CT Basics Module 4 Transcript PDF

CT Basics Module 4 Transcript ©2017 ASRT. All rights reserved.  CT Basics Module 4: Image Processing and Reconstruction 1. ASRT Animation 2. CT Ba...

CT Basics Module 4 Transcript ©2017 ASRT. All rights reserved.  CT Basics Module 4: Image Processing and Reconstruction 1. ASRT Animation 2. CT Basics: Module 4 Image Processing and Reconstruction 3. License Agreement 4. Module Objectives After completing this module, you’ll be able to:  Describe the steps required for computed tomography image reconstruction.  List the postprocessing techniques needed for image enhancement.  Define the tools used to view a CT image.  List the workstation applications used for specialized CT scanning.  Describe the methods of recording and archiving CT data. 5. Image Processing and Reconstruction Image processing and reconstruction are the meat and potatoes of computed tomography imaging. A great deal of data is acquired during the scanning process, but this information is only as valuable as the quality of the image presentation. The processing algorithms and how the data are reformatted determine how easily the images can be interpreted. This module explores many important aspects of CT image processing. The module briefly reviews the components and functions of the image reconstruction process and then discusses specific postprocessing techniques. We’ll also look at the basic functions of the CT technologist’s workstation and the tools technologists use to alter image appearance, and display and archive image data. 6. Minicomputer and Microprocessors CT scanners use minicomputers to rapidly process the collected scan data. The minicomputer contains a microprocessor that transforms the incoming signal into a binary code. The microprocessor determines how quickly data is processed. Data processing speed is important because there are more data bits than can be handled at one time. The faster the entire data set can be processed, the more quickly the viewer can see the resulting images. Parallel processing occurs when data are handled by more than one minicomputer. When minicomputers use parallel processing, a task is split up among more than one processing unit, and each processor works on part of the scan data. The processed information from each unit is merged and then, in most cases, is sent to the analog-to-digital converter, or ADC. The ADC converts incoming analog data to a digital format. Before the data can be displayed on a monitor, however, the digital signal must be changed back to an analog signal by the digital-to-analog converter, or DAC. 7. Array Processor Computed tomography uses a type of microprocessor known as an array processor. The array processor is responsible for receiving image data and performing very high-speed calculations. It divides up the various imaging tasks, such as convolution and planar reconstruction, and performs these multiple tasks simultaneously. The array processor typically works using parallel processing so that multiple tasks are performed at the same time. When parallel processing involves many tasks at once, the processor is usually referred to as a large array parallel processor, or LAPP. 8. Backprojection ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction At first, image data acquired during the CT scanning procedure was processed using backprojection reconstruction algorithms. During backprojection the reformatted raw data goes through convolution, or the addition of filters. The data are essentially copied to create a data set that looks like the original raw data. However, backprojection doesn’t produce an exact duplicate of the original reformatted raw data, but rather generates new data by making an educated guess as to what the data should look like. This process involves taking the incoming raw data and combining it with the data of a prototype scan, so that the new data set appears similar to the anticipated outcome of the scan. This process was initially difficult because of artifacts caused by the backprojection process. Backprojection is not used in today’s CT scanners because of numerous limitations, such as the unsharpness of the image, which may lead to an artifact, and the length of processing time. 9. Linear Interpolation Linear interpolation is a mathematical equation applied to data acquired spirally by a single-detector row scanner. Very simply, linear interpolation reconstructs the spirally acquired images into what looks like individual slices. Interpolation is necessary because although the scan is performed in a spiral fashion, the viewer must look at the images in a flat plane. If the image was viewed in the same way that it was acquired, it would be unreadable and appear skewed. The scan would be inconclusive because there would be no way of knowing where one slice began and another slice ended. Two types of interpolation have been used for image reconstruction: 360- degree linear interpolation and 180-degree linear interpolation. A 360-degree linear interpolation averages measurements taken from each circular path around the patient to generate the projection data. The measurements are taken at reference points 360 degrees apart from one another, that is, from slice one to slice two, slice two to slice three and so on. The technique then takes these measurements and approximates the information not seen between two consecutive image slices. The disadvantage of 360-degree linear interpolation is that the distance between each gantry revolution might be large, leading the 360-degree reference points to be too far apart, which in turn may cause misregistration or incorrect image reconstruction. The 180-degree linear interpolation method works in the same way as 360-degree linear interpolation, except that the reference points are 180 degrees apart. Because the reference points now overlap, there is less distance between each of the adjacent slices. The overlap reduces misregistration and artifacts, and increases spatial resolution. Most modern CT scanners use 180-degree linear interpolation. 