Introduction to Medical Imaging Week 2: X-ray and CT Course 046831 Introduction to Medical Imaging Week 2: X-ray and CT Guy Gilboa
X-Ray First imaging modality (Discovered by Röntgen in 1895).
X-ray discovery Wilhelm Rontgen 1895 - Discovered and detected X-rays Used it as a first medical imaging modality. Nobel 1901, first Noble prize in physics was awarded to him. First x-ray image. Hand of Anna Rontgen (wife)
X-ray machines Standard machine C-arm
X-ray tube
X-ray tube diagram [Board (1)] Taken from http://used-medicalequipmentblog.blogspot.co.il/2011/05/advances-in-ct-scanner-x-ray-tubes.html
X-ray on the spectrum
X-ray energy spectrum [Board (2)]
Energy units eV = electron- Volt 1𝑒𝑉≅1.6× 10 −19 𝐽 It is equal to the amount of kinetic energy gained by a single unbound electron when it accelerates through an electric potential difference of one volt. Photon energy: 𝐸= ℎ𝑐 𝜆 1𝑒𝑉 𝜆=1240𝑛𝑚 kVp = peak kilo-voltage. The maximum voltage applied across an X-ray tube. Determines the kinetic energy of the electrons accelerated in the X-ray tube and the peak energy of the X-ray emission spectrum. Noise model – Poisson (see details in CT part) h – Planck’s constant 𝜆 - wavelength c – speed of light
X-ray measures the total attenuation across a line [Board (3-4)] Linear attenuation coefficient Photoelectric and Compton absorptions. I – X-rays intensity (transmitted through a certain thickness L of tissue): 𝐼= 𝐼 0 𝑒 −𝜇𝐿
Attenuation coefficients Linear attenuation coefficient N – number of X-rays transmitted through a certain thickness x of tissue 𝑁= 𝑁 0 𝑒 −𝜇 𝐸 𝑥 𝜇 𝐸 = 𝜇 𝐸 𝑝ℎ𝑜𝑡𝑜𝑒𝑙𝑒𝑐𝑡𝑟𝑖𝑐 + 𝜇 𝐸 𝐶𝑜𝑚𝑝𝑡𝑜𝑛
X-ray detectors Previously – film, analog radiography. Today – digital radiography Indirect conversion X-ray to light using a scintillator (CsI:T1) Light to voltage using photodiodes Direct conversion Thin-Film-Transistors Cadium-Teloride, Cadium-Zinc-Teloride – technology not mature yet.
Signal to Noise Ratio Noise – mainly Shot noise - Poisson distribution. Variance is SNR 𝜎 2 =𝜇≈𝑁 𝑆𝑁𝑅≝ 𝑆 2 𝑑𝑥 𝑛 2 𝑑𝑥 = 𝑐𝑁 𝜎 ∝ 𝑁
Standard optics constraints affect the quality of the X-ray image Taken from [1] http://www.medphysics.wisc.edu/~block/bme_530_lectures.html
Popular for checking chest (lung problems) and bone fractures
X-Ray Summary Advantages: Drawbacks: Does not give 3D info. Cheap and simple Low radiation (compared to CT) Drawbacks: Does not give 3D info. Bones can occlude significant diagnostic data.
Digital Mummography Used to detect small tumors or microcalcifications in the breast. Very high spatial resolution and Contrast to noise ratio (CNR) are need for these type of pathologies (often <1mm in diameter) Low radiation is important – avoid tissue damage and allow frequent usage. Low energy (e.g. 26 keV) is used – high contrast, low radiation, low penetration.
Computed Tomography (CT) 3D imaging using X-ray radiography
History - Invention of CT Sir Godfrey Hounsfield (English Electrical Engineer), built first CT 1971, scanned head. Allan McLeod Cormack - math framework. Nobel prize for both in 1979 for the invention of CT. 1975 – first full body scanner. Hounsfield sketch
CT scanner
CT is based on the physics of X-ray Generation by an X-ray tube. X-ray spectrum (higher energies) Attenuation in the body (mu) Poisson noise model.
CT – operation principle Taken from http://www.cyberphysics.co.uk/topics/medical/CTScanner.htm
Hounsfield unit Water: 0HU Air: -1000HU What is displayed in CT images? Typical medical scanner display: [-1024HU,+3071HU], Range: 12 bit per pixel is required in display.
Hounsfield units of tissues Taken from [1]
Rendering based on different HU thresholds Taken from https://www.sharbor.com/news/MSG/index.html
Helical CT Taken from [1]
Gantry
Scanning Geometry of a CT System From [1]
Multi Slice Detectors Taken from http://tech.snmjournals.org/content/36/2/57/F1.expansion.html
Visualization (extra): X-ray tube movie (3:26 min.)
We learned today Basic principles of generating X- rays: braking and characteristic radiations. How X-ray is absorbed in the body – photoelectric and Compton absorptions. Linear attenuation coeffient Poisson noise model with SNR proportional to square-root of intensity. CT - Hounsfield Units CT system basic structure Next week: CT recon and clinical uses.