Image Reconstruction and Manipulation PDF

IMAGE RECONSTRUCTION AND MANIPULATION -​ Basic principle - Algorithms, fourier transform, convolution, interpolation -​ Image reconstruction from projections - Historical perspective, problem in CT -​ Reconstruction algorithms - Back projection, iterative algorithms, analytic reconstructi...

IMAGE RECONSTRUCTION AND MANIPULATION -​ Basic principle - Algorithms, fourier transform, convolution, interpolation -​ Image reconstruction from projections - Historical perspective, problem in CT -​ Reconstruction algorithms - Back projection, iterative algorithms, analytic reconstruction algorithms (fourier reconstruction algorithms, filtered back-projection) -​ CT Data (Types of Data) - Measurement data, raw data, convolved data, reconstructed data -​ Windowing - WW (contrast) WL (brightness) -​ Visualisation tools - Basic Tools and Advance Tools -​ Image manipulation - Reformatting Techniques Image Quality -​ Spatial resolution -​ Noise -​ Contrast resolution -​ Linearity -​ Uniformity -​ Artifacts - Beam hardening, photon starvation, partial volume artifacts, out field artifacts, ring artifacts, Cone beam CT (CBCT) -​ Basic operational and physical principles of CBCT -​ Instrumentation of CBCT Quality Control for CT scanner -​ Principles of quality control -​ QC tests for CT scanner ​Image Reconstruction and Manipulation 1.1 Basic principles 1.​ Algorithms are defined as a set of rules or directions for getting a specific output from a specific input. 2.​ Fourier transform is defined as a function that describes the amplitude and phases of each sinusoid, which corresponds to a specific frequency. In other words, the Fourier transform is a mathematical function that converts a signal in the spatial domain to a signal in the frequency domain. 3.​ Convolution is a digital image-processing technique used to modify images through a filter function. 4.​ Interpolation is a mathematical technique used to estimate the value of a function from known values on either side of the function. 1.2 Image Reconstruction From Projections Historical Perspective -​ Radon (1917) developed mathematical solutions to the problem of image reconstruction from a set of its projections. -​ In his initial work, Hounsfield’s images were noisy as a result of his chosen reconstruction technique. Then, convolution back projection algorithms were introduced. Problem in CT -​ The problem in CT is to calculate all values for the μ terms for a large set of projections. Projections can be obtained through both parallel and fan beam geometries. Image reconstruction from projections involves several algorithms - back projections, iterative methods and analytic methods to calculate all the μ terms from a set of projection data. 1.3 Reconstruction Algorithms Back projection (also called the “summation method” or “linear superposition method”) -​ Simple procedure that does not require much understanding of mathematics. -​ The problem of back projection technique is that it does not produce a sharp image of the object and therefore is not used in clinical CT. -​ The striking artifact appear in back projection images is the typical star pattern that occurs because points outside a high density object receive some of the back-projected intensity of that object. Iterative algorithms -​ How it works: an iterative reconstruction starts with an assumption and compares this assumption with measured values, makes corrections to bring the two into agreement, and then repeat this process over and over until the assumed and measured values are the same or within acceptable limits. -​ Techniques: -​ Simultaneous iterative reconstruction techniques  -​ Iterative least-squares techniques -​ Algebraic reconstruction techniques -​ Limitations (In early years of used) 1.​ It is difficult to obtain accurate ray sums because of quantum noise and patient motion. 2.​ The procedure takes too long to generate the reconstructed images because the iteration can be done only after all projection datasets have been obtained. 3.​ More projection datasets than pixels to produce a true image. -​ Advantages (IR algorithms have resurfaced due to high speed computing) is reduce image noise and minimize the high radiation dose. Analytic Reconstruction Algorithms -​ Is developed to overcome the limitations of back projection and iterative algorithms. -​ There are two analytic reconstruction algorithms which are Fourier reconstruction algorithms and filtered back projection. a)​ Filtered back-projection (convolution filter method) -​ How it works: it is filtered or convolved to remove the typical starlike blurring that is characteristic of the simple back-projection techniques. -​ The steps in the filtered back projection method: -​ Major problem: noise and streak artifact. b)​ Fourier reconstruction algorithms -​ Advantages: 1.​ The image in the frequency domain can be manipulated (ex: edge enhancement or smoothing) by changing the amplitudes of the frequency components. 2.​ A computer can performs digital image processing 3.​ Frequency information can be used to measure image quality through the point spread function; line spread function, and modulation transfer function. -​ The steps in the Fourier reconstruction:  1.​ The object to be scanned is represented by the function f(x,y). 2.​ Projection data are obtained from the object. A projection dataset for at least a 180-degree rotation is required for adequate reconstruction. These projections represent a spatial domain image. 3.​ Each projection is transformed into the frequency domain by the Fourier transform. This image must be converted into a clinically useful image. 4.​ Because CT scanners use a fast Fourier transform developed specifically for digital implementation,the frequency domain image must be placed on a rectangular grid. This is accomplished by interpolation. The fast Fourier transform requires that the pixels in the grid array be 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024 5.​ Finally, the interpolated image is transformed into a spatial domain image of the object through an inverse Fourier transform operation. 1.4 Types of data 1.5 3D algorithms 3D reconstruction techniques for surface display is based on at least two process: a)​ Preprocessing b)​ Display And it consists of the following operations: a)​ Interpolation b)​ Segmentation c)​ Surface formation d)​ Projection 1.6 Image Manipulation (image postprocessing) and Windowing -​ Image post processing uses various techniques that modify the reconstructed images displayed for viewing and interpretation by an observer. -​ Image post-processing techniques is gray-level mapping also known as contrast enhancement, contrast stretching, histogram modification, histogram stretching and windowing.  -​ Window width - manipulated image contrast -​ Window level - manipulated image brightness Explain the definition of window width and window level. -​ Window width is defined as the range of the CT numbers in the image. It determines the maximum number of shades of gray that can be displayed on the CT monitor. -​ Large window width is a relatively long gray scale or a large block of CT numbers that will be assigned some value of gray (result in low contrast). Wide WW used to encompass tissues of greatly different attenuation within the image. For example, bone structure and lung. -​ Narrow window width, the transition from black to white will take place over a relatively few CT numbers. Narrow WW used to display soft tissues within structures that contain different tissues of similar densities. For example, the brain and liver. -​ Window level is defined as the center or midpoint of the range of CT numbers. -​ As window level increases, the image gets darker because more of the lower CT numbers are displayed. -​ As window level decreases, the image gets brighter because more of the higher CT numbers are displayed. 1.7 Visualisation tools 1.8 Image Manipulation - CT Image Reformating Techniques (Multiplanar Reconstruction, MPR) Multiplanar Reconstruction (MPR) -​ Also known as image reformatting or image reformation. -​ MPR is a computer program that can create coronal, sagittal, and paraxial images from a stack of contiguous transverse axial scans. Advantages and disadvantages of MPR.  Advantages Disadvantages 1.​ Enables visualisation of specific 1.​ Loss of image detail. structures in relation to surrounding structures 2.​ Determine extent of lesions or fractures 3.​ Helps to localize lesions, bone fragments, or foreign bodies. What is the difference between reconstruction and reformation? Reconstruction Reformation From raw data From image data Axial orientation Any orientation Can use any algorithm Must use same algorithm as image data ​Image Quality