RAD 350 Module 4
Attenuation Coefficient
% or ratio of original x-ray beam intensity absorbed by each different type of tissue area within the body 3d projection onto 2d image receptor
Three Adjustments made to histogram before the image is displayed
1. Histogram analysis1. Identifies the useful pixels (values of interest) 2. Rescaling adjusts the pixel values to fit the average histogram of the body part and adjusts for small exposure errors 3. A look-up table (LUT) is used to give the image just the right amount of brightness and contrast
Filtering
A selected layer can also be entirely left out upon reconstructing the image Band-Pass Filtering: used in noise reduction (electronic mottle) Low-Pass Filtering: the algorithm will pass through the low-frequency layers Most commonly called smoothing function- The elimination of noise from the image generally requires that we accept some loss of fine detail along with it
Grayscale
AKA Image Contrast Property that comprises visibility of detail on the displayed digital image and is the differences between IR exposures Differences can range from clear white through various shades of gray to black
Spatial Resolution (definition, sharpness, recorded detail)
Ability for a system to demonstrate small details of an object The degree of geometric sharpness or accuracy of the structural lines actually recorded in the image Geometric property that controls the detail or sharpness of structural lines Is the dot distinguishable as separate from other dots
Contrast Resolution (Grayscale, Shades of Gray)
Ability of the digital system to display subtle changes in the shade of gray Higher contrast resolution-differences between adjacent densities are enhanced Directly related to the bit depth of the pixels in the image (more pixel bit depth = more shades of gray)
Digital Radiography - Direct Conversion
Active Matrix Array (AMA) is the main image capture component First panel has thousands of dexels (individual electronic detector elements) Direct Conversion Detectors; Surface of dexel made from amorphous selenium (high absorption efficiency for x-rays) --> AMA panel converts energy of remnant x-ray beam directly into electrical charges computer can read Each dexel has three compoents: Radio-sensitive detector surface (a-Se), thin-film transistor "switch" (TFT), small capacitor to store electric charge X-ray conversion to direct conversion: a-Se used to convert x-ray energy into electric charge Ionization of selenium atoms by x-rays creates an electric-hole pair (pos charge) --> Dexel electrode neg charged (pos charge move down toward dexel electrode) --> Neg charge collected at top and drained off, pos charge moved into storage capacitor Active Matrix Array (AMA) in DC Contains gate lines and data lines To "read" info on DR detector, bias voltage in gate line changed from -5 to +10 volts --> sudden change opens gate to allow electricity to flow through and the charge stored up on the capacitor flows into data line TFT Array
Intensity Domain Operations
After the initial image is displayed any adjusting done is effectively a readjustment to the gradation processing: Windowing Construction of the original histogram Histogram analysis
Interpreting the Waveform
Amplitude (height)= gray level or pixel value being displayed Wavelength= the number of pixels the object occupies across the row (lateral size) Taller waves= darker objects Wider waves= larger objects that occupy more pixels across the row A tall skinny spike would represent a dark but small object
Analog vs Digital Imaging
Analog = Measures continuously changing signal (level of brightness, shapes and colors) X-rays enter as analog, converted to digital Analog system = x-ray energy --> light --> recorded just the way it is Digital = Recorded as multiple numeric values and divided into array of small elements that can be processed in many ways Digital system = Analog signal --> numbers that are recorded
Noise
Anything that interferes with the formation of the image Anatomic, radiographic, equipment and quantum mottle
Pixel Waveform
Appearance of a waveform represents alternating densities: Top peak: darkest pixels Dips: brightest pixels
Brightness
Appearance on the display monitor of the computer and is a function of the monitor's ability to emit light through the surface of the display (luminance) Amount of light emission of a display monitor Too light has excessive brightness, too dark has insufficient brightness Brightness (after processing) does not equal IR exposure (before processing) Brightness is a monitor control function that can change the lightness and darkness of the image on a display monitor, but it is not related to IR exposure Brightness not controlled by the IR, but by post-processing applied to the image data Term used for the luminous intensity of the display monitor's light emission (measured in candela)
Digital Processing Domains
Approaches to digitally processing an image (domains) Spatial location: Pixels are acted upon according to their location within the matrix Shade (Intensity):Pixels are operated upon based on their pixel values (how bright/how dark) Size (frequency) Based on the number of pixel per row in matrix Low-Frequency: Large objects/details, long waves High-Frequency: Small details, short waves
