Seeram Chapter #3: Digital Imaging CT Seeram Chapter #3: Digital Imaging
Analog vs. Digital Information continuous information Can have any of an infinite number of values Digital discrete information Can have a finite number of values limited by # of digits on display # of bits used to represent value
Analog vs. Digital Images continuous spatial information Digital discrete spatial information
Digitizing a Picture Commercial scanner Renders a photograph into numbers 311, 255, 309, 78, 43, 99, 124,…
Analog vs. Digital Images continuous gray shade information Digital Discrete gray shade information
Digital Image Formation Screen Wire Mesh Clinical Image
Digital Image Formation: Sampling Place mesh over image Assign each square (pixel) a value based on density Pixel values form the digital image 194 73 22
Digital Image Formation: Sampling Each pixel assigned a value Value averages entire pixel Any spatial variation within a pixel is lost The larger the pixel, the more variation 194 73 22
Digital Image Formation The finer the mesh (sampling), the more accurate the digital rendering
What is this? 12 X 9 Matrix
Same object, smaller squares 24 X 18 Matrix
Same object, smaller squares 48 X 36 Matrix
Same object, smaller squares 96 X 72 Matrix
Same object, smaller squares 192 X 144 Matrix
The Bit Fundamental unit of computer storage Only 2 allowable values 1 Computers do all operations with 0’s & 1’s BUT Computers group bits together
Popular Bit Groupings Bit (binary digit) Byte Word Double Word Smallest binary unit; has value 0 or 1 only Byte 8 bits 28 = 256 unique values Word 16 bits 216 = 65536 unique values Double Word 32 binary bits (1110 0100 0000 1011 0101 0101 1110 0101)
# of values which can be represented by 1 bit 2 unique combinations / values 1 2
# of values which can be represented by 2 bits 1 2 4 unique combinations / values 3 4
# of values which can be represented by 3 bits 5 1 6 2 7 3 8 4 8 unique combinations / values
Digital Image Bit Depth bit depth controls # of possible values a pixel can have increasing bit depth results in more possible values for a pixel better contrast resolution 1 2 3 . 8 0, 1 00, 01, 10, 11 000, 001, 010, 011, 100, 101, 110, 111 00000000, 00000001, ... 11111111 2 1 = 2 2 2 = 4 2 3 = 8 2 8 = 256 Bits Values # Values
# of Possible Values & Contrast Resolution The more possible values for a pixel, the more gray shades & the better the contrast. 4 grade shades 256 grade shades
Digital Image Formation Quantization (A to D Conversion) Process of assigning a number to a gray shade Only discrete #’s assigned can lose information because of discrete # assignment 88 ? 89 The middle pixel attenuates between the other two. What # will the A to D converter assign it?
Analog to Digital Converter 88 ? 89 Since there are no #’s between 88 & 89 (88.5 not allowed), the A to D converter will assign pixel either a 88 or a 89. The fact that the center pixel is darker than the left one and lighter than the right one is forever lost.
Contrast Resolution difference in x-ray attenuation required for 2 pixels to be assigned different digital values 88 89
Gray Scale the more candidate values for a pixel the more shades of gray image can be stored in digital image The less difference between x-ray attenuation required to guarantee different pixel values See next slide
1 2 3 4 5 6 7 Setting pixel values 1 2 3 4 5 6 7 11 8 10 14 12 9 13
Display Limitations 17 = 17 = 65 65 = = not possible to display all shades of gray simultaneously window & level controls determine how pixel values are mapped to gray shades numbers (pixel values) do not change; window & level only change gray shade mapping 17 = 17 = Change window / level 65 65 = =
Presentation of Brightness Levels Pre-processing Assignment of values to a pixel In CT values assigned according to attenuation Post-processing Each pixel value assigned a brightness Dynamic process Assigned brightness for a particular pixel values can be changeg Window Level Does not affect image data 125 25 311 111 182 222 176 199 192 85 69 133 149 112 77 103 118 139 154 120 145 301 256 223 287 225 178 322 325 299 353 333 300
Digital Image Sources CT MRI CR DR Digital Subtraction Angiography Ultrasound Nuclear Medicine
Why Digital? Required for computer use Perfect image copies Compression Reduction of image size in computer
Why Digital? Image Manipulation Rotation White/black reversal Zoom Window/level Enhancement Edge enhancement Smoothing (noise reduction) Analysis Image statistics Pattern recognition
Image Processing Techniques Point Operations Spatial Frequency Filtering Geometric Operations Some Operation Output / Display Pixel Values (Gray shades) Input Pixel Values Don’t Change Altered by Operation
Point Operations Value of each pixel altered according to some rule New pixel values assigned on pixel by pixel basis independent of adjoining pixels Example: Gray level mapping: window / level Look-up table (LUT) altered Maps pixel value to gray shades
Histogram Graph showing # of pixels at each gray shade Altered by point operations 1 2 3 4 5 6 7 8 9 10 11 Pixel Value # 1 2 3 4 5 6 7 8 9 10 11 Pixel Value #
Local Operations AKA Area processes Group processes Modification of input pixel based upon values of pixels close by
Local Operations High frequency image Low frequency image brightness changes rapidly with distance Small pixels required Low frequency image brightness changes slowly with distance
Spatial Frequency Filtering Can increase or decrease brightness changes with distance Increasing Sharpens image Increases noise Decreases Blurs image Smoothes image Decreases noise
Sharpening Image Original Image Sharpened Image Note Increased Noise
Note Decreased Noise & Blurring Smoothed Image Smoothed Image Note Decreased Noise & Blurring Original Image
Geometric Operations Scaling Sizing Rotation Translation Modifies Orientation of pixels Spatial position of pixels
Image Processing Hardware Image Memory Image Processor Digital to Analog Converter Host Computer
Image Processing Hardware Image Memory Temporary storage used while processing / displaying image Image Processor Digital to Analog Converter Host Computer
Image Processing Hardware Image Memory Image Processor Computer responsible for processing (arithmetic) done on input digital image Digital to Analog Converter Host Computer
Image Processing Hardware Image Memory Image Processor Digital to Analog Converter Converts output pixel values from Look-up table to analog voltage required by monitor Host Computer 125 25 311 111 182 222 176 199 192 85 69 133 149 112 77 103 118 139 154 120 145 301 256 223 287 225 178 322 325 299 353 333 300
Image Processing Hardware Image Memory Image Processor Digital to Analog Converter Host Computer Directs above hardware Holds stored images Directs archiving