10. Filtered Backprojection Filtered backprojection also uses the process of convolution, or filter assignment, much like the original backprojection process; however, filtered backprojection can automatically remove artifacts that occur in anatomy of different densities, such as soft tissues and bone. Filtered backprojection assigns the data a specific kernel, sometimes referred to as a convolution algorithm, to the scanned image data. The filter enhances certain types of anatomy based on the desired appearance of the scan. After the filters are added to the incoming scan data, the newly created information is put through the traditional backprojection method to produce an artifact-free image. It’s important to know that assigning filters to the image data is what creates images without processing artifacts and is why backprojection alone cannot be used. 11. Longitudinal Interpolation with Z-axis Filtering Longitudinal interpolation with z-axis filtering typically is used by scanners with multiple detector element rows. The z-axis is perpendicular to the helical path the x-ray tube and detector array follows. This ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction reconstruction algorithm was developed because the x-ray beam in multidetector row scanners is much wider than the beam used in a single-detector row scanner. The beam is so wide that the primary path of radiation often doesn’t fall directly within each scan slice. Longitudinal interpolation with z-axis filtering calculates not only the distance from one slice to the next slice, or pitch, but also how the data are displayed for each image slice within the detector rows. The pitch affects interpolation calculations because if the image slices are farther apart, the reference points for the interpolation calculation also are farther apart. The calculation that determines how the slices fall on the detector rows is referred to as the z-gap. The z- gap algorithm measures the distance between two adjacent slices and is directly related to image quality. As the pitch decreases, the space between slices narrows, and the slices overlap. The z-gap is used because the x-ray beam might not be perpendicular to each detector during image acquisition. If this is the case, the peripheral, rather than the central, portion of the x-ray beam collects the scan data because of cone-beam geometry. The interpolation algorithm uses the information provided in the z-axis to fill in unavailable data. 12. Interlaced Sampling Interlaced, or interleaved, sampling refers to the ability of the CT scanner to calculate image data from multiple, overlapping acquisition data sets. CT image data are reconstructed into sections that assume the data are new or discrete, and are located immediately next to the adjacent slice. This assumption can cause problems because one slice often overlaps the previous slice, and the calculation is identical to that of the previous calculation, leading to repeated information. Modern CT scanners have algorithms that separate the acquired scan data from the reconstructed scan data. 13. Knowledge Check 14. Knowledge Check 15. Reconstruction Reconstruction can be performed at different times — during the scanning process, after the scan is completed or much later in another setting. During scanning, the initial data are preprocessed to eliminate various artifacts, such as those caused by faulty detector elements or beam hardening, and to ensure that there are enough data samples to fulfill the scan requirements. The preprocessed data is referred to as reformatted raw data. Filtered backprojection is used for image reconstruction immediately following the scan. At this point, a kernel is applied to the incoming scan data. The kernel provides a reference range of desired image outcomes and represents what the scan should look like, much like a set of guidelines the incoming scan data should mimic. The filter adjusts the data by accentuating the anatomy of interest. During the convolution process, the scan data are manipulated to show the desired, enhanced anatomy. Convolution removes artifacts present during the original backprojection method. After the appropriate windows are applied to the information, the data continue through the interpolation process. Reconstruction that is performed at a much later time is known as retrospective reconstruction. This type of reconstruction can create a variety of images, from 3-dimensional images to multiplanar reconstructions. 16. Reformatting The term “reformatting” is sometimes applied to the postprocessing technique of multiplanar reconstruction, or MPR. Multiplanar reconstruction rearranges the axial data set so that it can be seen from front to back, left to right and in any plane desired. We’ll discuss MPR in another part of this module. ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction Reformatting is used when the axial data set needs to be manipulated for methods such as for maximum intensity projection, or MIP and minimum intensity projection, or minIP. We’ll also discuss these reformatting methods in much greater detail at a later time. A disadvantage of reformatting images is the loss of image detail. Loss of image detail can occur for several reasons such as patient motion during image acquisition; however, loss of detail in reformatted images is a postprocessing issue that occurs after scanning. One reason for a loss of detail is that multiplanar reconstructions use only 10% of the data set for reformatting. If that part of the data set is not of sufficient quality, the user might have to switch to a reconstruction method that uses 100% of the data. 17. Specific Postprocessing Techniques CT technologists use many postprocessing techniques to manipulate imaging data. The techniques we’ll discuss in this section affect how the final image is viewed. Although these techniques are designed to enhance the appearance of the displayed image, it’s important to understand that they also remove image data during the process. The raw, unformatted data set collected during the acquisition process contains the most image information. 18. Maximum Intensity Pixels Maximum intensity pixels, more commonly known as a maximum intensity projections, or MIPs, are created using less than 10% of the total data set. To create an MIP, the computer software allows only the brightest, or whitest, shades to remain and removes all other densities not needed. The brightest shades are now the focus of attention for the viewer. This type of processing is especially useful in computed tomography angiography, or CTA. A CTA scan enhances the arteries with a positive iodinated contrast agent. The IV contrast appears as the brightest pixel values on the scan after the bones. In addition to artery reconstructions, this method sometimes is used for exams to evaluate the kidneys, ureters and bladder. 