Dexel
Average attenuation coefficient that is measured by dexel must be digitized or rounded to the nearest available value in dynamic range
Image Quality Characteristics
Brightness, grayscale (contrast resolution), noise, spatial resolution (sharpness) and distortion Visibility (brightness and grayscale) and sharpness (spatial resolution and distortion) determine overall quality of the image
Sharpness
Characteristic of final displayed digital image Determined by matrix size, pixel size, and grayscale (pixel) bit depth Measured by point spread function (PSF), line spread function (LSF), modulation transfer function (MTF) and system noise MTF Measures the accuracy of an image compared to the original object on a scale of 0-1 (0 is no image, no signal - 1 is extremely high exactness (records image perfectly)) Measures the % of object contrast that is recorded Trueness or fidelity of an image Increase spatial frequency, decrease MFT Factors affecting sharpness: Spatial resolution problems should be approached in this order: eliminate motion reduce OID reduce focal spot size increase SID
Digital Radiography is a term used for
Computed Radiography and Direct Capture Radiography
Controlling Brightness
Controlled by Window Leveling. Adjusting window level changes that average gray level (center gray shade on the dynamic range) If window level is unchanged (when other parameters are changed), the average brightness level remains unchanged (setting correct mAs still critical, minimum change necessary to cause a visible shift in brightness is about 30% of mAs) Correct technique = adequate brightness (may display appropriate level of brightness but can be under or over exposed) Window level (L) corresponds to the pixel value that corresponds to the mid-gray brightness level. Increasing level makes image darker, decreasing image makes it lighter
Casette-Less Based Systems (Direct Capture/Digital Radiography)
DETECTOR AND READER ARE A PERMANENT PART OF A TABLE OR WALL UNIT No casette Direct ADVANTAGE: ENHANCED CONTRAST RESOLUTION
Multiscale Processing
Decomposing the original image into eight or more frequency layers, performing various operations on selected individual layers, and recomposing the image The original image is repeatedly split into a high-frequency component and a low-frequency component: The high-frequency component is set aside, and the low-frequency component is subjected to the next division until there are eight remaining levels After this is completed the image layers are complied back together to form the final displayed image using the inverse Fourier transformation
Magnification Factor
Degree of magnification and calculated by M = SID/SOD
Dynamic Range
Describes the concept of contrast as it is displayed on the displayed station and is a characteristic of the overall image system The range of brightness of the display monitor light emission Number of the shades of gray that can be represented Limited by the capacity of the computer system
Spatial Domain Operations
Digital image starts and ends in spatial domain Operations include magnification, translation (flipping), inversion (flopping), image subtraction and all kernal operations Spatial domain further divided: Point Processing Operations: Perform specific algorithm on each individual pixel in sequence, pixel by pixel. Ex: image subtraction. (Value contained in each specific pixel is subtracted from the value contained in the corresponding pixel from another image). Ex. Angiography Area processing operations: Execute a mathematical function on a subsection of the image. Ex: magnification (demarcate a portion of the image and push zoom, value for each single pixel will be spread out across an area of 4 hardware pixels) Global operations: Massive operation across the entire image. All image reorientations (inversion, flipping, or translating) --> translating leaves only the very middle column intact
Spatial Frequency
Digital imaging recorded detail Described as high or low A shorter wavelength (higher frequency) signal represent pairs of lines that can be visualized very close together (high resolution, capable of imaging smaller objects) Longer wavelength (lower frequency) signal represents pairs of lines that are further apart (low resolution, images larger objects)
Digitizing Data Info and Steps
Digitizing Data Limitations: Limit number of pixel values that can exist between pitch black and blank white (no limit = limitless values, we want discrete (limited) values) ADC (Analog to Digital Converter) = changes analog to digital Before processing Step 1: Scanning How many pixels involved? Image divided into matrix of pixels. Collimation like scanning because it selects which of these detector elements make up the initial matrix Step 2: Sampling Look @ each pixel and determine differences in pixels Detection and measurement of the intensity of signal at each pixel location Image formed through multiple samplings of the signal rather than one exposure of an analog image Step 3: Quantization Level out brightness level for every pixel to the nearest available gray level in the pre-set dynamic range Selecting displayed brightness level of each pixel
