and MinIP reformatting do provide additional information to the 2D study[Quint, 1995 #219; Remy-Jardin, 1996 #220; Remy, 1998 #109] but volume rendering is preferred for more complex pathologies [Remy-Jardin, 1998 #223; Remy-Jardin, 1996 #220]. For high-grade stenosis we assimilate this 3D information with 2D data to assess the lung parenchyma beyond and, in particular, air trapping and the likelihood of re-expansion of a post-obstructive collapse once the stenosis has been relieved. Signs of drowned lung with fluid filled bronchi or areas of necrosis imply re-expansion is less likely and may indicate aggressive interventions may are unwarranted. Customized multidimensional images are utilized to better assess the response or progression of inflammatory airway conditions such as Wegener’s Granulomatosis or post-lung trasnsplant stenoses [Ward, 2000 #241] [Schafers, 1991 #231] through consistent serial tailored reconstructions. Similarly volume-rendered images can provide objective and reproducible evidence of response to bronchoscopic interventions such as tumor shaving or laser photocoagulation, electrocautery or cryotherapy.



3D Airway CT-Airway Stents

There is an increasing role of self-expandable metal endoprostheses to restore and maintain airway patency [Wilson, 1996 #242; Lehman, 1998 #194; Mehta, 1999 #203; Ward, 2000 #241] to palliate major airway obstruction. Airway CT has been used widely for assessing cancer patients for such stents [Nicholson, 1998 #212] and for those patients with non-neoplastic airway disease (e.g. as following chemical aspiration and anastamotic strictures in lung transplantation recipients) [Lee, 1997 #192]. 3D airway CT helps triage suitable patients through assessing the feasibility and has replaced conventional bronchography in this role[Doi, 1999 #165]. Once planar CT has established there is salvageable lung parenchyma the main pre-bronchoscopy question asked of a 3D airway study is the distance from the cricoid or carina to beyond the stenosis and the ability of the bronchoscope to traverse the stenosis. Once the patient is selected the scan aids in the choice of optimal stent design and size [Zwischenberger, 1997 #246]. After the stent has been placed an immediate 3DCT documents the baseline positioning and re-expansion often better than an endoluminal approach. It is then possible to reproduce the same imaging perspective in a series of studies so that any subtle migration or collapse of the stent can be detected and quantified. Similarly any stent lumen compromise due to benign granulation tissue or tumor encroachment may be seen so that bronchocopy may then be reserved for those cases where CT suggests intervention for salvage is indicated.



3D Airway CT-Airway Bronchiectasis

Bronchiectatic change is a permanent abnormal dilatation of the airway and CT examination is part of any work-up [Kornreich, 1993 #188]. Transient dilation can be seen in the setting of an acute inflammatory process. It may be congenital (Mounier-Khun, Cystic Fibrosis) or acquired. Acquired conditions include infection (e.g. MAI, ABPA) and traction from adjacent fibrosis (e.g. radiation or usual interstitial pneumonities. Volumetric helical CT is advantageous for bronchiectasis [Engeler, 1994 #247; Lucidarme, 1996 #248] and for similar factors that apply to airway stenoses, characterization of bronchiectatic change is limited with axial planar interpretation alone. Transparerent bronchography-like images are used to depict the bronchiectatic change alone and classic lung parenchyma trapezoids show the bronchiectasis in relation to other lung changes such as interstitial abnormalities. Unlike airway stenosis, bronchiectasis is more commonly a multisegmental process and for this reason the whole 3D volume rendered whole lung images are well suited to mapping the disease process . In patients with cystic fibrosis, for example, the distribution of disease and its relative severity in different lung locations can be better depicted with selected oblique coronal and sagittal views. Whole lung display with clip planes orientated along airways of interest is superior for the appreciation of bronchiectatic change in an individual airway or segment of lung. This approach may also be valuable in the future to better assess response to therapies [Remy-Jardin, 1998 #224; Remy-Jardin, 1998 #223] and for mapping localized disease for possible resection.



3DCT and Diffuse Lung Disease

One of the features of many diffuse lung diseases is that they can affect the lung in a variable but characteristic patterns of distribution between the upper and lower lobes and can be quite asymmetric between lungs. Some conditions such as bronchiolitis obliterans and organizing pneumonia (BOOP) can have a ‘flitting’ pattern with spontaneous temporal change in distribution and all conditions may change their imaging features in response to therapy or due to progression. An appreciation of the whole lung distribution or the partial lung predominant distribution of such diseases is readily apparent with coronal and sagittal reformatted images that display the abnormal and normal lung in a single image. Subtle disease which may not be apparent on axial images may become more conspicuous when, for example, upper lung predominant disease is seen on the same imaging slice as normal lower lung. For interstitial lung disease volume rendered classic lung parenachyma-like settings are useful for appreciating the distribution of disease and minimum intensity projection (MinIP) images display the low densities of centrilobular emphysema, cysts, air trapping or ground glass opacities [Remy-Jardin, 1996 #221; Remy-Jardin, 1996 #222; Engeler, 1994 #247]. MinIP can be applied to multiplanar reconstructions or to user defined sliding slabs of data which reduce data from a greater proportion of lung to a single slice which may be more sensitive [Remy-Jardin, 1996 #221; Remy-Jardin, 1996 #222; Napel, 1993 #208; Bhalla, 1996 #156]. Remy-Jardin et al. suggested MIP may increase sensitivity to centrilobular micronodular change and may better delineate their location in the secondary pulmonary lobule compared with high resolution CT [Remy-Jardin, 1996 #221]. If prone, supine and dynamic inspiratory and expiratory images are performed they can be reformatted into three-dimensions for direct comparison. Together with multi-dimensional nuclear and CT ventilation and perfusion imaging with dynamic volume measurements [Mergo, 1998 #204; Brown, 1999 #160], CT may ultimately provide a comprehensive morphological, functional and quantitative [Park, 1999 #215] evaluation [Kinsella, 1990 #185; Park, 1999 #215; Kauczor, 1996 #183; Mergo, 1998 #204] of the entire lung with information on distribution of disease, gas transfer and perfusion derangements [Thurnheer, 1999 #237]. When lung volume reduction surgery is considered 3D reformatted images can be consulted when contemplating removal of an anatomic area of air trapping [Holbert, 1996 #178].



3D CT- Focal Lung Disease

There is much current interest in assessing MDCT and screening for lung nodules in the hope of reducing lung cancer mortality [Schoepf, 2001 #232; Obuchowski, 2001 #214]. The lung nodule interpretation involves assessment of morphology, density, activity (Fluorodeoxyglucose uptake), temporal change and contrast dynamics. However in smaller lung nodules much of the 2D morphological features of malignancy are not present and we must look to other features. 3DCT images may be a useful adjunct to the rotating whole body PET and CT PET images to accurately localize abnormal activity for diagnostic accuracy and planning resection. Nodules are often not perfect spheres and measurement differences between 2D and 3D may have significant clinical impact. 3D segmentation has been shown to have the spatial resolution to accurately measure nodule volume within 3% and to detect volume change with 30d for doubling times 30-180days and nodules larger than 5mm [Yankelevitz, 2000 #244]. Further it has been suggested there volume measurements may be more consistent with final pathology than 2D techniques [Yankelevitz, 2000 #244] and may be better suited to longitudinal follow-up to assess for potential malignancy and response to therapy.