19. Minimum Intensity Pixels Minimum intensity pixels, more commonly known as a minimum intensity projections, or minIPs, also are created using less than 10% of the total data set. The computer software enhances the darkest shades within the scan, while removing the brighter shades. The darker shades are now the focus of attention for the viewer. This type of processing is especially useful in CT virtual colonography exams. Virtual colonography uses a negative contrast medium, usually carbon dioxide, to enhance the colon. The minIP software “highlights” the darker pixels, leaving the carbon dioxide-filled colon and subtracting the rest of the tissues. A minIP reconstruction also may be used for dedicated lung studies. Lung tissues fall in the darker part of the Hounsfield number spectrum because the lungs are filled with air. 20. Image Smoothing The image smoothing filter is applied to scans that must demonstrate subtle soft tissue differences. The filter evens out areas that look blurred or have subtle changes within the shades of gray. The smoothing filter also helps improve low-contrast resolution. Low-contrast resolution is the ability for the CT scanner to differentiate between very minor changes in shades of gray. Image smoothing takes place during the assignment of a kernel, or convolution algorithm. The use of a smoothing filter is by far the preferred technique for a soft tissue study. 21. Edge Enhancement An edge enhancement kernel, also known as an edge enhancement convolution algorithm, adds detail to the denser anatomy within a scan. This filter usually is applied to studies in which it’s important to show bony structures. Kernels and algorithms are related in that the kernel can only be applied to the scanned data through a series of algorithms, or mathematical processes. ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction Used during the convolution process, the kernel is preset according to a histogram. The kernel is what gives the scanned data a bony, a soft tissue or a cleaned-up appearance. A histogram is a curved line that indicates what the scan parameters should resemble. The histogram on the left has a narrow slope, with curvatures only at the top and bottom of the Hounsfield scale. In other words, the computer software should only use the highest and lowest Hounsfield numbers for image reconstruction. The histogram in this example indicates how an exam yielding blacks and whites should appear. The histogram on the right has a much more rounded curve overall. This means the entire Hounsfield scale should be used for image reconstruction rather than just the upper and lower Hounsfield numbers. 22. Gray-scale Manipulation Gray-scale manipulation is the process of adjusting the range of grays within an image. Gray-scale manipulation also occurs when filters are applied to the scan data.The Hounsfield scale represents the given range of numbers the computer can assign to each pixel element. The numbers range from +1000, which represents white or bone, to -1000, which represents black or air. Each pixel within the CT scan is assigned a CT/Hounsfield number to give images a desired appearance. Images with a wide window width contain many shades between black and white, while those with a narrow window width have a smaller range of shades. Gray-scale manipulation is necessary because the appearance of the image can be changed drastically based on which pixels are enhanced and which pixels are minimized. 23. Gray-scale Manipulation Window Width The window width (WW) is the total number of CT/Hounsfield numbers available for any given exam. Window width typically is described as either narrow or wide. A narrow WW is associated with scans requiring fewer shades of gray. Each pixel within the image is assigned a different shade of gray than the pixel next to it. This is especially important when imaging anatomy like the brain because the densities within the brain are so similar. Gray and white matters differ from one another by only a few CT/Hounsfield numbers. A wide WW is used when an exam needs many shades of gray to adequately show the changes in the different tissues. 24. Gray-scale Manipulation Window Level Another important aspect of gray-scale manipulation is the window level (WL). The window level is the midpoint number in the CT/Hounsfield range for any given exam. For example, the window level for a CT/Hounsfield range of 0 to 100 is 50. The appearance of an image is affected in part by the window width but also by the WL, or midpoint of the range. 25. Gray-scale Manipulation – WW/WL Equation The gray-scale range is determined using the window width/window level equation. Let’s look at an example. Assume that the scan parameters are set to a window width of +800 and a window level, or midpoint, of +200. The first step in this equation is to take the WW number and divide it in half. In this example, we divide the WW of +800 by 2, which equals +400. Next, the new WW number is both added and subtracted from the window level value. In this case, we add +400 to +200, which equals +600, and we subtract +400 from +200, which equals -200. So, for this example, the entire range for a window width of +800 and a window level of +200 is -200 to +600. 26. Gray-scale Manipulation – WW/WL Equation Let’s look at one more example of the WW/WL equation. What is the range of CT/Hounsfield numbers if a window width of +800 is set, but a different window level, or midpoint number, of +400 is used? As you remember, the first step of the equation is to take the given WW number and divide it in half. So, for this equation, +800 is divided by 2. ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction Next, the new WW number is both added and subtracted from the WL value. So for this example, the entire range for a window width of +800 and a window level of +400 is 0 to +800. 27. Gray-scale Manipulation Grayer shades usually fall in the -200 to +200 area of the CT/Hounsfield number range. Blacker shades typically start around the -100 area of the scale. The window width in the first example included CT/Hounsfield numbers from white shades to black shades, so the resulting image could correctly display more variations in tissue densities. The window width ranges in each of the previous examples show their relationship to the CT/Hounsfield range. If a window width range falls higher up on the CT /Hounsfield scale, the resulting image will appear brighter, and vice versa. Radiologists use the window width range to determine how the CT/Hounsfield numbers are assigned for the scan, and thus demonstrate what pathology is present. 