w4 Flashcards
What conditions may impose limitations on the quality and information content of medical images?
Accessibility of the organ of interest Variability of information Physiological artifacts and interference Energy limitations Patient safety
Features of the brain are difficult to image because it is protected by the skull (the brain is not accessible). What is one way someone could image the arteries in the brain?
requires the injection of an X-ray contrast agent and the subtraction of a reference image.
what are image artifacts and why do they occur
any feature which appears in an image which is not present in the original imaged object.
An image artifact is sometime the result of improper operation of the imager, and other times a consequence of natural processes or properties of the human body.
similar in meaning to interference and noise.
The distinction between a normal pattern and an abnormal pattern is often clouded by significant overlap between the ranges of the features or variables that are used to characterize the two categories. The problem is compounded when multiple abnormalities need to be considered. Imaging conditions and parameters could cause further ambiguities due to the effects of subject positioning and projection.
How may this variability of information affect the imaging of a cancerous breast tissue? (3 ways)
- most are irregular and spiculated in shape, whereas benign masses are smooth and round or oval. However, some may present smooth shapes, and some benign masses may have rough shapes.
- A tumor may present a rough appearance in one view or projection, but a smoother profile in another. Furthermore, the notion of shape roughness is nonspecific and open-ended.
- Overlapping patterns caused by ligaments, ducts, and breast tissue that lie in other planes, but are superimposed to a single image plane in imaging, could also affect the appearance of tumors in images. Can use multiple views and spot magnification, but at the cost of additional radiation dose to the subject.
what is the measure used to represent the “energy” of an x-ray beam
The kVp, standing for kilo-volt-peak, is a commonly used indicator of penetrating capability, and often referred to as the “energy” of the X-ray beam.
how may energy limitations affect the acquisition of an X-ray mammography
low-energy X-ray photons are absorbed more readily than high-energy photons by the skin and breast tissues, meaning more energy is required to penetrate through the body. However, breast tissue is mainly soft tissue, meaning a lower kVp would be desired in order to maximize image contrast. A compromise is required.
how is contrast defined in medical imaging
the difference between the image parameter in a region of interest (ROI) and that in a suitably defined background.
e.g. If the image parameter is expressed in optical density (OD), contrast is defined as: C_OD = forground_OD - background_OD.
Contrast needs to be normalized, what are the two ways of normalizing
normal contrast: Cn = (f – b) / (f + b)
simultaneous contrast: Cs = (f – b) / b
what is simultaneous contrast, and why is its distinction from normal contrast important
Describes the way color is perceived when two image objects are placed next to each other e.g. if two images share color values but have different contrasts, and they are placed next to each other, the color values will not appear the same. Compared to normal contrast, simultaneous contrast better emphasizes this effect. Used for calculating Just-Noticeable Difference.
for a greyscale range of 0 to 255, what color is 0 and what color is 255
0 - black
255 - white
what is Just-Noticeable Difference and Weber’s Law
an experimentally derived threshold indicating at which simultaneous constrast value is the difference between forground and background noticible. Important in analyzing contrast, visibility, and the quality of medical images.
Experiments have shown that the JND is almost constant, at approximately 2% over a wide range of background intensities; this is known as Weber’s law.
how is Histogram defined in medical imaging
a characterization of image quality. A histogram of the intensity profile (probability density function) of an image.
h(rk) = nk
where rk is the k-th gray level/partition and nk is the number of pixels in the image having gray level rk.
How are histograms used to characterize image quality?
lets the user see the distribution of contrast throughout the image. low contrast imaging, such as MR Angiography, have narrow pixel distributions. This data can be used in medical image enhancement algorithms to increase the contrast based on the pixel’s bin.
what are some sources of noise
sources of noise could be physiological, the instrumentation used, or the environment of the experiment.
what is Digital subtraction angiography / Temporal Subtraction
a fluoroscopy technique used in interventional radiology to clearly visualize blood vessels in a bony or dense soft tissue environment. Images are produced by subtracting a “pre-contrast image” from an image taken when a contrast medium/agent has been introduced into a structure. Pixels are subtracted in a pixel-wise way.
what are the three common types of noise
Salt-and-pepper noise: random occurrences of black and white pixels
Impulsive noise: random occurrences of white pixels
Gaussian noise: variations in intensity drawn from a
Gaussian normal distribution
What are the two major categories of Digital Image Processing
spatial domain processing techniques: based on direct manipulation of pixels in an image
frequency (transform) domain processing techniques are based on modifying the Fourier (or others) transform of an image.
what is the mathematical expression which defines spacial domain processing techniques.
g (x,y) = T [ f (x,y) ]
where f (x,y) = input image, g (x,y) = processed image, and T is an operator on f, defined over some neighborhood of (x,y).
A neighborhood is an n x m region (> 1 and odd) surrounding (x,y) e.g. a 3×3 neighborhood about a point (x,y) has one layer of pixels circling it’s center, (x,y).
T is applied to each (x,y), which each have different neighborhoods, basically a convolution.
what are some other names for spacial domain processing
pixel group processing
mask processing
filtering
What is the special case called when the transformation function T (spacial domain processing) has a neighborhood size 1 x 1.
T becomes a gray-level / intensity / mapping transformation function of the form
s = T ( r )
where r and s are variables denoting, respectively, the gray level of f(x,y) and g(x,y) at any point (x,y).