For tissue diagnosis 3D CT further exploits the "CT bronchus sign", facilitating the path into lesions far in the periphery [Aoshima, 2001 #151; Midthun, 2001 #205] and choosing safe avenues for tissue sampling for transbronchial biopsy sampling [McAdams, 1998 #201; McAdams, 1998 #202], with identification of the shortest needle path into mediastinal nodes and those nodes amenable to endoscopic ultrasound guided biopsy. CT may help identify those with small lung nodules who are more likely to benefit from either bronchoscopic tissue sampling or percutaneous approach [Aoshima, 2001 #151; Midthun, 2001 #205]. It has also been suggested that virtual bronchoscopy guidance of ultra-thin bronchoscopes may improve the accuracy and duration of conventional bronchosopy for peripheral lung lesions [Asano, 2002 #154]. 3D renderings designed to depict medial lung neoplasms are more intuitive for assessing suitability for sleeve resection by showing the tumor in relation to the bronchus or pulmonary vessels in their entirety together with the medial extent of tumor. If a patient has poor respiratory reserve the 3DCT can be interpreted with the ventilation-perfusion for anatomic segmentation to quantify what proportion of normal lung will be removed by tumor resection and what lung parenchyma preserving approaches are feasible [Ravenel, 2001 #108; Wu, 2002 #243]. Though once the preserve of MRI current coronal and sagittal 3D images can better define superior sulcus tumors, differentiate basal lung tumor from hepatic dome masses and demonstrate the distribution of mesothelioma. Radiation therapy requires an anatomically accurate field for focal lung disease [Leibel, 1992 #195] and images have been found [Armstrong, 1998 #153] useful in planning the portal for delivery of external beam irradiation by providing measurements and thoracic landmarks that can then be applied to a conventional radiograph to maximize tumor dose and minimize injury to normal tissue.



3DCT-Chest Wall and Diaphragm

The orientation of the bone and muscle of the chest wall are not well seen with 2D imaging and the exact ribs and their orientation are difficult to localize so that any deformity of curvature is hard to appreciate or measure. 3D chest wall CT is requested to evaluate congenital abnormalities such as pectus excavatum or "acquired Jeune’s syndrome" [Haller, 1996 #176] or resection and reconstruction of chest wall tumors such as direct invasion of lung carcinoma, primary rib osteosarcoma or infection [Haller, 1996 #176; Pretorius, 1998 #216; Pretorius, 1999 #217; Pretorius, 1999 #218]. Optimal trauma renderings such as for sternal or clavicle fractures are better seen on sagittal or coronal reconstructions respectively and multidimensional views have long been favored for diaphragm imaging including for cases of rupture [Israel, 1996 #181; Brink, 1994 #158; Killeen, 1999 #184]. Volume rendering is preferred to maximum intensity projection or shaded surface display as it preserves the densities of costochondral cartilage, bone and muscle and does not require laborious editing [Kuszyk, 1996 #190; Pretorius, 1999 #217; Leitman, 1983 #196; Kuriyama, 1994 #189].



Part II-Aortic Dissection

Aortic Dissection-Introduction

The aorta has limited ways of expressing itself in disease states and though penetrating atherosclerotic ulcer, intramural hematoma and classic aortic dissection can each represent distinct pathological entities their clinical assessment is imprecise and their pathophysiology and morphological expression clearly lie upon an imaging spectrum [Waller, 1997 #140; Waller, 1997 #141; Waller, 1997 #142; Waller, 1997 #143; Waller, 1997 #144]. Transesophageal echo(TEE), MRI and CT have all proven efficacious in dissection imaging but CT is the initial test in two thirds of cases and is the second test in the majority of those who have multi-modality evaluation [Treasure, 1991 #134; Cigarroa, 1993 #20; Petasnick, 1991 #99; Flachskampf, 2000 #41; Hartnell, 2001 #51]. The accessibility of CT together with its rapid and accurate evaluation has changed our understanding of dissection disease and reduced its mortality [Erbel, 2001 #32; Erbel, 2001 #33; Ledbetter, 1999 #79; Novelline, 1999 #94]. Multidetector row CT (MDCT) acquisition and volume rendering techniques permit high quality CT angiography (CTA) through volume data acquisition and interpretation [Fishman, 1997 #39; Fishman, 2001 #40; Fishman, 1991 #38; Prokop, 2000 #107; Smith, 1998 #124; Smith, 1999 #125; Rydberg, 2000 #116; Rydberg, 2000 #117].



Aortic Dissection- MDCT Imaging Technique and Protocols



Careful attention to patient preparation is important in dissection imaging. Possible sources of beam hardening artifact should be removed from the field of view with alternate placement of metallic leads and placement of the arms above the head. 18-22G venous access through a right antecubital vein is preferred for power injection of contrast that causes minimal opacification of the left brachiocephalic vein as it crosses the aortic arch. Femoral vein injection [Prokop, 1993 #104] has also been advocated for this reason but is less favored by patients.

Routine coverage for thoracic aortic dissection CT is from the celiac axis to the proximal great vessels with caudal-to-cranial scan direction allows some clearance of high-density contrast from the upper thoracic veins [Rubin, 1995 #112]. Superior extension of the imaging can be performed for carotid CTA if there is concern for extension here. Likewise when there is concern for further dissection extension below the diaphragm or organ involvement or if there are plans for an endovascular approach The high pitch of MDCT and the heat loading capacity of its detector ceramics permits for imaging to the femoral vessels in a single acquisition during the contrast plateau [Katz, 2001 #69]. If abdomino-pelvic imaging is performed negative oral contrast is preferred for post-processing.

Non-contrast studies are not performed as a matter of routine but have value in discriminating displaced intimal calcification, displaying high-density acute intramural hematoma or depicting acute mediastinal or pleural blood. Most of these diagnoses will not however be missed with careful evaluation of a contrast enhanced study. On occasion non-contrast studies are requested to assess the burden of atherosclerosis prior to placing an intra-aortic balloon pump or when cross clamping the aorta is anticipated. Non-contrast studies are recommended for endovascular stents follow-up to assess for endoleak. When used, dose exposure for non-contrast studies may be reduced with 10mm slice widths, 2-3mm inter-slices gaps and high pitch values (~1.7-2.0) and low mAs (~50-80mAs) [Prokop, 1993 #104; Prokop, 2000 #107; Coulam, 2001 #24; Jeffrey, 1998 #58].

