Presentation on theme: "Principles of CT. Limitations of Radiography n Inefficient x-ray absorption: typically ~25% for par speed cassette (prior to rare earth technology) n."— Presentation transcript:
Limitations of Radiography n Inefficient x-ray absorption: typically ~25% for par speed cassette (prior to rare earth technology) n High Scatter-to-Primary Ratios: may have >50% scatter at receptor with large beams even with high ratio grid n Receptor Contrast vs latitude: required film dynamic range limits film contrast n Superposition/Conspicuity: overlapping structures with 3D anatomy rendered on 2D image
Early Attempts at CT n Gabriel Frank: 1940 Patent: described CT principles using optical backprojection reconstr (but no filter) n Takahashi (Japan, 40s, published 1956): describes equipment to image slices by backprojection n Tetelbaum et al (Russia, 1957): Accurate formulation of inverse Radon Transform; TV-based reconstruct n Kuhl & Edwards: (1963): cross-sectional NM images by back-projecting transmission data on oscilloscope n Alan Cormack: built simple CT to measure densities for radiotherapy. Shared Nobel Prize.
Godfrey Hounsfield and EMI: 1967 n Considered areas where much information available but inefficiently used: radiography n Estimated that if efficient detection/analysis, attenuation coefficients measurable within 0.5% from transmission measurements ---> sufficient to distinguish soft tissue differences n Invisioned slice divided in small voxels n Experiments using Americium source (9-day acquisition) verified 0.5% accuracy achievable
Hounsfields CT Formulation n Measurement Ni written as sum of attenuation of pixel along path n Solve simultan- eous equations from data at many positions and angles n Experiments achieved 0.5% accuracy.
1st Generation Data Collection n 1 Pencil Beam and 1 NaI detector n 160 samples/traverse n 1 o increms over 180 o n 28,800 samples n Solved simultaneous equations (Fortran) n 160 2 image matrix but reduced to 80 2 for practical clinical use
CT Numbers: Hounsfield Units Example 1: voxel contains water (u p = u w ): CT# = 1000 x (u w - u w )/ u w = 0 CT# = 1000 x (u w - u w )/ u w = 0 Example: voxel contains air (u p 0): CT# = 1000 x (0 - u w )/ u w = 1000 x (-1) = -1000