Beam Restriction
Easiest ways to improve contrast and lower patient dose
Fluorescence vs Phosphorescence
Fluorescence: Immediate emission of light be substance under stimulation Phosphorescence: delayed emission of stored energy in the form of light In CR: X-ray exposure constitutes a "first stimulation" of the phosphor plate resulting in immedate fluorescence, second simulation applied by laser beam (stimulated phosphorescence)
Distortion
Geometric property that affects image quantity Misrepresentation of size or shape of the structures being examine Understanding normal radiographic anatomy can help to recognize distortion Factors affecting size distortion: magnification all size distortion controlled by OID and SID post processing of digital images can resize reduced magnification increases spatial resolution Objects further from IR will be magnified Factors Affecting Shape Distortion: Misrepresentation by unequal magnification of the actual shape of the structure If the object plane and image plane are not parallel Degree of distortion affected by object's angle of inclination and lateral position from the central axis Reduce shape distortion by making body part and IR paralle, and central ray perpendicular Three conditions contribute to image distortion: image thickness, object position, and object shape Thicker object more distorted than thinner object Elongation caused by changes in tube angle or when IR is not aligned Foreshortening caused by body part being improperly aligned
Windowing
Gradient curve graphs tell us both the average brightness level (left-to-right position) and its contrast level (how steep the slope of the curve is) Default gradation processing (using the LUT) is done on every image before it is initially displayed on a monitor Brightness and contrast (grayscale) are the aspects that are controlled by windowing: Effectively adjustments to the anatomical LUT and are a reapplication of gradation processing Window level controls brightness Window width controls the grayscale
Rescaling
Has the power to align image brightness levels perfectly, but can only align overall image grayscale partially by aligning the min and max Q values The anatomical LUT is automatically set when the radiographer selects a radiographic procedure from the menu on the X-Ray console
Describing Contrast
High Contrast/ Short Scale = The differences between adjacent IR exposures that comprise contrast are great (fewer discernible shades of gray) Low Contrast/ Long Scale = The differences between adjacent IR exposures are minimal (more discernible shades of gray)
Other Frequency Detail Processing
High-Pass Filtering- passes through the highest frequency layers- Most commonly called edge enhancement Background suppression- elimination of the very lowest frequencies- In both cases, small details and the fine edges of structures visually will stand out better against the surrounding larger anatomy
SNR (Signal to noise ratio)
How much noise can be tolerated by the image Describes the strength of the radiation exposure compared to the noise apparent Increasing signal to noise ratio (means higher signal) means image quality is imrpoved But a decrease in noise will result in higher patient dose (increased mAs, more PE interactions) Optimizing SNR: Noise intolerance can lead to exposure creep (increase mAs) Set an appropriate target exposure value (IE) Routine monitoring The optimal image may not be the best image
Preprocessing: Segmentation
In CR, more than one image can be recorded on a single PSP plate Computer must sort out how many images are on one plate Inability to segmenent or separate individual exposure areas = segmentation failure Does not occur in DR systems
Conventional Vs Digital Imaging
In conventional X-Ray systems the film measure exposure by how much chemical change had occurred. (Ionizations led to chemical changes in the film) Conventional processing consisted of development, fixing, washing and drying In all modern digital X-Ray systems, the image receptor is an electronic detector Exposure is measured (X-Rays are counted) by how much electrical charge is built up. (Each measurement is recorded as a pixel value) Modern digital image processing consists of many generic steps
Latent Image formation
Invisible image on IR created by the unseen change in the atomic structure of the photostimulable phosphors
Source to image receptor distance (SID)
Major effect on magnification Greater SID = smaller magnification
Spatial Distortion
Misrepresentation in the image of the actual spatial relationships among objects As object position is shifted laterally from the central ray, spatial distortion can be more significant Single image not enough to define 3D configuration
Final Displayed Image
Mostly impacted by LUT (look up table), can be further adjusted by window width kVp and mAs are contributing factors
Matrix
NUMERICAL VALUES THAT IS STORED IN THE COMPUTER'S MEMORY Cells stored as rows and columns Each cell = pixel (picture element) Better quality = bigger matrix = more pixels