Dissection CTA is performed with 200-300 mA and 120-140kV depending on patient size. A maximum 1cc/kg of 350mg/L iodinated non-ionic contrast is administered at a rate of 3-5cc/second [Bae, 1998 #5; Bae, 1998 #6; Bae, 1998 #7; Prokop, 1993 #104; Prokop, 1997 #106; Prokop, 2000 #107] with power injector for optimal lumen opacification. Timing is a critical factor in accurate dissection interpretation). Larger contrast volumes or a saline chaser [Prokop, 1997 #106] do prolong the contrast plateau but with fast MDCT acquisition times are not necessary even for ilio-femoral vessel visualization unless there is large body habitus. 200HU measurements have been found with doses as low as 80cc [Costello, 1992 #398]and further refinement of MDCT techniques promises decreased contrast doses [Prokop, 2000 #107; Coulam, 2001 #24; Rubin, 2000 #114]. The initial HU within the aorta rises linearly with early injection. The contrast plateau is in fact an asymmetric hump formed during continued injection combined with the onset of contrast recirculation After the total volume has been administered there is a precipitous fall in density of the true lumen as slower opacification in the false lumen or systemic veins develops [Petasnick, 1991 #99; Bae, 1998 #5; Bae, 1998 #6; Bae, 1998 #7; Bae, 2000 #8; Burbank, 1984 #15]. The high injection rates do give high arterial contrast [Bae, 1998 #5; Bae, 1998 #6; Bae, 1998 #7] and though they do shorten the acquisition time window this is not usually a problem with the fast data acquisition of MDCT. A minimum true lumen contrast plateau of 2-300HU [Prokop, 1993 #104; Prokop, 1996 #105] should be consistently placed within the MDCT scan duration [Rubin, 1995 #112] of 13-20seconds using an empiric scan delay of 25-30seconds from the start of injection. Approximately 4 second scan delays are added for perceived diminished cardiac output [Bae, 1998 #5; Bae, 1998 #6; Bae, 1998 #7], manual injections (e.g. central lines) or injections in the far periphery such as the foot. Injections rates should be slowed for tenuous access or access in tight limb compartments (e.g. dorsum of hand). A timed bolus can be designed from a time density curve obtained from a 10-20cc test bolus administered at 4-5cc/s with sequential single level imaging every 2 seconds after an 8 second delay [van Hoe, 1995 #137]. Though tailored test bolus [van Hoe, 1995 #137] or automated bolus tracking and triggering [Prokop, 1996 #105; Prokop, 1993 #104; Prokop, 1997 #106; Prokop, 2000 #107] and tailored multiphasic injections techniques [Bae, 2000 #8; Fleischmann, 2000 #44; Fleischmann, 1999 #42; Tello, 1994 #133] are advocated by some and do give some numerical improvement in consistency, uniformity and duration of opacification they may require additional contrast and radiation dose [Kalender, 1994 #66]. Empiric delays will suffice in most cases for consistent opacification [Sheiman, 1996 #123; Macari, 2001 #81] and delineation of the flap and lumina. The peak opacification level may be increased with increased rate of administration or increased iodine concentration but there is little to be gained with values greater than 300HU and indeed in many cases the MDCT capture of the bolus is so good the windows need to be adjusted from the conventional settings so as not to miss the flap. Seventy -second delayed post contrast imaging is performed to assess for delayed flow in the false lumen or graft leak.

MDCT continues to build on the advances in breath-hold imaging made with slip ring technology [Rydberg, 2000 #116; Rydberg, 2000 #117; Mesurolle, 2000 #85][[Heiken, 1993 #53; Fleischmann, 2000 #43] Kalender, 1990 #63; Kalender, 1990 #64; Kalender, 1991 #65; Kalender, 1994 #66]. Rubin et al. have shown that multidetector systems benefit CTA of the arterial system in terms of scan duration, contrast use, thinner sections and cost savings [Rubin, 2000 #114]. in comparison with single detector systems. Multidetector row CT provides improved Z-axis spatial resolution for similar coverage and extended coverage for smaller resolution [Hu, 2000 #54; Berland, 1998 #11; Klingenbeck-Regn, 1999 #74; Rydberg, 2000 #116; Rydberg, 2000 #117]. Smaller slice widths and high pitch values provide higher quality images than larger slice widths with smaller pitch. Fusion of small slices into a broader slice minimizes the artifact of wider slice widths and diminishes inter-slice misregistration. For optimal 2D imaging and for a good 3D substrate a beam collimated to four 1mm detectors of an adaptive array design (~ four 1.25mm matrix array detectors) is employed to provide 1.25mm slice widths with overlapping 1mm reconstruction interval and a 25-30cm field of view [Fleischmann, 2000 #43]. The higher MDCT pitch values (pitch defined as table increment per gantry rotation divided by single slice collimation) required for aorta imaging come without the single detector penalty in widened slice sensitivity profile and effective slice width through [Brink, 1992 #13] the use of multiple detectors and weighted interpolation [Fuchs, 2000 #45; Klingenbeck-Regn, 1999 #74]. Temporal resolution of 130-250ms is attainable with simultaneous 4-detector acquisition systems reducing pulsation artifact somewhat. The dissection flap is clearly defined by a 512 in plane matrix and narrow z-axis resolution. Tube current and patient dose exposure is modulated according to the geometric of body habitus with diminished dose for antero-posterior imaging compared with transverse imaging. The penumbra of focal spot wobble of the MDCT cone beam of wasted irradiation is reduced further as a proportion of larger detector arrays such as 16 detector compared with 4 detector systems. Near-istropic data with overlapping reconstruction suffices for most dissection evaluations with 4 detector systems though 16 detector designs will likely permit routine true-isotropic data with 0.5mm detectors. For patients who are poor breath-holders or for those in whom greater coverage is required 2.5mm detectors can be used and the image quality is still better than single detector systems. For hard-copy interpretation larger slice widths are assimilated from smaller ones to limit the amount of film required for the 5-600 slice acquired. Data is sent to a PACS workstation for scrolling and also sent to a 3D workstation for post-processing.



