Transcript MEGN 536 * Computational Biomechanics

MEGN 536 – Computational Biomechanics
Prof. Anthony J. Petrella
Basics of Medical Imaging
Introduction to Simpleware Software
Medical Imaging
 Used for measuring anatomical structures… size,
shape, relative position in body
 Can reconstruct geometry for modeling purposes
 X-ray techniques
 planar x-rays, mammography, chest x-ray, bone fracture
 CT scans – computed tomography
 Nuclear imaging, radioactive isotope
 planar imaging, bone scan
 positron emission tomography (PET)
 MRI – magnetic resonance imaging
 Ultrasound
Medical Imaging
Ultrasound (~1mm)
Ionizing
Non-ionizing
Broken molecular
bonds, DNA damage
May produce heating,
induce currents
Non-thermal, low
induction
X-ray Imaging (Roentgenogram)




Wilhelm Röntgen (1845-1923)
Nov 1895, announces X-ray discovery
Jan 1896, images needle in patient’s hand
1901, receives first Nobel Prize in Physics
Röntgen’s
wife, 1895
X-ray Imaging
 X-ray film shows intensity as a negative ( dark areas,
high x-ray detection)
X-ray Imaging
 X-ray film shows intensity as a negative ( dark areas,
high x-ray detection) = radiolucency
CT Imaging
 Computed tomography
 Tomography – imaging by sections or sectioning,
creation of a 2D image by taking a slice through a 3D
object
 2D images are captured with X-ray techniques
 X-ray source is rotated through 360° and images are
taken at regular intervals
 CT image is computed from X-ray data
CT Imaging
 Developed by Sir Godfrey Hounsfield,
engineer for EMI PLC 1972
 Nobel Prize 1979 (with Alan Cormack)
 “Pretty pictures, but they will never replace
radiographs” –Neuroradiologist 1972
early
today
Inhalation
Exhalation
How a CT Image is Formed
 X-ray source is rotated around body for each slice
 Patient is moved relative to the beam
 Figure below does not show it well, but the X-ray
beam has a thickness  each slice has a thickness
Note: slice thickness
http://www.sprawls.org/resources/CTIMG/module.htm
How a CT Image is Formed
 Figures below show only two views 90° apart
 A process of “back projection” is used to indicate
regions where X-ray attenuation is greater – i.e.,
tissue is more dense
How a CT Image is Formed
 Example at left w/ only 2
views shows poor image
 Clinical CT uses several
hundred views for each
slice
 Data collected in matrix
CT Image Data
 Recall that each CT slice has a thickness  each
element in the data matrix for a single CT slice
represents a measurement of X-ray attenuation for a
small volume or “voxel” of tissue
 X-ray attenuation is expressed in terms of the X-ray
attenuation coefficient, which is dependent primarily
on tissue density
CT Numbers
 CT numbers are expressed in Hounsfield units (HU)
and normalized to the attenuation coefficient of water
(atomic number)
CT Numbers & Viewing a CT Image
 CT numbers usually recorded as 12-bit binary number, so they
have 212 = 4096 possible values
 Values arranged on a scale from -1024 HU to +3071 HU
 Scale is callibrated so air gives a value of -1024 HU and water has
a CT number of 0 HU
 Dense cortical bone falls in the +1000 to +2000 HU range
0-2000 HU
1000-2000 HU
MR Imaging
 Magnetic resonance imaging
 1946: Felix Block and Edward Purcell discover
magnetic resonance
 1975-1977: Richard Ernst and Peter Mansfifield
develop MR imaging
 An object is exposed to a spatially varying magnetic
field, causing certain atomic nuclei to spin at their
resonant frequencies
 An electromagnetic signal is generated and varies
with spatial position and tissue type
 Hydrogen is commonly measured – hence, good
contrast for soft tissues that contain more water than
hard tissues like bone
MR Imaging – 30 Years Later
 “Interesting images, but will never be as useful as
CT” –Neuroradiologist (different), 1982
First brain MR image
Contemporary Image
Notes on CT v. MR Images
 CT image based on X-ray beam attenuation,
depends on tissue density
 CT images generally regarded as better for
visualization & contrast in bone imaging
 Bone density and modulus can be estimated
 MR image based on resonance of certain atomic
nuclei, e.g. hydrogen
 MR images generally regarded as better for
visualization & contrast in imaging soft tissues, which
contain more water than bone
3D Reconstruction
 CT & MR images represent 2D slices through 3D
anatomic structures
 2D slices can be “stacked” and reconstructed to form
an estimate of the original 3D structure
Simpleware Software
 Simpleware (http://simpleware.com/) is a leading
commercial software program for reconstruction of
CT & MR image data
What Data Format Does Simpleware Read?
 Most medical images are saved in the DICOM image
format
 What is DICOM?
 The standard for Digital Imaging and Communications in
Medicine
 Developed by the National Electrical Manufacturers
Association (NEMA) in conjunction with the American
College of Radiology (ACR)
 Covers most image formats for all of medicine
 Specification for messaging and communication between
imaging machines
 You don’t need to know the details of the format, but
software is happiest when reading DICOM images
What If You Don’t Have DICOM Data?
 You will need to use manual input methods with to
read the data
 You need to know something about the images
 A CT or MR scan consists of many slices
 We will be focused on bone modeling, so CT data will
be our main interest
 It is also important to remember how a CT image
slice is formed and what data it contains
Data in an Image File
 The format of CT numbers in the data file depends on the
precision of the binary data
 For CT numbers, we only need to cover the 12-bit range,
-1024 to 3071
 short has 2 bytes = 2 × 8 bits/byte = 216 binary values = 65,536
 When using unsigned shorts the data is shifted so all CT
numbers are positive  0 to 4095
Data in an Image File
 Recall a single CT slice is a matrix of data
 512 x 512 is a common size  262,144 pixels
 Each element in the matrix represents a pixel value
with a binary format of “short”, therefore each pixel
contains 2 bytes of data
 262,144 x 2 = 524,288 bytes, any additional data is
part of the “header”
Data in an Image File
 Visible Human link on class website
 Data are available for download
 Download sample of Visible Human
data from today’s Class Notes page
 These images are 512 x 512 and the data format is
unsigned short
 How large is the header (bytes)?
Data in an Image File
 512 x 512 = 262,144 pixels
 Each element in the matrix represents a pixel value
with a binary format of “short”, therefore each pixel
contains 2 bytes of data
 262,144 x 2 = 524,288 bytes, any additional data is
part of the “header”
 Total file size is 527,704  header is 3416 bytes
Starting Simpleware
 DO NOT use the search box on the Start menu
 You should find a ScanIP icon in the EG Apps folder
 Run Simpleware with the ScanIP icon
 Since we do not have a DICOM dataset, choose to
Import Raw Stack
Importing Visible Human CT Data
 Download ZIP file from class notes page and
uncompress in a separate folder
 Use following settings to import raw stack…
Working with Visible Human Data
 Now you are ready to work on reconstructing
geometry of the knee (next time)