Object to image receptor distance (OID)
Objects further from IR = magnified OID must be minimized to decrease magnification
Frequency of Objects
Objects within the image also have a frequency related to the number of pixels they occupy in each row Large objects have large wavelengths and low frequency because fewer can "fit" across the screen Small objects have short wavelengths and high frequency because a high number of them will "fit" across the screen
Data Clipping
Occur when bit depth of the hardware or the dynamic range of the system are too limited to allow for windowing adjustments Data clipping limits the radiologist's ability to windoww the image and can aversely affect diagnosis
Frequency Domain Operations
Operations that are executed on structures or objects within the image rather than on the pixels Objects are identified, sorted and grouped by size in "bins" Detail processing operations: Smoothing Edge-enhancement Background Suppression Local contrast of only fine details
Background and Scatter Radiation
PSP plates used for CR are very sensitive to background radiation (1 mGy = noticeable fog, pre-fogging (before image was taken) different than scatter (after image was taken)) Background radiation (fog) can influence how the image is processed Scatter radiation is almost always corrected for during processing
Imaging Chain
Patient placed between x-ray source and IR Image technical and geometric factors selected Image capture and latent image formed
Digital Radiography - Indirect Conversion
Phosphorescent screen is laid over AMA (Direct conversion does not have this) --> Phosphor converts x-rays into light, which strikes AMA panel below --> Each dexel made of amorphous silicon (better interaction with light) Remnant x-ray beam strikes a phosphor screen (scintillator) that fluoresces when exposed to x-rays --> beneath this phosphor layer, we see the same active matrix array that was used in DC systems EXCEPT used with silicon instead of selenium
Steps of Digital Image Processing
Preprocessing: Operations designed to compensate for flaws in image acquisition Field Uniformity Corrections: Uneven distribution of the background density (even background density = easier to diagnose), compensate for aeras that are outside the range of uniformity, anode heel effect is partially compensated Del Drop-out Corrections: Computer looks for dead pixels. Computer uses noise reduction software (Kernel) --> reads and averages the pixel values in selected pixels surrounding the dead pixel and inserts the number into the dead spot --> called interpolation Noise/Mottle Drop-out Corrections: Quantum Mottle (random) --> within the x-ray beam and is randomly distributed (poisson distribution) --> best corrected using Kernal. Electronic Mottle (periodic) --> Apperas with consistent size at regular intervals, best corrected using frequency filtering algorithms (frequency processing) Image and Histogram Analysis: Constructing a histogram --> count is made of all pixels sharing the same pixel value (density or brightness) and is done for all pixels at dynamic range. Histograms displayed as bar graph that represents the brightness value of each pixel. Read left to right (light to dark). Each body part displays a specific histogram shape. Computer compares actual histogram to expected histogram. Eliminates extreme data that will skew the rescaling. Rescaling/Normalizing (Processing): Brightness and gray scale/contrast are manipulated until looks "normal". Rescaled (not raw) image is subjected to post-processing. Final displayed image has an ideal level of brightness and balanced grayscale regardless of technique on initial exposure (gives flexibility w/ technique) Types of Histogram Analysis: Must be selected at menu Type 1: Designed to analyze two-lobe histograms with a tail spike representing background densities (chest imaging) Type 2: Designed for a single-lobe histogram (abdominal imaging) Type 3: Designed for three lobes with some metallic material present (contrast exam) Errors in Histogram Analysis Segmentation Errors Failure to match histogram to actual image acquired Extreme circumstances Pre-fogging Extreme amount of scatter radiation Postprocessing: Customized refinements specific to the radiographic procedure Gradation Processing (LUTs) Detail Processing Preparation for Display
Final Displayed Image
Primary role of radiographic technique is to produce sufficient signal reaching the IR for the computer to successfully process Digital processes almost always produces a displayed image that is diagnostic Corrections only fail under unusual and extreme circumstances
Exposure Indicator
Provides a numerical value indicating the level of radiation exposure to the digital image receptor The much wider dynamic range of digital image receptors is more forgiving of exposure errors, and images can be produced with a wide range of receptor exposures, spanning three to four orders of magnitude. Lack of attention to this wide dynamic range gives rise to a phenomenon known as dose creep
Subject Contrast