Aortic dissection CT image processing and data evaluation

3D CTA has become a useful adjunct to a well-performed 2DCT for dissection evaluation [Rubin, 1995 #112; Rubin, 1996 #227; Rubin, 1996 #228; Adachi, 1994 #3; Zeman, 1995 #150; Kalender, 2001 #67]. Whatever post-processing techniques are employed to generate 3DCTA they must be real-time (frame rate over 20F/s) without laborious editing for clinically practicality and current options include; multiplanar (MPR) and curved multiplanar (CMPR) reconstructions, maximum intensity projection (MIP), shaded surface display (SSD) and volume rendering techniques (VRT) [Kalender, 2001 #67; Addis, 2001 #4; Prokop, 2000 #107; Johnson, 1996 #59; Johnson, 1998 #60; Remy-Jardin, 1999 #225; Wu, 1999 #149]. Multiplanar (MPR) and curved mulitplanar (CMPR) techniques are on many CT consuls and require minimal computer power [Prokop, 2000 #107]. These are not true 3D techniques but rather a reordering of voxels into a tomographic section, which can be of value but are often limited by a tortuous non-planar aorta and dissection flap [Zeman, 1995 #150]. Maximum intensity projection (MIP) [Schreiner, 1996 #399]displays only density values above a chosen threshold and reduces the length of data encountered by the ray to a single plane. MIP is akin to digital subtraction angiographies (DSA) projectional technique and lacks depth cues without movement of and can be problematic amongst the complex enhancing mediastinal vascular structures. The chest wall must be edited from the projection or it will be included in the reconstruction of the vessels. Shaded surface display (SSD) will reconstruct the data of an assigned interface through polygon building blocks and uses lighting models for perspective. Both MIP and SSD use only 10-20% of the acquired data but will obscure the intimal flap and internal lumen information on thrombus burden [Zeman, 1995 #150].

Volume-rendered techniques are now preferred for 3D CTA of dissection [Prokop, 2000 #107; Smith, 1998 #124; Smith, 1999 #125; Johnson, 1996 #59; Johnson, 1998 #60; Fishman, 1991 #38; Fishman, 1997 #39]. This method is of high fidelity to the original data set and all density values are preserved in the final segmentation [Calhoun, 1999 #16; Johnson, 1996 #59; Johnson, 1998 #60; Remy-Jardin, 1999 #225]. Since volume rendering uses a percentage based classification of the values encountered by the ray including the intimal flap, slow flow or thrombus are all represented as well as areas of volume averaging. Areas of interest are such as intimal calcification, thrombus and flaps are highlighted through trapezoid histogram manipulation of level and width and opacity and brightness [Wu, 1999 #149]. MDCT acquisition together with infinite VRT clip planes editing, projections and trapezoids permits visualization of the variable aorta, dissection flap and branch vessels with perspectives that are independent of the acquisition plane and tailored to the individual patient. When there are associated aneurysms true orthogonal measurement can be obtained and consistent reproducible views can be generated for follow-up studies. Though endoscopic modes for 3D display can be performed they may not add to the 2D interpretation [Smith, 1998 #124; Kimura, 1996 #72]. Conventional perspectives such as anterior and left anterior oblique view which the surgeons are used to are routinely produced. These are supplemented with the unique tailored views for the individual patient and include information on branch vessels and end-organ supply where indicated. With 3D post-processing surgical consultation with simulation of the possible approaches to therapy becomes part of routine practice and tables of operatives measurements required become standard.





Dissection Epidemiology and Pathophysiology



As we age the elastic fibers of the aorta fragment and the smooth muscle cells diminish in number with increasing collagen fibers and mucoid ground substance resulting in multidimensional dilatation (ectasia). Atherosclerosis results in lipoid deposition, fibrosis and calcification of the tunica intima with underlying tunica media atrophy [Waller, 1997 #144].

Classic aortic dissection is defined as a disruption of the aortic wall, forming and intimal flap and therefore separating a true and false lumen [Erbel, 2001 #33; Nakashima, 1990 #92]. It is the most common catastrophe of the aorta with rupture occurring twice as commonly than with aneurysm [Miller, 1984 #425]. Most risk factors for dissection are those entities that cause some weakening of the aortic wall. 90% are hypertensive, which increases transmural pressure and may lead to aortic sclerosis and stiffer vessels more vulnerable to wall stress [Erbel, 2001 #32; Erbel, 2001 #33; Coulam, 2001 #24]. Other risk factors include congenital (connective tissue disorders, valve abnormalities, cystic medial necrosis), atherosclerosis (PAU or IMH), inflammation[DeSanctis, 1987 #403], toxic agents (e.g. cocaine)[Chang, 1995 #409] and trauma (acute traumatic aortic injury (ATAI) or iatrogenic) [Crawford, 1988 #402; Fisher, 1992 #406], third trimester of pregnancy[Oskoui, 1994 #405] and more rare conditions such as neurofibromatosis [Chew, 2001 #18]. Less than 10% of Marfans patients suffer dissection but it accounts for a large proportion of their deaths. Traumatic deceleration dissections tend to occur at points of fixation such as the aortic root or isthmus an account for up to 20% of high-speed deceleration fatalities [Patel, 1998 #97]. Iatrogenic etiologies are largely due to catheter manipulation, cardiac bypass or balloon pump or cross clamping. Dissection is thought not to be related to atherosclerois unless the sequela of PAU or IMH thus explaining its preponderance in the relatively atherosclerosis free ascending aorta [Roberts, 1981 #111; Nakashima, 1990 #92]. It is an acute dissection if diagnosed within 14days of symptom onset and beyond that are considered chronic [Sebastia, 1999 #122]. It has a prevalence of 0.5-2.95/100,000/year with and incidence of 2.95/100,000 [Asfoura, 1991 #411; Fuster, 1994 #413; Fuster, 1999 #412] The mortality is 3.25-3.6/100,000, three times higher in men than women with a high acute mortality of 1.4% per hour. The overall mortality rate is 21% within the first 24 hours, 60% within two weeks and 90% within 3 months [Dmowski, 1999 #28; Pretre, 1997 #103[Fowkes, 1989 #414; Meszaros, 2000 #415]. 80% of deaths are due to aortic rupture whereas tamponade, torrential aortic regurgitation and coronary or carotid compromise accounting for the rest. Thinning of the aortic wall, aneurysmal dilatation, hypertensive transmural pressure and cystic medial necrosis associated with dissection are all factors that can predispose to rupture according to the Law of Laplace. Aortic rupture is found in 0.9% of cases of sudden death and 62% of these patients are found to have dissection [Young, 2002 #416].

The dissection usually begins as a thoracic transverse intimal tear involving 25-50% of the circumference in most patients [Waller, 1997 #144]. Primary abdominal dissection tears are rare but for iatrogenic etiologies. There may be an entry site and a more distal re-entry site and these sites may be multiple allowing multi-directional flow [Mohr-Kahaly, 2001 #89] with a pressure gradient of 10-25mmHg [Erbel, 2001 #33]. Blood flow and dissection propagation tends to be antegrade and re-entry tears are commonly in the abdominal aorta or iliac vessels. Retrograde dissection occurs in less than 20% of patients [Erbel, 1993 #417]but fibrous pericardium is blended with the ascending aortic wall and both the ascending arch and pulmonary trunk is contained within a common sleeve of serous pericardium. The distal extent of a dissection flap is thought to be limited by atherosclerosis, media scarring or atrophy, branch points or coarctation [Waller, 1997 #144; Roberts, 1981 #111]. In Marfan patients with limited scarring of the aorta one will commonly find the entire aorta involved.