28. Shaded Surface Rendering Shaded surface rendering is used to create the illusion of three dimensions on a two-dimensional computer screen. Shaded surface rendering uses 10% of the total data set. For this type of reconstruction, the data set should be acquired using thin slices, or a pitch of less than 1. Shaded surface rendering gives shape and depth to the contour or periphery of an image for a more life- like appearance. In effect, the technique manipulates the position of a simulated light source so that some regions of the image are illuminated while other regions are in shadow. The voxels farthest from the area of interest are darkened to create the illusion of roundness, shape and texture. By moving the light source in 90 degree increments around the object of interest, different parts of anatomy appear more detailed. Shaded surface display, or SSD, is the end result of shaded surface rendering. Often the terms are used interchangeably, but shaded surface rendering involves the actual creation of the shaded surface, and shaded surface display consists of the presentation of the image on the 2-D computer screen. 29. Multiplanar Reconstruction Multiplanar reconstruction is performed after the patient has left the department. The scan data is manipulated in different ways to enhance certain characteristics. These reconstructions do not add any information to the data set, but rather take the information already acquired and present the data in alternative ways. Multiplanar reconstructions use 10% of the image data, changing the image from an axial projection to a coronal, sagittal or oblique projection. To perform the reconstruction, the computer software rearranges the pixels and changes the pattern of image slices. 30. Curved Multiplanar Reconstruction Like multiplanar reconstruction, curved multiplanar reconstruction, or curved MPR, is a postprocessing technique that allows the viewer to look at images in coronal, sagittal or oblique projections. The technique works in the same way as traditional multiplanar reconstruction except that the edges of the reconstructed images can be curved or irregular rather than just straight. The only difference between traditional and curved MPR is that for traditional MPR, the computer selects the direction of the reconstruction based on the desired viewing position. In a curved MPR, the viewer sets the boundaries for reconstruction. Only 10% of the axial data set is used for a curved MPR. In the example on this page, the broad black lines show where the viewer wants to begin the image reconstruction. The computer then determines the voxels that will be used for the reconstruction process. 31. Volume Rendering ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction Volume rendering is a complex technique used to create images that appear three-dimensional. Reconstructions can be adjusted to show the surface of an object or deep into the most internal structures. This rendering method uses 100% of the data set, so more information is available for reconstruction and the cost of the computer software is higher. Volume rendering sometimes is referred to as a four-dimensional, or 4-D, reconstruction in that the added pixels give the illusion of a transparent layer on top of the image. In volume rendering, the voxel elements go through the process of segmentation, much like in shaded surface rendering. Segmentation involves creating a boundary of Hounsfield numbers in which the anatomy of interest is highlighted. In addition to a number boundary applied to the voxels, there’s an additional pixel element added to the structures demonstrated on the scan. These additional pixels give the illusion of partial visibility, similar to a soapy film left on a dish. The film is visible, but everything under the film also is visible. The image has the appearance of a shaded surface rendered image, except that the light source used for the periphery of the image is used for the rest of the image. In other words, the viewer is able to remove the outermost surface of the image, revealing the internal anatomy. The volume rendering technique continues to use the light source calculations to add depth to the image, as though the viewer is looking at the image head-on. 32. 3-D Reconstruction Like other postprocessing reconstruction techniques, 3-D is used to give the acquired images a more life- like appearance. The 3-D technique requires a data set acquired in thin, overlapping slices, although the computer software only uses 10% of the data set. The software takes each pixel and creates what looks like coronal, sagittal and oblique projections — in other words, multiplanar reconstructions. The axial data set is simply restacked so that the projection of interest determines the direction the pixel elements are viewed. The 3-D software either enhances the brightest pixels while minimizing or removing the darker pixels, as in maximum intensity projections, or enhances the darker pixels and removes the brighter shades, as in minimum intensity projections. With 3-D viewing, the radiologist and other medical professionals can better understand the relationship of anatomical structures and can locate pathology or trauma more precisely. 33. Virtual Reality – VR Virtual reality, or VR, imaging is a postprocessing application used to demonstrate what the inside of structures look like. The abbreviation VR can sometimes be mistaken for the term “volume rendering,” but the two techniques are not identical. Essentially, VR software allows the viewer to travel through the inside, or lumen, of a tubular structure, as if the patient had undergone an actual procedure such as an endoscopy or colonoscopy. Virtual reality imaging uses 100% of the axial data set to create images. The data set should be acquired using thin slices and a pitch of less than 1. When there’s greater slice overlap, more data is available for reconstruction, although this comes at the cost of increased radiation exposure. The viewing route can be created automatically by a set of established protocols or drawn on top of the image by the user. For an automatic reconstruction, the computer identifies a reference point within the lumen to start the process. For example, if a colon is visualized, the computer singles out the Hounsfield number assigned to the lumen’s opening. The software calculates a set of reference numbers and then uses all numbers that resemble the reference set to create the reconstruction. Any numbers that vary greatly from the reference numbers are excluded from the reconstruction, or may be used as the wall of the structure. The viewer also can hand draw the reconstruction pathway while viewing the axial data set in a multiplanar reconstruction format. After the images are displayed in the coronal, sagittal and axial planes, ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction the viewer traces the target line onto the images with the mouse cursor. The MPR images allow the viewer to trace just the tube-like structure as it appears. After the viewer has outlined the structure, the computer includes only the Hounsfield numbers that are close in value to the drawn line. Hounsfield numbers that vary greatly from the viewer’s hand-drawn line are excluded from the reconstruction. These numbers can be included as the wall of the lumen, or if many are present, may be shown as an abnormal mass or pathology within the lumen. 