Range of differences in intensity of the x-ray beam after it has been attenuated by subject results from differential attenuation Depends on kVp and the amount and type of irradiated material
Spatial Detail Processing with Kernals
Remember a kernel is a submatrix- A small matrix that is passed over the larger matrix of the whole image in order to change all of the pixel values mathematically The kernel is passed left to right along the entire first row of pixels, then indexed down to the next row and repeats, applying a mathematical algorithm along the way For every detail processing effect that can be achieved in the frequency domain, there is an equivalent process in the spatial domain that can achieve the same type of result through kernels The resulting effect is smoothing, noise reduction, edge enhancement or background suppression This can be done without transitioning between image domains
Dynamic Range Compression (DRC)
Removal of the darkest and the lightest extremes of the pixel values from the grayscale of a digital image Saves computer space Allows for windowing Portion of the dynamic range displayed is the grayscale of the image itself
Q Values and the "Permanent" LUT
Rescaling done either electronically or with software Q-Values are standardized labels that are assigned preset pixel values that represent certain brightness levels for pixels in the image Computer program takes incoming S values (data that has not been rescaled) and reassigns them as Q values (processed data) and puts them in the permanent LUT Regardless of what the incoming pixel values were, the output pixel values are always adjusted to the same output Q values set by the permanent LUT
Pixel size
Size of pixel directly related to the amount of spatial resolution or detail in image Smaller pixel = greater detail Bigger pixel = less detail Small pixel + greater/larger bit depth = great spatial resolution Pixel size can change when matrix changes Size of matrix determines size of pixel
Dynamic Range
Why digital film is forgiving Range of pixel values (shades of gray) made available by the entire computer system (hard and software) to build final displayed image Dynamic range = subset of the bit depth of system Displayed grayscale (contrast) = subset of dynamic range Bit depth system --> dynamic range --> displayed grayscale (contrast)
X-ray conversion to direct conversion:
X-ray conversion to direct conversion: a-Se used to convert x-ray energy into electric charge Ionization of selenium atoms by x-rays creates an electric-hole pair (pos charge) --> Dexel electrode neg charged (pos charge move down toward dexel electrode) --> Neg charge collected at top and drained off, pos charge moved into storage capactior
Computed Radiography
Stimulate film-based radiography IR is a photostimulable phosphor (PSP) plate Cassette is plastic w/ low absoprtion Components of the PSP Plate Image Stimulated Phosphorescence: Occurs in compounds like barium fluorochloride and barium fluorobromide that have tiny defects (metastable site) in their crystals Each metastable site can "trap" free electrons and store them until something excites them Each time the barium-halide molecules are ionized, some of the freed electrons become trapped in these metastable sites CR Processing: PSP plate removed from cassette by the processor and scanned by helium-neon red laser beam that moves across the plate and then indexes down one row at a time --> metastable sites activated by EM energy from laser beam and emit (dim) light --> image has to be electronically amplified before it can be displayed CR Reader (Processor) PSP plate pulled by suction cups and rollers Laser beam deflected off a rotating mirror to make it scan across the PSP plate (fast scan) Direction in which the PSP plate itself is advanced by the rollers is called the slow scan Size of CR pixels determined during processing (DR pixel size determined by IR) Fast scan - pixel width Slow scan - pixel length Fast and slow scan frequency (how fast they're moving) must be equal to prevent distortion Erasing process: After scanning, PSP plate moved by rollers into the eraser section of the reader --> PSP plate exposed to intense white light (electrons moved back to ground state) to remove any info --> plate reloaded and ejected from machine
Detail Processing
Structures in the image can be selected according to their size and singled out for contrast enhancement or suppression Very small details can have their local contrast increased- making them stand out Mid-size structures can be suppressed by contrast reduction and "moved" into the background How does it work? Digital algorithms treat fine details of an image separately The image is sorted by the size of the object and placed in its own file (bin) Contrast or brightness can be manipulated on each individual object Decoupling local contrast from global contrast- Close inspection shows that the fine details are more visible, but the image as a whole has about the same amount of contrast
Pixel
THE SMALLEST ELEMENT IN A DIGITAL IMAGE MATRIX THAT CAN REPRESENT ALL AVALIABLE PIXEL VALUES EACH PIXEL ASSIGNED SINGLE NUMBER PIXEL VALUES REPRESENT BRIGHTNESS ASSIGNED TO PIXEL'S LOCATION PIXEL IS DEFINED BY ITS PIXEL VALUE AND LOCATION IN THE MATRIX CORRESPONDS TO A SPECIFIC LOCATION (AREA OF THE PATIENT'S TISSUE)