The false lumen is located in the outer half of the media giving a thinner outer wall than inner wall [Waller, 1997 #144] and it has high pressure and wall stress [Erbel, 2001 #33]. Since outer wall of the false lumen lacks smooth muscle so it tends to be larger and to compress the true lumen False lumen flow may be less than the true lumen, equal to it or thrombosed and rarely has greater flow than the true lumen. Though false lumen may provide blood flow to organs, thrombosed false lumens are generally thought to be more stable. By definition the wall of the false lumen is thinner than the normal aorta and thus according to the Law of Laplace more prone to continued expansion and pseudoaneurysm formation under the influence of systemic transmural pressure. Aneurysms with communicating dissections will enlarge at a rate of 2-3mmmm per year with systemic transmural pressure and those with non-communicating dissection will enlarge at a rate of 1mm per year [Erbel, 2001 #33].





Dissection Classification



Dissections were initially categorized based on site, which largely dictates management. The Stanford classification [Daily, 1970 #25] is the most widely used where all dissections involving the ascending aorta are Type A regardless of the site of tear or distal extent (same as DeBakey Type I/II) [DeBakey, 1982 #27] and all dissections limited to beyond the left subclavian artery origin are Type B regardless of distal extent (same as DeBakey Type III [DeBakey, 1982 #27]). The DeBakey classification further subdivided the Type B dissections into those that involve the thoracic aorta only (IIIa) or both the thoracic and abdominal aorta (IIIb). 65% of dissections are of the ascending aorta with an entrance tear 2cm above the sinotubular junction [Waller, 1997 #144], 10% involve the arch, 20% affect the descending thoracic aorta and 5% are abdominal. One year survival for Type A dissections is around 30% with 60% mortality at 24 hours [Dmowski, 1999 #28] representing their greater propensity to propagate retrogradely into the pericardium, cause coronary arteries, torrential aortic valve regurgitation or carotid flow compromise. Type B dissections fare better at one year with 85% one-year survival [Erbel, 2001 #32; Erbel, 2001 #33; Glower, 1991 #418; Masuda, 1991 #420; Eagle, 1989 #397]

Svenson more recently attempted to classify dissections based on etiology in an attempt to reconcile the multiple conditions that are thought to overlap with classic aortic dissection [Svensson, 1999 #130] (FIGURE). This is a useful classification for clinical interpretation as it goes beyond the morphological system alone. Under this classification a Class I dissection is the classic aortic dissection with identifiable flap and true and false lumen with or without communication of the two lumina. A Class II dissection represents an intramural hematoma (IMH). Class III dissections are a subtle wall bulge and are thought by some to represent an abortive dissection with a subtle tear and healing thrombus overlying it. Under this classification penetrating atherosclerotic ulcer (PAU) is named a Class IV dissection. Finally tears in the intima due to iatrogenic or non-iatrogenic trauma are assigned to Class V dissections though they can lead to class I or II dissections.





Class 1,3,4 Dissection-CT Imaging Interpretation



There are two fundamental rules for aortic dissection CT interpretation; (1) survival is directly related to the timeliness of diagnosis and (2) imaging should always be finally interpreted in the imaging context for correct triage and management though all suspected dissections must be initially treated as acute potentially fatal processes. A comprehensive CTA for aortic dissection demands high accuracy of diagnosis and classification, localization of tears and documentation of the nature of false lumen flow. The status of branch vessels and the viscera or bowel they must be established. The need for therapy and the type of surgical or endovascular approach are in large part determined by this CT interpretation MDCT and 3D Volume rendering may not add greatly to the established sensitivity and specificity of SDCT but will add to the sophistication and precision of interpretation and measurements.

The intimal flap usually appears as a single homogenously low-density linear or curvilinear structure with or without intimal calcification. The flap can appear more complex or appear as more than two lumens with more extensive tears, intimal redundancy or greater in-folding [Sebastia, 1999 #122; Karabulut, 1998 #68]. It may also appear as falsely as multiple flaps due to pulsation artifact but these are usually apparent from their parallel nature. The flap tends to spiral postero-laterally for a variable distance and can extend into and divide among the ostia of branch vessels[Quint, 1996 #437] [Zeman, 1995 #150].

Accurate discrimination of the true and false lumen has become increasingly important with the use of interventional therapies. The true lumen is usually along the inner curvature of the aorta in continuity with the diseased aorta. It is often under lower pressure than the false lumen and so may be partially compressed. The false lumen tends to be larger and under higher pressure though it may be compressed in systole. It tends to be posterolateral in the chest and posterior at the level of the common iliac arteries and is noted deep to any intimal calcification. Calcification may be a useful sign for acute dissection but becomes less reliable with chronic dissections where re-endothelialization occurs and can calcify within the false lumen [LePage, 2001 #80]. It has a more variable flow rate dependant on the size of the entry and re-entry tears and this can manifest as a swirling pattern of contrast or thrombus formation in the form of the ‘beak’ sign, ‘cobweb’ sign or complete thrombosis [LePage, 2001 #80]. It is important to document whether the dissection is communicating (flow present in the false lumen) or non–communicating. The former may serve to supply some branch vessels but have a greater tendency to progress and rupture.

CTA sensitivity and specificity is over 95% for assessment of branch vessel involvement by the dissection process [Sommer, 1996 #424; Nienaber, 1999 #93]. The true lumen usually supplies the celiac axis, superior mesenteric artery and right renal artery with the false lumen supplying the left renal artery [Coulam, 2001 #24]. True or false lumen caliber may be compromised in diastole or systole respectively, On occasion the false lumen remains as the sole supply of blood to the organ and it may be insufficient due to slow flow or obstruction. There are two types of branch vessel compromise: static and dynamic [Sebastia, 1999 #122]. Static obstruction is due to the actual dissection flap or thrombosis extending into the vessel ostium and dynamic obstruction is due to a ball-valve effect of the intimal flap covering the ostium. Such branch vessel compromise is related to poor outcome due to dissection [Cambria, 1988 #17]. Renal artery involvement is most commonly reported whereas clinical mesenteric ischemia is found in 2.5% of cases and SMA or celiac involvement is found in 10% of autopsies [Cambria, 1988 #17]. Dissection mortality is due to renal insufficiency in 50-70% [Cambria, 1988 #17],[Laas, 1992 #450] of patients and mesenteric ischemia in 87% [Cambria, 1988 #17]. Asymmetric renal perfusion is suggestive of but not confirmatory for renal ischemia and bowel ischemia is suggested by thickeing, abnormal enhancement and mesenteric stranding with infarct suggested by pneumatosis or air in the portal system. Any involvement of iliofemoral vessels must be carefully documented to aid planning of endovascular approaches and the relatively small ilio-femoral vessels can be occluded with diastolic collapse of the true lumen, which may require a bypass procedure. Those with peripheral vascular compromise have high mortality [Fann, 1995 #36] [Miller, 1984 #425; Miller, 1991 #400]. Coronary artery involvement is rare but cardiac ischemia is seen in 3-4% of patients, which can be multifactorial [Erbel, 2001 #33].