34. Knowledge Check 35. Knowledge Check 36. Image Display, Manipulation, Recording and Archiving As you’ve just seen, image reconstruction and postprocessing are two very important parts of CT imaging. The processes we’ve looked at so far affect the quality and integrity of the image data. However, the CT technologist has other tools that play an equally important role in creating a quality image for interpretation. The final section of this module explores the various tools, viewing modes, functions and applications of the computed tomography workstation. If you’re familiar with the workstation functions of cassette-less and cassette-based imaging systems in diagnostic radiology, you might recognize some of these functions. Many of these tools are similar to the those used by technologists who perform digital imaging. 37. Pan Image pan is a display tool that allows the user to move the image on the display field in any direction desired. The viewer uses the mouse to electronically select the image and move it in the desired direction. Panning is an especially useful tool when the image is enlarged using the zoom tool. When the image is enlarged, only a small section can be seen on the monitor. The pan tool allows the user to move the image around until the desired anatomy is displayed on the monitor. 38. Zoom Viewers use the zoom tool to increase the size of an image. Zoom typically enlarges the image but doesn’t add information to the scan. The pixels in the image appear larger because the image is brought closer to the viewer’s perspective. The image matrix doesn’t change either, but image quality may be degraded if the viewer zooms too much. As the pixels within the image are viewed more closely, small details are lost. Although zooming the image may decrease the resolution, smaller anatomical structures are easier to measure, and sometimes it’s easier to relate anatomy to surrounding tissues. 39. Image Scrolling The image scrolling tool typically is controlled via the computer mouse. The viewer rolls a raised wheel in the middle of the mouse to move the images in sequential order. When the viewer places the mouse cursor over the “dog-ear” at one of the corners of the data set, the images can be viewed from beginning to end. Image scrolling gives viewers a fast way to flip forward and backward through the image slices so that the data set can be interpreted as a whole. Image scrolling also allows the CT technologist to see where to set the parameters for postprocessing reconstructions. The image scrolling tool typically is controlled via the computer mouse. The viewer rolls a raised wheel in the middle of the mouse to move the images in sequential order. When the viewer places the mouse cursor over the “dog-ear” at one of the corners of the data set, the images can be viewed from beginning to end. Image scrolling gives viewers a fast way to flip forward and backward through the image slices so that the data set can be interpreted as a whole. Image scrolling also allows the CT technologist to see where to set the parameters for postprocessing reconstructions. 40. Swivel ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction The swivel tool rotates an image to the left or right, as seen by the viewer. This action is similar to shaking your head when saying “no.” The swivel tool is especially useful when viewing images that have been loaded into a 3-dimensional software program. 41. Roll Rolling is similar to moving an image head over heels. This tool is used to show different parts of anatomy and is especially useful in viewing reconstructed 3-D images. 42. Rotate The rotate tool allows the image to be moved in any direction. The viewer can work with the image as if holding the anatomy in his or her hand. The ability to rotate anatomy allows the user to view an area of interest in greater detail and at all angles. The rotate tool is used most often with a volume-rendered data set. Volume rendering creates a cube-like display of individual pixels. The cubed pixels, now called voxels, are used to stack the image slices, so that the image no longer looks flat, but has a more textured appearance. Rotation is used for a variety of exams, but is seen most frequently in head-over-heels and left-to-right spin reconstructions for CT angiography. 43. Measurement The measurement tool calculates the size of abnormal pathologies seen within the anatomy. The tool usually measures distance and diameter. The radiologist uses the distance measurement tool to determine the length of an object or how close an object is to surrounding tissues. The diameter measurement tool is used to calculate an object’s circumference. Measurements typically are made during postprocessing, but measurements also can be compared from one scan to the next. Examinations that use measurement tools include chest scans to evaluate pulmonary nodules or scans to rule out an abdominal aortic aneurysm. 44. Magnify The magnification tool is used to look more closely at an image. When the tool is activated with the mouse, a small square appears over the image. The part of the image inside the square appears closer than the surrounding anatomy. This tool is similar to zoom, but instead of the entire image being brought closer to the viewer and appearing larger, only the portion under the magnification square appears bigger. 45. Viewing Modes The CT technologist’s workstation has several types of viewing modes. Depending on their needs, radiologists or technologists can select two-dimensional, three-dimensional, slab, planar or cine viewing modes. 46. 