Digital Image Processing

X-ray cannot show differences in tissue contrast that are less than..
10% (contrast resolution)
____ and ____ are fixed and limited in film?
Optical range and contrast
(you cannot window them)
The 4 steps in the digital process
Data Aquisition
Image Processing
Image display, storage, and communication
Image Formations and Representation
Data Aquisition
First step in the digital process
Attenuated data is collected by the detectors
Output signal is an electrical signal which is analog
Must go through the ADC to become a digital signal for computer processing
Image Processing
Performed by the digital computer
Takes the input digital image and processes it to produce an output image in the digital binary form
Image Display, Storage, and Communication
The output digital image must be converted into an analog signal before it can be displayed on the monitor using the DAC
Can be stored on a disk or sent electronically through networks
Image Formations and Representations of Analog Images
Analog images are continous images - X-ray images
Image Formation and Representations of Digital Images
Numerical representations or images of objects
Requires a digital computer
Any info entered into a computer must be converted into digital form (numbers)
ADC - converts continuous signals to discrete signals
Digital Image Processing
Converting an input image into an output image
Both analog - analog processing
Both digital - digital processing
CT is based on reconstruction processes where a digital image is changed to a visable physical image
A series of operations or actions leading to a desired result
Spatial Location Domains
Used in CT and radiography
The digital image is a numerical image arranged in a grid so the location of each number can be identified using the x-y coordinate system
x - columns
y - rows
(0,0) upper left corner
Spatial Frequency Domains
Used in MRI
Image domains
Digital image processing can change one domain into another domain
Use the Fourier Transform for this task
Characteristics of a digital image
Bit Depth
A 2 dimensional array of numbers that make up a digital iamge
Consists of columns (M) and rows (N) that are broken up into pixels
Dimensions are M x N x k bits
Matrix size = FOV - selected by the tech
In CT we use 512 x 512
Make up the matrix
Contains a number that represents the brightness level
This number is related to the atomic number and mass density of the tissue
Pixel size = FOV/Matrix size
Larger the matrix the smaller the pixels and better spatial resolution
Volume of tissue in the patient represented by pixels
Info is converted into numbers contained in the pixels
Numbers are assigned brightness levels
Higher numbers - high signal intensity, bright white
Low numbers - low signal intensity, dark
Bit depth
Number of bits in a pixel
M x N x k bits
Binary system uses a base of 2
Each pixel will have 2x gray levels
In CT we use a bit depth of 12
Matrix size effects ______?
spatial resolution (detail)
Smaller the pixel, the better the detail
Bit depth effects ______?
Contrast Resolution (The number of shades of gray)
More bit depth - more shades of gray
Image Digitization
Converting an analog signal into numerical data
3 steps of digitizing
Gray Scale
Total number of grays
Number of grays depends on bit depth
ADC Conversion
Measures the analog signal at various times to measure its strength at different points
Sampling is followed by quantization
Time taken to digitize the analog signal
More smaples you take the more accurate the representation of the digital image
Not enough samples are taken and the image is not accurate after computer processing
Point Operations
Most frequently used image processing technique
Value of the input pixel is mapped on the corresponding output pixel
Gray level mapping
Also know as windowing or point operations
1. Uses a LUT
2. Changes the brightness
3. Which modifies the histogram
Local Operations
Image processing which the output pixel is determined from a small area of pixels areound the corresponding input pixel
Deals with spatial frequency
Spatial Frequency Filtering
If a brightness of an image changes quickly it is high spatial frequency
Can use this is spatial domain and frequency domain
Filter in the spatial location domain
Output pixel is calculated by multiplying each input pixel by its corresponding weighting factor
Process is applied to all of the pixels in the input image
Each calculation requires 9 multiplications and 9 additions
Spatial Frequency Filtering Categories
High Pass Filtering
Low Pass Filtering
Unsharp (blurred) Masking
High Pass Filtering
Edge enhancement or sharpness
Sharpen an input image in the spatial domain that appears blurred
Spatial location is converted to spatial frequency and then uses a high pass filter to surpress the low spatial frequencies
Low Pass Filtering
Smoothes the input image
Output image will appear blurred
Reduces noise and brightness
Detail is comprimised
Unsharp (Blurred) Masking
Takes the blurred image produced by the low pass filtering and subtracts it from the original image
Global Operations
Entire image is used to compute the value of the pixel in the output image
Common global operation is the FT
Geometric Operations
Modifies the spatial position or orientation of the pixels in an image
Results in the scaling and sizing of images and roations and translation
Image Compression
Using software and hardware to reduce information by removing uncessary data
Remaining data is encoded and either transmitted or stored
User's equipment later decodes the informations and fills in a representation of the data that was removed
Lossless (reversible)
No info lost
Does not use quantization
2:1 3:1
Lossy (irreversible)
Some info is lost
Uses quantization
10:1 or higher
Low compression ratios are?
8:1 and lower
Image is still virtually acceptable
Compression Tolerance
Maximum compression in which the decompressed image is acceptable for interpretation and aesthetics
As compression ratios increase?
Need less storage space and faster transmission speeds
But bad image quality
Study by Haung 8:1 or less is visually acceptable
Image Prcessing Hardware
Image Storage
Image Display
Image Processor
Host Computer