Gradation Processing
Tailor final image brightness and contrast according to the anatomy and predominant pathologies to be displaued Gradient curve used to describe both the brightness and grayscale as it related to IR (body of each curve represents a range of exposures that are typically used in daily practice) High gradient = high contrast Low gradient = low contrast
Magnification Radiography
Technique used principally during interventional radiology and mammography Enhances the visualization of small structures by delibreately increasing the OID while keeping the SID constant Degree of magnification is given by the magnification factor Small focal spot must be used for magnification radiography to help reduce the image detail Grids are not needed because of large OID Disadvantage: increased patient dose and a reduction in image detail/spatial resolution
Transition Between Image Domains
The image must be transition from the spatial domain into the intensity domain so that histogram analysis and gradation processing can be performed The computer must keep track of where the pixels are so they can be reconstructed after the intensity operations are completed Frequency detail processing requires that the image be transitioned from the spatial domain into the frequency domain then placed back into the spatial matrix after detail processing is completed
Fourier Transformation
The mathematical process that allows for the ability of frequency processing to separate structures according to their size Breaks up the complex waveform (sum of pulses with different wavelengths) into component waves, long, medium and short
Actual Dynamic Range
The maximum number of shades of gray that can be represented by a digital imaging system is the numeric range of each pixel or "bit depth." The actual dynamic range (capabilities of the overall system) may be less than the bit depth (hardware components)
Understanding the Frequency Domain
The zero-point of each individual wave corresponds to the transition border between each pair of pixels The wavelength of each pulse represents the pixel size The frequency of the waveform is from the left edge of the display monitor screen to the right edge and the number of up-down cycles that make up this distance- 600 pixels in a row= 300 cycles or 300 hertz (Hz)- Each cycle is two pulses and each pulse equals 1 pixel Smaller pixels result in a higher frequency
Digital Radiography (Review)
Two types: DR (Digital) - Directly connected to digital processor electromagnetically, Uses TFT (thin film transistor), two types (direct conversion and indirect conversion) CR (Computer) - Uses PSP (photostimulable phospher), manual process
Cassette Based Systems (Computed Radiography)
USE A PHOTOSTIMULABLE STORAGE PHOSPHOR IMAGING PLATE INSIDE A CASSETTE (like casette-based film) Indirect: Radiographer moves plate
Exposure Errors
Underexposure: Exposure to IR is too low, excessive quantum noise may be visible Overexposure: When IR is extremely overexposed, saturation may occur (dexels become overwhelmed with info)
Adjusting the Gray Scale
Window Width Digital systems algorithms provide linear range (histogram and look up table) Image processing will provide the proper grayscale regardless of most variations in kVp and mAs Look Up Table: Contains stored data to substitute new values for each pixel during processing Provides the proper grayscale, regardless of variations in kVp and mAs Consistant images Needs correct histogram selection Can't compensate for exposure values far outside normal range Window Width (W) determines the range of pixel values that will be incorporated into the display width Increasing W = reduce display contrast Decreasing W = increase display/brightness interval between two consecutive pixel values
Voxel
Volume of tissue within patient Data will be collected and averaged by dexel (square detector element) below it
Quantum Noise (Mottle)
When there are two few x-rays captured by the IR to create a latent image Can originate from materials used in the IR, electrical current, and computer algorithms
Detective Quantum Efficiency (DQE)
a measurement of the efficiency of an image receptor in converting the x-ray exposure it receives to a quality radiographic image No information lost during conversion = 100% or 1.0 DQE Higher DQE = Easier to produce good image, lower patient dose
Subject Contrast and kVp
kVp prime controller As long as kvp is adequate to penetrate the part being examine, low kvp will produce high subject contrast Subject contrast varies as kVp varies No amount of mAs can ever compensate for inadequate kVp Use 15% rule If electron's cant push (kvp), the amount (mas) won't matter
Pixel bit depth
max range/amount of pixel values that are stored (number of bits within a pixel) number of gray tones a pixel can produce is 2 to the power of bit depth. pixel bit depth of 8 is 256 shades of gray (2^8) Gray level factor in determining image contrast resolution shades of gray (gray scale) determined by pixel bit depth more gray scale = less pixelation