2-D and 3-D Two-dimensional, or 2-D, refers to an object displayed as a horizontal and vertical representation. Two- dimensional images have no depth or roundness, but appear flat. The ability to view objects in 2-D is important because axial scanning acquires and displays image data in 2-D. CT image data acquired by helical scanning can be reconstructed in other ways. The ability to collect volumetric data led to the development of 3-D reconstructions. To create 3-D images, the computer software changes each pixel element into a cube. The cubed pixel is referred to as a voxel. 47. Slab A slab is similar to a multiplanar reconstruction in that the voxels of each image slice are stacked in a desired order based on the anatomy of interest. A slab takes a chunk of the scan and creates coronal, sagittal or oblique perspectives. ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction The viewer can manipulate the direction of the slab by changing the position of the voxel elements. Individual voxel elements are selected as a reference or starting line. Simply reassigning or changing the reference line moves or changes the direction of the slab. 48. Planar Planar refers to the point of view of the data set. Images appear in different planes depending on the anatomy of interest. To produce an image in the axial plane, spirally acquired data first must undergo 180-degree linear interpolation and multiplanar reconstruction. During interpolation, the images are converted from a spiral, skewed appearance, to flat individual slices of data. Multiplanar reconstructions change the direction of the data set to present a coronal, sagittal or oblique projection. If a coronal reconstruction is needed, the voxels from each slice are stacked on top of each other, starting with the first horizontal row of slice one, the first horizontal row of slice two and so on. The first row of all the slices are then stacked on each other, producing an image seen from front to back. This same process occurs for images seen in a sagittal projection, in that the entire vertical row of voxels in slice one is taken and stacked on the vertical row of voxels in slice two and so on. 49. Cine Cine, or loop viewing, refers to the ability of the viewer to watch the series of images move sequentially from the beginning to the end of the data set. This type of viewing mode typically is created during postprocessing, most often using 3-D reconstruction. A cine film usually requires the entire data set, which entails the use of an advanced computing system at an increased cost. 50. Technologist Workstation The technologist workstation is the control center of the entire CT scanning process. Besides controlling data acquisition, image display and specific reconstruction software, the workstation performs the administrative duties associated with handling patient and image data. 51. Directory or Patient List The directory, or patient list, is a register of patients who are scheduled for an exam in the radiology department. The list is generated from the radiology information system, or RIS. When patients arrive at a hospital, their information typically is entered into the hospital information system, or HIS, database. Radiology examination requests are entered into an ordering system and automatically routed to the RIS. The RIS and HIS are integrated so that care is not delayed, and the patient data is identical in both databases. The RIS assigns each exam a unique number, similar to a bar code, which typically is referred to as an accession number. Accession numbers can be filtered according to each modality in the radiology department. 52. Archive Button Archiving is the process of saving examinations to a database so that images can be recalled when needed. Archiving images is essential, and an archiving system should be set up in such a way that the images are easy to retrieve and have the same quality as the initial images. Archived information should be stored in a separate location such as a backup server so that images can be used for comparison at a later time. 53. Query The query process is based on a question-and-answer format. The user requests specific information from available databases and receives the answer via the computer. Another example of the query process is when the computer requests patient information updates from the radiology information system database. This query can be set up automatically, or the user can make updating the work list a selectable option. 54. Send/Network The CT technologist selects the send option after completing the exam. The technologist chooses the appropriate image data sets and then exports them to the desired location. 55. Copy/CD ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction A copy is an exact duplicate of the original image data. Copies allow more than one person to have the images. It’s important to remember that a copy can become degraded if it’s created from a duplicate image rather than the original data set. As copies are made from other duplicates, image data can become degraded, and the image detail will be poor. A copy of an examination typically is sent to a storage area in a digital format, although the digital archiving of a scan is not a true copy. Copies of the scan can then be created at a later time using a digital storage medium such as a compact disc, DVD or optical disc. We’ll cover optical discs and DVDs later in this module. Compact discs are commonly used to archive or copy medical images. A CD is a small, relatively inexpensive optical disc that has been digitally imprinted with the scan information. To view the images, an optical scanning device reads and displays the digital information. CDs can store and retain several exams for extended periods of time. 56. Online Help Equipment manufacturers often provide an online help service. This assistance can be interactive; that is, the CT technologist can submit questions to a knowledgeable vendor’s representative at an offsite location. Or, help may be available through a manufacturer’s Web site. A dedicated Web site often contains frequently asked questions, also known as FAQs, or short online tutorials that provide a step-by- step guide for users. 57. Workstation Applications Workstations can be set up to perform specific functions depending on their location in the hospital. Special software for various CT applications that perform highly specialized functions may be installed on a specific scanner. This page lists some of the major applications available on CT scanners. We’ll briefly discuss a few of these different uses; however the final module in this series examines the applications in greater detail. 