Medische Beeldvorming -en analyse

Ga naar: navigatie, zoeken


Klik hier om de samenvattingen te bekijken


Ontvangen via VTK (meeste mensen die dit vak volgen zijn Master Burgerlijk Ingeniuer: Biomedische Technologie):


1. Ultrasound Imaging

a) What is the frequency range is used in US imaging. When is the high end of the spectrum used and when is the low end used?
b) Colorflow imaging. The incoming wave has a frequency of 2.5 MHz. How deep can the scanner make images without aliasing? Using this depth calculate the time resolution fo this image. Is this setup applicable?
c) Explain Doppler tissue imaging and strain imaging.
d) Which analysis imaging technique can be used to calculate velocity?

2. Medical Resonance Imaging: Three different MRI images were given of two syringes with fat and water in a aquous environment.

a) Draw the spin-lattice and spin-spin relaxation curves for fat and water and explain T1-, T2 -and proton density weigthed images. Explain the intensity difference in the three given images.
b) Draw the echo sequence for 3D SE.
c) TR=2440 ms and TE=20 ms. Calculate the acquisition time. Is this setup applicable? Why (not)? How can you shorten the acquisition time?
d) Adept the above drawing to a FLAIR 3D Turbo SE
e) Forgot last question

3. Image Analysis: Five top view mri images of the heart were given. Of the first picture the myocardial wall was delineated. A sixth colour figure was given. This was the average of the five figures.

a) Explain three sources of motion artefacts.
b) How are these artefact visible in the image?
c) Explain the parametrization of the transformation matrix.
d) What measure can you use for image registration?
e) Which image would you use as reference?
f) Which artefacts can be corrected for with a tranformation matrix
Take that a statistical shape model of the myocardial wall exists: v = \bar{v} + \sum\limits_{i=1}^{k}c_{i}q_{i}, where q_{i} are the eigennodes with eignevalues \lamda_{i}.
g) Derive an objective function that maximizes sharp edges.
h) How would you optimize this function?
i) Forgot this question.
j) (bonus) Why is it good for motion artifact correction to also do this optimization to the statitical model as well as all the individual pictures?


Image Processing & Analysis

  (a) Convolution of a given an image matrix and a 3x3 mask
  (b) Given training set of contours. Using PCA the contours can be modeled as v= avg(v) + ∑(k->v)c_k*q_k, the eigenvectors.
   - How q_k calculated from training set. 
   - Given contour v*, how are c_k calculated.
  (c) Explain a method that uses statistical shape shape fitting and global properties to delineate the left lung in a 2D chest radiograph.


  (a) T2 weighted FLAIR 3D SE sequence
   - Draw schematically this pulse sequence (RF,Gz,Gy,Gx,signal)
   - TR=4800ms, TE=400ms. How long is acquisition time?
   - How can acquisition time be reduced?
  (b) Given images explain the following artifacts.
   - Wrap-around artifact image.
   - Chemical shift artifact image.
   - Phase cancellation artifact image.

X-ray & CT

  (a) Biological effect of X-rays on tissue? Explain?
  (b) Which measures exists to express the delivered radiation dose? What is their definition? Unit?
      Assume patient undergoes radiation therapy with gamma rays as part of the treatment of a brain tumour.
      The radiation is targeted to the tumour with high precision. Which 
      measure(s) would you use to express the delivered dose to the tumour? Why?
  (c) What is the definition and unit of CT dose index?
  (d) In 2D fan beam CT we have seen reconstruction formulas that use projection from 360°. However, 
      it is possible to perform the reconstruction from a smaller rotation angle 
      β<360°. This reduce the acquisition time and rotation dose. What is the minimum β needed to complete the dataset?


Question 1 Given 2D gradient echo sequence with the following trajectory in k-space: (given on exam)

  (a) Draw schematically the corresponding pulse sequence (RF,Gz,Gy,Gx)
  (b) What is the influence on the imaging quality? 
   - If the trajectory is prematurely terminated, i.e. the outside of the k-space is not scanned? 
   - If the trajectory is completely traversed in opposite direction, i.e. from the outside to the inside of the k-space. 