Rupture features include an irregular aortic wall, high attenuation blood around the aorta or in the serosal spaces or active extravasation of contrast. Pericardiac, mediastinal or pleural fluid do not necessarily imply impending rupture but must be give great attention as they are frequently seen in cases of sudden death [McDonald, 1981 #426; Kaji, 1999 #427] and have been associated with higher mortality [Erbel, 2001 #33]. Fluid greater than 1cm around the aorta separating it from the left atrium and esophagus or pericardial effusion separating epicardium and pericardium are associated with 50% mortality [Kaji, 1999 #427].

Acute traumatic aortic injury (ATAI) from blunt chest trauma may manifest as a class I, II or V dissection. When there is a focal transection it is a transverse laceration and may appreciated on axial images as a focal wall irregularity and caliber change and on 3D imaging may demonstrate a distinctive kink in the aorta and focal outpouching [Patel, 1998 #97; Kuhlman, 1998 #77; Marotta, 1996 #84; Gavant, 1996 #48; Gavant, 1995 #47; White, 1995 #146].

A discordant CT result with no evidence of class I classic dissection in the presence of classic history, symptoms and signs should lead to a search for more subtle IMH or PAU variants of dissection as well as other causes of acute chest pain.



Class 2 Dissection-Aortic Intramural Hematoma



The tunica intima and inner one third of the tunica media are supplied by diffusion of nutrients from the circulating blood with the vasa vasorum supply remaining wall. Aortic intramural hematoma (IMH) is hemorrhagic dissection of the media (Class 2 dissection) without and intimal tear (MOHR) or intramural flow [Banning, 1995 #428; Harris, 1994 #50] and will remain occult on conventional angiography. IMH has been found in 13-27% of suspected dissection cases [Song, 2001 #126; Vilacosta, 1997 #139; Mohr-Kahaly, 2001 #89; Nienaber, 1999 #93; Kang, 1998 #430]. Krukenberg in 1920 first ascribed this to spontaneous rupture of the vasa vasorum in the outer layers of the media. Traumatic etiologies have since been described since been described [Murray, 1997 #91]. There are two non-traumatic etiological types: cystic medial necrosis Erdheim-Gsell or atherosclerosis [Mohr-Kahaly, 2001 #89]. Traumatic tend to have a better prognosis than non-traumatic and all do worse in aneurysmal aortae [Vilacosta, 1997 #139].Unlike class I dissection AIH occurs in older hypertensive patients with in an equal male: female ratio [Sawhney, 2001 #119] and is found in an atherosclerotic milieu so it is more likely to occur in the descending thoracic aorta. It is classified and managed like Type A and B classic dissections and carries a high mortality (~80%) with a wide range reported [Ide, 1996 #55; Alfonso, 1995 #431]and higher mortality for Type A (50-100%) than Type B (20-40%) AIH [Mohr-Kahaly, 2001 #89; Maraj, 2000 #83] usually due to rupture through the adventitia. The natural history is that 15-41% will convert to class I dissection, 5-26% will rupture and 11-75% will spontaneously heal [Mohr-Kahaly, 2001 #89; Sueyoshi, 1997 #129; Erbel, 2001 #33; Maraj, 2000 #83; Ide, 1996 #55; Murray, 1997 #91; Vilacosta, 1997 #139], with progression more likely in aorta greater than 5cm [Choi, 2001 #19](KAJI). The hematoma may progress longitudinally and in a circumferential manner and by weakening the aortic wall can subsequently lead to aneurysmal dilatation [Vilacosta, 1997 #139; Sueyoshi, 1997 #129[Ohmi, 1998 #432; Kaji, 1999 #427] Findings such as type A IMH, thick hematoma, compression of the true lumen or the presence of pleural or pericardial effusions are thought to predict progression to overt dissection [Choi, 2001 #19]. AIH can be occult to DSA and may not be appreciated with SSD or MIP techniques.

On axial 2D and 3D volume rendered interpretation it appears as a circular or eccentric discrete crescent segmental thickening of the aortic wall (>0.5cm) [Mohr-Kahaly, 2001 #89; Sommer, 1996 #423] which may be high attenuation on non-contrast study, may displace the intimal calcification and does not enhance. It maintains a constant anatomical relation to the aorta and may extend over multiple slices but does not cause significant lumen compromise. Associated pleural (75%) and pericardial (88%)effusions are common [Mohr-Kahaly, 2001 #89; Sueyoshi, 1997 #129] and though not necessarily indicative of a leak all should be viewed with caution for impending rupture. IMH may be distinguished from mural thrombus by its site under the intima and its smooth inner surface but it can be difficult to distinguish from a class I dissection with thrombosed false lumen (non-communicating dissection) [Song, 2001 #126; Sommer, 1996 #423]. The thickness of the aortic wall will vary over time reflecting resorption or progression [Vilacosta, 1997 #139] and resolution can occur within 7-10 days [Mohr-Kahaly, 2001 #89; Kaji, 1999 #427]. The full longitudinal extent is best appreciated on sagittal 3D reconstructions.



Class 4 Dissection-Penetrating Atherosclerotic Ulcer



Penetrating atherosclerotic ulcer (PAU) like IMH is an entity appreciated on CT though often occult with aortography [Quint, 2001 #435; Stanson, 1986 #128; Cooke, 1988 #23; Hayashi, 2000 #52; Welch, 1990 #145]. It is a class 4 dissection by Svensson’s criteria due to a focal violation of the internal elastic lamina by plaque rupture and washout of the lipid core. It is principally a disease of elderly men [Troxler, 2001 #135] first ascribed to atherosclerotic ulceration of the internal elastic lamina by Stanson in 1986 with an incidence of 2.3% of patients who had aortography for suspected dissection.