58. Ejection Fraction Ejection fraction is an advanced coronary application used to determine how much blood is pumped from chamber to chamber. Ejection fraction also may be used to evaluate blood flow volume from each side of the heart. The heart scan is performed using gating software that tracks the patient’s heartbeat during the exam. Scanning takes place during the resting portion of the patient’s heartbeat. The gating software tracks the heartbeat on what looks like an electrocardiogram strip, and then the scanner acquires the images when the patient’s heart is at rest. After the examination, the scanner creates a postprocessing reconstruction of the heart cycle. These reconstructions, when loaded into the appropriate software, give the illusion of a beating heart. The images help determine which phase of the heartbeat is not pumping sufficient blood. A color map is applied to the heart reconstruction to indicate deficient function at certain times during the heart cycle. 59. Calcium Scoring Calcium scoring is an advanced cardiac CT application used to calculate the amount of calcium in the coronary arteries. It is performed retrospectively, after the patient has left the department. The software assigns a preset Hounsfield number to calcium and then evaluates the areas that appear like calcium buildup. Calcium possesses a range of numbers at the upper end of the Hounsfield unit scale. Because calcium is one of the building blocks for bone, the range is near +1000, or the Hounsfield number for bone. Next, the software calculates the amount of calcium in the artery based on the given threshold. This number is subtracted from the remainder of the coronary arteries. Calcium scoring is important because calcium buildup leads to a lack of blood flow within the arteries. Depending on the total amount of calcium blocking the arteries, the radiologist can determine whether the patient has, or is at risk for, developing arteriosclerosis. 60. CT Fluoroscopy Real-time CT imaging is a tool used for CT-guided biopsies. Real-time imaging often is referred to as CT fluoroscopy. The images are similar to conventional spiral CT images; however, there are differences in how many images are acquired at one time and where the patient is during the procedure. The x-ray tube and detector array rotate around the patient in a 360-degree path, and an image is produced for each rotation. Instead of acquiring as many slices as there are rows of detectors, the detector array produces a single image, and the patient stays in exactly the same position for the entire procedure. In conventional ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction CT, the table moves at the speed needed to gather the images in relation to the rotation of the x-ray tube and detector array. During CT fluoroscopy, the images are acquired in one area only, and a single image is taken from each 60-degree portion of the 360-degree rotation. So, a total of six single images in the same position are acquired for each complete rotation. The series of images looks like a live action sequence that allows the radiologist to see the exact location of the biopsy. In addition, the patient remains stationary within the gantry because the position for the very first image is used to gauge exactly where the biopsy will take place. The process of image reconstruction varies significantly in CT fluoroscopy. An interpolation algorithm isn’t required because image acquisition is similar to a single-slice scanner. The x-ray tube and detectors don’t follow a helical path, and the patient doesn’t move through the CT gantry. The images are reconstructed using the initial 360-degree image as a point of reference. This first image contains the full amount of scan data from an entire rotation. The next image is reconstructed from the first image based on the complete data set collected in one full rotation. The second image acquired uses only 60 degrees of the 360-degree rotation. The rest of the information pertinent to the image is averaged based on what the initial image reveals. After the first 60-degree image, the second 60-degree image is acquired at the next rotation increment. Reconstruction occurs by subtracting the first 60-degree image data from the new data of the second image. The CT technologist then scrolls through the image data set, allowing the radiologist to verify where the biopsy will take place. When the exact location is determined, the technologist sets the biopsy position to zero so that if the scanner needs to be repositioned, the zero setting will move the patient back to the selected spot. The radiologist places a stereotactic tool on the patient and determines what anatomy lies beneath the device. The stereotactic tool transmits an exact coordinate of its position to the computer. The computer then uses the initial, full scan of the area of interest to create an image of what that anatomy will look like with a biopsy needle. 61. Radiation Oncology Treatment Planning CT imaging is used to help plan where radiation is targeted during treatment. The patient is positioned for the CT scan in exactly the same position as the set up for the radiation therapy table. Occasionally the patient’s skin is marked to indicate the treatment set up, and sometimes the radiation therapy immobilization masks or fiducial markers are used for the CT scan to ensure reproducible positioning when the patient undergoes treatment. After the CT scan is performed, the image data are loaded onto the treatment planning software. The CT scan is used to define and measure the radiation treatment volume, with the treatment plan superimposed on the CT scan. The CT scan also helps to define beam parameters and radiation dose distribution. 