Bioengineering 508: Physical Aspects of Medical Imaging http://courses.washington.edu/bioen508/ Bioengineering 508: Physical Aspects of Medical Imaging Organizer: Paul Kinahan, PhD Adam Alessio, PhD Ruth Schmitz, PhD Lawrence MacDonald, PhD Introduction to Medical Imaging 1. Medical Imaging Modalities 2. Modern Image Generation 3. Intro to Image Quality Imaging Research Laboratory http://depts.washington.edu/nucmed/irl/ Department of Radiology University of Washington Medical Center Adam Alessio, PhD Department of Radiology University of Washington Medical Center aalessio@u.washington.edu Nature of Medical Imaging For this class: Medical Imaging: Non-invasive imaging of internal organs, tissues, bones, etc. Focus on: 1. Macroscopic not microscopic 2. in vivo (in the body) not in vitro ( in glass, in the lab) 3. Primarily human studies 4. Primarily clinical diagnostic applications Nature of Medical Imaging QUICK CAVEAT Powerpoint Slides are just a vehicle for major topics These do not have all the information discussed in class! Taking notes to supplement slides is probably a good idea! 1
Types of Medical Imaging (Modalities) Grouped by underlying physics: X-Ray/CT Ultrasound Magnetic Resonance Imaging (MRI) Nuclear Medicine Optical Magnetic Field Electric Field Thermal Mainly research based Optoacoustic Elastography Major 4 that dominate clinical imaging, focus of this course Primarily microscopic Types of Medical Imaging (Modalities) Electromagnetic Spectrum Nuclear medicine For comparison, this is wavelength/frequency range of US, but US is NOT electromagnetic! Types of Medical Imaging (Modalities) Modern Image Generation Classifications of Medical Images 1. Anatomical vs. Functional Anatomy/Structure/Features vs. Physiology 2. Emission vs. Transmission Where does energy imaged originate? 3. Projection vs. Tomographic Projection--> 2D imaging, single plane, no depth information Tomographic ( tomo = slice, graphy=image) --> volumetric From continuous real world to a meaningful image (on computer): 1. Sampling Continuous Information Information and sampling technique varies widely for each modality- Topic for later lectures Computer can only hold discrete chunks of data Pixel = a single picture element; Voxel = a single volume element 2. Quantizing Samples Each discrete chunk must be represented by certain number of bits 3. Visualization Techniques of quantized, sampled image volumes 2
1. Sampling Continuous Information Given a signal such as a sine wave with frequency 1 Hz: We can sample the points at a uniform rate of 3 Hz and reconstruct the signal: We can also sample the signal at a slower rate of 2 Hz and still accurately reconstruct the signal: However, if we sample below 2 Hz, we don t have enough information to reconstruct the signal, and in fact we may construct a different signal (an alias): 3
Aliasing occurs when your sampling rate is not high enough to capture the amount of detail in your image Can give you the wrong signal/image an alias Where can it happen in graphics? During image synthesis: sampling continuous signal into discrete signal e.g. ray tracing, line drawing, function plotting, etc. During image processing: resampling discrete signal at a different rate e.g. Image warping, zooming in, zooming out, etc. Nyquist criterion: Must sample at two times the highest frequency in the signal for the samples to uniquely define the given signal F Nyquist = SamplingRate 2 Sampling below the Nyquist frequency can cause aliasing (CD sampling example) To perform sampling correctly in image space, need to understand structure of data/image Fourier: Any periodic function can be rewritten as a weighted sum of sines and cosines of different frequencies. - Fourier Series A sum of sines Our building block: Asin("x +!) Add enough of them to get any signal f(x) you want Which one encodes the coarse vs. fine structure of the signal? What would an image look like with a lot of high frequency content? What could you do to reduce speckled noise from an image? Fourier Transform 1D Example: A signal composed of two sine waves with frequency 2 Hz and 50 Hz The Fourier Transform of the signal shows these two frequencies In 2D: Usually represent low frequencies near origin, high frequencies away from origin Low Freq Signal f(x) Fourier Transform of f(x) frequency 4
2D Fourier Transforms Image in space domain (magnitude of frequency component) 2D Fourier Transforms (log magnitude of frequency component) Image in space domain (magnitude of frequency component) (log magnitude of frequency component) Original After low-pass After high-pass Frequency Content Frequency Content 5
Modern Image Generation 2. Quantization From continuous real world to a meaningful image (on computer): 1. Sampling Continuous Information Information and sampling technique varies widely for each modality- Topic for later lectures Computer can only hold discrete chunks of data Pixel = a single picture element; Voxel = a single volume element 2. Quantizing Samples Each discrete chunk must be represented by certain number of bits 3. Visualization Techniques of quantized, sampled image volumes Only have finite storage available for each picture element Digital images have digitized intensity values. Continuous values are quantized into discrete values. Example: Truecolor on computer displays use 24 bits for each pixel (8bits blue, 8 bits red, 8bits green=256x256x256 possible colors) Many medical imaging modalities use intensity values of 12 bits per pixel. (2^12=4096 possible gray levels) Color depth 8 bits per pixel 5 bits per pixel 4 bits per pixel 3 bits per pixel 2 bits per pixel 1 bit per pixel 6