On CT PAU is typically manifest as a single focal ‘collar button’ ulcer with perhaps a small rind of thrombus, extending beyond the expected circumference of the aortic wall [Kazerooni, 1992 #71; Hayashi, 2000 #52; Vilacosta, 1997 #139]. There is no dissection flap or false lumen and it is seen usually in the setting of extensive atherosclerotic disease of the aorta. It tends not to extend far longitudinally unlike IMH possibly due to the limitations imposed by the atherosclerosis present. PAU is uncommon in the ascending aorta and most are found in the mid-descending thoracic aorta [Kazerooni, 1992 #71]. Axial 2D and cine scrolling is best suited to their detection but their anatomic location is best depicted with 3D. PAU has a high incidence of complications [Coady, 1999 #22] and it is thought to be more likely to rupture than class 1 classic aortic dissection [Vilacosta, 1997 #139; Stanson, 1986 #128; Welch, 1990 #145; Cooke, 1988 #23; Troxler, 2001 #135]. Though the risk of PAU rupture is greater in larger aortae [Troxler, 2001 #135]or in the ascending aorta [Kazerooni, 1992 #71; Movsowitz, 1994 #90]it may occur in normal sized aorta too [Vilacosta, 1997 #139]. It may spontaneously resolve, progress to IMH (87%)or pseudoaneurysm (26%) or rupture (8%) (MOVSOWITCH) and 10-20% become class 1 aortic dissections [Coady, 1999 #22; Vilacosta, 1997 #139; Kazerooni, 1992 #71; Harris, 1994 #50[Hussain, 1989 #433; Quint, 2001 #435]. Any sign of progression or embolic phenomena may require resection of the ulcer and graft interposition [Cooke, 1988 #23]. Endovascular repair has been reported with some success [Dake, 1999 #26; Mergo, 1998 #204].



DISSECTION CTA-THE DATA

CTA for dissection is accurate (~96%-100%) with sensitivity over 90% and specificity over 85% for detection even with older technologies [Nienaber, 1999 #93; Sebastia, 1999 #122; Vasile, 1986 #138[Sommer, 1996 #424]and it is seldom compared to digital subtraction angiography (DSA) any longer. The negative predictive value of a normal study is close to 100% though minimal aortic injury (MAI) [Malhotra, 2001 #82] which represents a slight intimal disruption without blood in the media can be overlooked. CTA is changing our understanding of diseases and can detect injuries limited to the intimal and inner media [Malhotra, 2001 #82]. This latter condition however tends to be self-limited. CT has proven efficacious in the emergency room setting with rapid triage of those with traumatic and non-traumatic dissections [Novelline, 1999 #94; Gavant, 1995 #47; Gavant, 1996 #48] and 3D CTA is useful in preoperative assessment [Adachi, 1994 #3]. Even with the accuracy of today’s CTA false positives and needless sternotomies exist though they are uncommon. Four detector acquisition systems are not adequate for coronary flow assessment but 16 detector systems show great promise in this area. TEE is still favored to quantify and establish the flow direction across the proximal intimal tears and both TEE and MRI can better assess aortic regurgitation. CT is reliable with 90%-100% sensitivity, 82-99% specificity and 47-90% positive predictive value [Mirvis, 1996 #87; Gavant, 1995 #47; Gavant, 1996 #48]in excluding aortic injury in the trauma setting [Dyer, 1999 #31] and diagnosing it [Mirvis, 1996 #87] though aortography may still be required in the setting of extensive mediastinal hematoma and no evidence of ATAI of obvious cause for bleeding [Scaglione, 2001 #120]. Small or subtle injuries can be problematic for all modalities [Patel, 1998 #97], but may not have great clinical significance.

False positives may be minimized by attention to suboptimal technical factors, recognition of common pitfalls including artifact and periaortic normal structures and variant anatomy [Batra, 2000 #9]. Streak artifact is one of the more common causes of a false positive diagnosis due to either metallic leads, arms or venous streak artifact due to transmitted pulsation and beam hardening at high contrast interfaces. Proper patient preparation, appropriate scan direction and larger volumes of lower concentration contrast will minimize some of the more common causes. Streak artifact can be discriminated by its arbitrary trajectory on sequential slices in parallel lines or as a radiation from a single point and on occasion it can be clearly followed beyond the vessel wall. They tend to be less curvaceous than the intimal flap. If it remains a problem multiplanar reconstruction (MPR) or alternate 3D reconstructions may help.

Ghosting artifact (aortic motion artifact) [Loubeyre, 1997 #439; Loubeyre, 1997 #440; Duvernoy, 1995 #30; Posniak, 1993 #101; Posniak, 1993 #102] due to vessel wall misregistration is the next most common false positive seen on almost 60% of patients usually between 12 and 1 O’Clock or between 6 and 7 O’Clock [Duvernoy, 1995 #30]. It occurs because non-gating scanning is occurring arbitrarily in the systolic and diastolic phases of the cardiac cycle R-R interval and there is up to 1cm motion artifact of the ascending aorta even with gantry rotation times of 0.5seconds. It is usually not contiguous on serial images and is curvilinear and ill defined often sited beyond the aortic wall. Sagittal 3D reconstructions may reveal the repetitive nature of this artifact on serial images. Ghosting artifact can be minimized with the improved temporal resolution of multidetector row CT or 180-degree linear interpolation [Loubeyre, 1997 #439; Loubeyre, 1997 #440]. Prospective or retrospective ECG gating will also reduce this artifact. As yet CT cannot precisely evaluate and quantify aortic valve or coronary artery involvement and though the intima tears may be localized their size, quantity of flow and flow direction cannot be accurately estimated estimated. An imaging result, regardless of modality, should never be treated in isolation. The symptom of dissection pain in particular is an important measure of the significance of many signs on CT.





Dissection Therapy and Follow-Up



Follow-up studies are recommended at 1,3,6 and 12month intervals after the acute event. Surgery is thought to reduce the mortality of type A dissections [Kouchoukos, 1997 #76; Chang, 2001 #408; Walker, 2001 #447; Kang, 2001 #429]. For blunt chest trauma surgery is indicated to prevent transection. Surgery has a 20-35% perioperative mortality[Chang, 2001 #408] [Hagan, 2000 #49], which has been reduced 50% by newer imaging techniques [Erbel, 2001 #33]. Many type B dissections may spontaneously heal leaving only wall thickening in 31% [Hara, 1991 #441]. Dissections proximal to the left subclavian are usually treated surgically by graft or composite valved conduit graft [Cigarroa, 1993 #20; Riley, 2001 #110; Hagan, 2000 #49] [Fann, 1995 #36] [Chang, 2001 #408]which aims to replace the entire root, occlude the entry tear, induce false lumen thrombosis and prevent the 2% retrograde extension that can occur [Mohr-Kahaly, 1989 #455]. If the root is not compromised repair may be limited to the area of abnormality [Borst, 2000 #442; Lemole, 1995 #443; Najafi, 1976 #444]. This conventional wisdom has been challenged with the use of medical therapies for type A dissections. The false lumen is patent after surgery in 90% of cases [Erbel, 2001 #33] and should be monitored for progression [Ergin, 1994 #34]. Dissections limited to the descending aorta can usually be treated by medical therapy aimed at reducing the underlying hypertension with surgery considered in the setting of persistent pain, aortic expansion or any suggestion or rupture or organ compromise [Chang, 2001 #408; Waller, 1997 #144; Cambria, 1988 #17] [Walker, 2001 #447; Elefteriades, 1999 #451; Elefteriades, 1992 #452; Laas, 1992 #450]. Surgery for type B dissections has a high rate of mortality and paraplegia [Hagan, 2000 #49]. When surgery is required PAU are usually treated with interposition graft [Movsowitz, 1994 #90; Harris, 1994 #50]. AIH is treated much like class I dissection but older patients are thought to do better on medical therapy as the atherosclerotic process may limit expansion [Mohr-Kahaly, 2001 #89].