62. Fusion Fusion is a technique that retrospectively combines images from two modalities. Positron emission tomography and computed tomography, or PET-CT, are commonly merged as a fusion study. The first step is to acquire the images from both modalities. For PET-CT, the radiopharmaceutical is administered first and then the CT scan is performed, with or without IV contrast, followed by PET imaging. Although images can be acquired independently by a CT scanner and nuclear medicine camera and then fused, hybrid machines are now manufactured that can perform both imaging methods. During the examination, it’s important for the patient to maintain the same position for both the CT scan and PET imaging. If the patient moves out of position, it will be difficult to correctly align the anatomy in the CT and nuclear medicine images. This is the reason patients usually don’t hold their breath during the CT scan. The nuclear medicine scan takes approximately 15 minutes to complete, and it would be impossible for patients to hold their breath for that length of time. After image acquisition, the technologist uses a computer software program to perform retrospective reconstructions, loading the data from the CT scan first and then placing the PET images over the CT scan. 63. Neurology Neurology applications are used for many different reasons, including fracture assessment, cerebral artery evaluation and assessment of brain damage and recovery potential. Volume rendering and advanced subtraction methods are two of the techniques used in neurology applications to demonstrate anatomy. Volume rendering creates images that appear three-dimensional and can show anatomy from the periphery of an object down to the most internal structures. For example, volume rendering provides a clear representation of the skull to evaluate fractures. Subtraction methods typically are used for ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction evaluating the arteries of the brain. Subtraction works by determining the change in two separate data sets acquired at different times. The change usually is demonstrated using IV contrast. For example, the CT technologist performs a head scan without a contrast agent and then performs the same scan using contrast. The computer software removes all the identical Hounsfield numbers in the two studies, leaving the remaining, different numbers to be displayed. Xenon studies are performed to determine if the patient is receiving sufficient blood flow to the brain. After a baseline CT scan of the head is performed, the patient inhales xenon gas. The computer software calculates the Hounsfield numbers before the xenon gas was inhaled, the Hounsfield numbers during inhalation and finally the Hounsfield numbers at the washout phase. The washout phase occurs when the xenon gas leaves the brain and the original Hounsfield numbers begin to reappear. 64. Vessel Analysis A vessel often may be obscured by the anatomy around it, especially in the cranial bones. Vessel analysis helps demonstrate the exact location of a vessel, while excluding all nonessential anatomy. The technique works by determining the threshold number for the vessel. Vessels have a certain range of Hounsfield numbers, and based on those numbers, the computer software can remove, or suppress, the numbers for all other tissues, leaving the vessels of interest visible. Vessel analysis simply gives the computer software a guideline of which Hounsfield numbers it should keep and which numbers it should discard. 65. Vessel Tracking Vessel tracking takes place after the vessel analysis process. Vessel tracking is similar to virtual colonoscopy in that a specific vessel is used as the starting point for reconstruction. The computer uses designated Hounsfield numbers to recreate the vessel by adding only the pixels that are close to the initial reference number. The CT technologist selects a region of interest, and the computer then inserts the similar numbers. These additions slowly begin to form what looks like a vessel. Vessel tracking is an important application because other anatomical structures can hide a vessel. This application allows the CT technologist to reconstruct a vessel and know the entire vessel is present. 66. Recording and Archiving After a scan is complete, the patient has left the imaging department and the data have been processed and reconstructed, the images need to be displayed and archived appropriately. Historically, CT scans have either been printed for viewing on a light box, or saved on optical discs or tape for review at a workstation. Recording and archiving CT images is a critical component of patient care because previous studies frequently are used for comparison purposes. 67. Archiving Computed tomography images can be saved to CDs, DVDs or optical discs. Although a picture archiving and communication system, or PACS, is the primary method for long-term storage, images may need to be recorded on a transportable device for a referring physician or the patient. DVDs are an ideal archival medium because of their small size and ability to store thousands of images. DVDs last for years, provided they are not exposed to intense heat or scratched, and most viewing devices can read DVDs without a problem. Optical discs were widely used in the past because of their ability to record many more studies than older magnetic tapes. Although their use has declined with the introduction of CDs and DVDs, optical disks still provide excellent image retention and, if properly stored, don’t lose any of the image quality. 68. PACS Most people think of a PACS when image archiving is discussed. The central database of the PACS receives and stores image data from different modalities in the department. All other computers connected to the PACS network, either within the hospital or at off-site locations, can then retrieve or input information as needed. In addition to ensuring the integrity of the central database, the PACS must be compatible with the Digital Imaging and Communications in Medicine, or DICOM, standards. DICOM compatibility ensures that all modalities within a radiology department can communicate with the PACS and with one another, and users can view various imaging modalities in a consistent format. The PACS ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction usually is set up for long-term image archiving. In addition, an off-site storage area retains archived images in the case there is a system failure; this storage area acts like a backup for the backup. 69. Knowledge Check 70. Knowledge Check 71. Conclusion This concludes CT Basics Module 4: Image Processing and Reconstruction. You should now be able to: 1Describe the steps required for computed tomography image reconstruction.2 List the postprocessing techniques needed for image enhancement. 3 Define the tools used to view a CT image. 4 List the workstation applications used for specialized CT scanning. 5 Describe the methods of recording and archiving CT data. 72. References Dervin JE, Miles J. A simple system for image-directed stereotaxis. FR J Neurosurgery. 1989;3(5):569- 574. Papp J. Quality Management in the Imaging Sciences. 3rd ed. St. Louis, MO: Mosby; 2006. Seeram E. Computed Tomography: Physical Principals, Clinical Applications, and Quality Control. 3rd ed. St. Louis, MO: WB Saunders; 2009. ©2017 ASRT. All rights reserved. CT Basics Module 4: Image Processing and Reconstruction