This approach is referred to as pixel point processing.
what are function curves and what is its advantage.
a function that represents an attribute of a whole image and can be manupulated, e.g. brightness, where x = input pixel value and y = output pixel value. These attributes can be easily modified by manipulating the function curves without having to alter the image "manually" with a retouching tool.
Gray level transformations are among the simplest of all image enhancement techniques.
how are Gray Level Transformations implemented?
type of pixel point processing
the mappings from original to new values are implemented (stored) via lookup tables, and the values of the transformation function typically are stored in a one-dimensional array
What are some examples of Gray Level Transformation functions?
Linear Logarithmic Power-law Window-Level operation Pseudo-color table
why may using a linear gray level transformation function be useful in mammograms? more specifically, why may a negative image of a mammogram be useful?
it is easier to see fine details in a white background than it is in a black background, especially when the image is a film displayed on a light box. e.g. it is easier to see small, fine lumps when the lumps are black on white than they are white on black.
why may logarithmic gray level transformations be used?
Sometimes the dynamic range of an image exceeds the capability of the display device, and only one end of the spectrum is visible on display.
Logarithmic transformations allow you to extend one end of the spectrum while compressing the other (log expands darker range and inverse log expands lighter range)
why may one use a power law gray level transformation as opposed to a logarithmic gray level transformation, as they are both good for expanding one end of the spectrum while compressing the other.
it is simpler to get a variety of different transformation shapes, as you only need to change the exponent.
why may a gray level transformation that increases the contrast in dark areas and decreases the contrast in bright regions be useful in a clinical setting?
It can be used when the clinically relevant information is situated in the dark areas, such as the bronchi in the lungs, and the bright areas are not as relevant.
with respect to the the power law gray level transformation, what is the effect of having gamma (the power) < 1, gamma = 1, and gamma > 1
gamma < 1 : expands dark and compresses light (use when the detail you want is hidden within dark)
gamma = 1 : same as input.
gamma > 1 : expands light and compresses dark (use when the detail you want is hidden within light)
What is a Window-Level Operation gray level transformation?
an interval or window is selected in the original gray level range, determined by the window center or level l, and the window width w. Contrast outside the window is lost completely, whereas the range inside the window is stretched to the original gray level range a.k.a. Contrast Stretching.
why are window-level operations useful in clinical settings?
useful for highlighting an intensity band of interest. can use this method to isolate a grey-area of interest and increase its contrast, while totally blacking or whiting out the areas of non interest. e.g. the histogram of a lung would be bimodal, as the lung would be in the dark region and the bone would be in the light region. if you wanted to isolate the lung, you can use a window function where L is in the middle of the lung’s peak in the histogram.
what is the special case of window-level operations called when the window width = 0
binary image thresholding
basically turning the image black and white, where L determines the threshold at which pixels turn black or white.
why would you want to color grayscale images
Humans can discern thousands of color shades and intensities, compared to about only two dozen shades of gray.
how is color transformed in Pseudo-color table gray level transformations
usually the gray spectrum is mapped to a rainbow spectrum. The rainbow spectrum can be modified to fit intensity regions that highlight certain features of the image.
what is a Temporal Average gray level transformation, and why would you perform this operation?
averaging images
Averaging can be useful to decrease the noise in a sequence of images of a motionless object. The random noise averages out, whereas the object remains unchanged.
what does the histogram look like for the following types of images:
Dark
Bright
Low Contrast
High Contrast
Note that it is possible to develop transformation functions that can acheive these effects based only on information available in the histogram of the input image.
Dark: mostly low values
Bright: mostly high values
Low Contrast: low distribution
High Contrast: high distribution
what is histogram equalization
A mapping to increase the contrast in an image by stretching its histogram to approximately uniformly distributed.
The image that has been histogram equalized always has pixels that reach the brightest grey level.
how is histogram equalization performed
1) calculate the ratios of : # pixels in current gray level / total # of pixels.
2) calculate the new gray levels as so: current new gray level = the sum of the current gray level ratio + all of the previous gray level ratios (or the previous new gray level).
3) scale the new gray levels by multiplying them by the total number of gray levels. Round levels if necessary.
what are the other names for the neighborhood around a point (x,y) / the sub image in spacial domain processing
(spatial) filter
mask
template
window
what is the difference between spatial filtering and normal filtering
normal filtering usually refers to frequency domain signal processing. thus, the term spatial filtering is used to differentiate filtering operations on pixels in an image.
what is the other name for the mask coefficients in spatial domain processing
kernel
what does smoothing spatial domain processing do to an image and why is this used
used for blurring and for noise reduction. Removal of small details from an image prior to (large) object extraction. Bridging of small gaps in lines or curves.
what are some examples of linear/averaging smoothing filters and non linear smoothing filters
Linear/averaging smoothing filters : box filter, weighted filter
Non-linear smoothing filters : order-specific, median filter, max filter, min filter
what is the difference between box filters and weighted filters
box: each coefficient is weighted equally (produces average)
weighted: each coefficient is weighted differently
what are order-specific smoothing filters
response is based on ordering (ranking) the pixels contained in the image area encompassed by the filter.
e.g. median filter = value is replaced by median value within neighborhood. force points with distinct gray levels to be more like their neighbors