The use of endovascular approaches is in the early stages [Bortone, 2001 #12; Nienaber, 1999 #93; Milner, 2001 #86; Taylor, 2001 #132] but it has already been applied for thoracic aneurysm and trauma [Mita, 2000 #88; Taylor, 2001 #132]. Stents have been applied for class I, IV and V acute and subacute type B dissections [Bortone, 2001 #12; Sailer, 2001 #118; Taylor, 2001 #132; Erbel, 2001 #33; Erbel, 2001 #32; Ide, 2001 #56; Nienaber, 1999 #93; Dake, 1999 #26] [Kang, 2001 #429]to occlude the entry tear. They have been used to fenestrate a flap to cause communication between true and false lumen that is causing dynamic limitation of organ blood flow [Laas, 1992 #450; Elefteriades, 1990 #454] [Williams, 1997 #147] with operative mortality 21-67%[Williams, 1997 #147] [Walker, 1993 #448; Saito, 1992 #453]or to create a re-entry tear where false lumen thrombosis would lead to organ compromise. Stents are not used to push the intima back but to occlude entry tears, buttress a flap and prevent propagation into a branch vessel (LEE) or to maintain vessel patency in the face of an encroaching false lumen static obstruction [Williams, 1997 #147]. Endovascular stent has been suggested for PAU patients considered too grail for surgery [Troxler, 2001 #135]. The blooming artifact of wire components can limit MRI studies for follow-up but are less of a problem for CT. It is important in these cases to assess for failure of the stent to prevent progression or complication of dissection and to evaluate the stent for failure such as endoleak, migration, thrombosis. Ninety percent of false lumen will be patent at follow-up and 20% will have gone on to pseudoaneurysm formation [Kawamoto, 1998 #70] and less commonly aneurysm of the true lumen. Survival rates for Type A dissections surgically repaired are better when the false lumen is thrombosed [Ergin, 1994 #34]. 15-30% of those surgically treated for aortic dissection will have aortic disease leading to life-threatening conditions that require surgery after 10 years. Large re-entry tears promote flow preventing thrombosis and emobolism formation [Elefteriades, 1992 #452]but also increase the likelihood of pseuodaneurysm formation and rupture. Non-communicating dissections are thought to be more stable over time [Flachskampf, 2000 #41]. Aneurysms of the aorta in association with dissection enlarge at a rate proportionate to their size and they require 3D true orthogonal measurements to accurately detect 1-3mm changes. Through 3D post-processing consistent views may be produced that are better for assessing for stent migration and kinking or device failure. IMH is treated for the most part like Type A and B class I dissections depending on its site [Mohr-Kahaly, 2001 #89] though there have been some recent reports suggesting medical therapy can be successful for selected cases affecting the ascending aorta [Song, 2001 #126] .The rate of re-operation is 10% at 5years with a higher risk in Marfan patients. Particular attention must be given to measurements and angulation for thoracic stent placement [Milner, 2001 #86].





Conclusion

The present and future of CTA of aortic dissection rests with multidetector row CT acquisition and volume rendering post-processing which are inter-dependant. It is hoped advances in MDCT imaging will soon allow simultaneous assessment of the coronary arteries [Schoepf, 2001 #232; Schroeder, 2002 #121; Knez, 2001 #75; Janowitz, 2001 #57; Cline, 2000 #21] [Becker, 2000 #10; Fallenberg, 2002 #35; Achenbach, 2000 #1] and that volume rendering will aid the development of endovascular techniques through greater precision of measurements and pre-procedure triage. Current CTA assessment of aortic dissection involves high quality data acquisition with MDCT and post-processing techniques with interpretation through a combination of axial two-dimensional planar images and 3D volume rendering with customized images. CTA is hopefully improving the outcome of aortic dissection, permitting the exploration of new minimally invasive therapies and changing our understanding of the spectrum of disease including minimal aortic injury and aortic intramural hematoma. There is no role for CT screening for dissection at this time for selected high-risk groups such as Marfan’s patients. Studies have looked at aneurysm screening but found low compliance rates. Most developing whole body screening protocols currently in use are non-contrast studies diminishing their utility for detection of dissection. Most dissections have the type of pain that causes a presentation for evaluation and but for the small portion of patients with IMH or PAU there is not a precursor that can be detected by screening. Anything in the ascending aorta is high risk. Changing disease understanding.



Radiation dose strategies



Pediatric body habitus

Judicious use of multphasic scans

Screening





Conclusions.

Any current assessment of 3D airway CT must address the latest developments in MDCT and volume rendering post-processing, which are the imaging standard for the foreseeable future. This imaging advance has paralleled major progress in bronchoscopic technology and clinical applications. 3D volume rendering of the airway is only as good as the original CT data and the advances in MDCT of the airway cannot be fully realized on 2D planar images alone. A successful study requires appropriate protocol design and radiologist experience in the use of rendering software and the virtual imaging concept of anatomy and disease [Neumann, 2000 #209]. Virtual bronchoscopy alone is only one aspect of volume rendered 3D airway CT and does not fully reflect the capabilities of this technique which allow novel imaging perspectives ideally suited to airway imaging. Conditions which span a series of planar images may be assimilated in a single volume rendered illustration and conditions which only involve a single planar slice may be better depicted on an alternate projection. 3D airway imaging has a role in a number of defined airway conditions as an adjunct to axial planar studies and as a complement to bronchoscopic examination though it can also now be performed with ease in addition to routine MDCT imaging. The optimum application of this rapidly evolving technology requires further investigation, with particular focus upon patient outcomes. In these clinical investigations, as in the care of the individual patient, the dialogue between the radiologist and bronchoscopist will remain essential. Thus, images that depict the entire airway data and its extra-luminal relationships in ways that are unique to the capabilities of CT, rather than virtual bronchoscopy images alone may better serve the complementary attributes of each modality applied to each patient’s unique needs. When high quality data sets are routinely acquired it becomes less of a question of 3D airway added to a 2D study and more of a question whether additional post-processing is performed. Follow-up (HAPONIK, KAUCZOR). Structure function and volume. It is hoped many of these new approaches will improve upon the already high diagnostic accuracy of helical CT through more accurate and sophisticated interpretation of the data sets in ways that bear closer relation to clinical investigations including bronchology. MDCT challenge for data management (RUBIN). 3D CT of airway requires appropriate use of protocols and experience [Neumann, 2000 #209]. Managing data [Rubin, 2000 #113]