Question 2 US and image analysis

  (a) What are the biological effects of ultrasound waves on tissue? 
  (b) Echocardiography uses US to acquire 2D or 3D images of the heart. Explain a method for the non-rigid (flexible) 
        registration of subsequent images in the heart cycle. 
  (c) Explain how the result of image registration can be used to calculate tissue velocity, strain and strain rate. 

Question 3 Iterative reconstruction Given the equations to derive Maximum Likelihood iterative reconstruction algorithm for PET/SPECT.

  (a) What is the meaning of mu_j , c_ij, q_i and r_i in the equations given? 
        What is known and what are the variables that must be solved in this equations? 
  (b) For CT, the following equation holds: I_i = I_o*exp(-sum(c_ij*mu_j)) 
        What is the meaning of I_i, c_ij and mu_j?
        Rewrite the above (given) equations for iterative reconstruction of CT images. Explain the new variables and/or 
        parameters that you use in these equations. 
        (Note: in CT the poisson distribution can also be used)

13/06/2014, 14:00

Duration of exam: 4 hours max. Type of exam: written examn, closed book, list of formula's allowed 1. X-ray imaging Assume a dedicated maxillofacial cone-beam CT scanner operating at 100 kV and 40 mAs. The beam thickness in the center of the FOV is 60 mm. CTDI_w = 90 mGy and the average regional conversion factor k = 2,0 * 10^(-3) mSv/mGy cm. a) Calculate the effective dose. Show the details of your calculations. b) Draw the X-ray beam intensity as a function of the wavelength and show how the kV and the mA influence the intensity? How do the kV and the mA influence the absorbed dose? c) What is the influence of dental amalgan fillings (a mixture of metals) on the image quality? How can the quality be improved in this case? Explain. 2. MRI T1-weighted spin-echo sequences of one 2D slice of phantom containing three different tissues are acquired. The repetition time (TR) is 400 ms. Field-of-view 300x300 mm. Number of phase encoding steps 240. Slice thickness 8 mm. - Tissue A with T1 = 80 ms and T2 = 50 ms - Tissue B with T1 = 400 ms and T2 = 100 ms - Tissue C with T1 = 2000 ms and T2 = 1000 ms Assume that the tissues all have the same proton density. a) Draw the T1-relaxation of each tissue as accurately as possible. b) Draw the T2-relaxation of each tissue as accurately as possible, taking the T1-relaxation and the repetition time into account. c) Assume an image with no T2-weighting

 - On a gray scale from 0 (dark) to 100 (bright), specify the gray values of each of the three tissues.
 - What is the acquisition time? Show the details of your calculations.

d) TE = 50 ms. Same questions as c 3. Image analysis A T1-weighted and a T2-weighted MRI scan of a patient's brain need to be spatially aligned. a) Which similarity measure would you use and why? b) below are the joint histograms of the intensities given for four geometric transformations. The intensity at position (x,y) in the joint histogram represents the number of voxels with intensity x in the T1-image and intensity y in the T2-image. Order the histograms from best to worst, i.e. start with the histogram that corresponds to the best transformation and end with the histogram that corresponds to the worst transformation. Explain your answer. 4. Ultrasound A 3D ultrasound system acquires an image of the heart with the following parameters: - transmission pulse frequency: 6000 Hz - ulgrasound frequency: 2,5 MHz - ultrasound speed: 1500 m/S - cone angle: 60° in lateral and in elevation direction - line density (lateral and elevation): 1 line per degree - number of scan lines detected in parallel: 12 - heart rate: 60 bpm a) What is the maximal depth of the tissue that can be visualized? Show the details of your calculations. b) Calculate the frame rate. Show the details of your calculations. c) Does Color Flow (CF) Doppler imaging influence the frame rate? Give an example. d) How can signals reflected by blood in the heart chambers be distinguished from those reflected by the heart muscle (tissue Doppler)?

18 juni 2008 (14u00)

1. Een CT-scanner met 64 detectoren met breedte 0.5mm wordt gebruikt bij pitch 1 voor een perfusie-opname van de hersenen (een gebied van 16 cm). a) gegeven het schema voor een cardiale CT. Pas dit aan voor het hierbovenstaande geval. b) hoeveel omwentelingen zijn er nodig om het gehele gebied in beeld te brengen? Duid dit ook aan op het aangepaste schema.

2. Gegeven 2 (onvolledige) pulsschemas voor diffusie-MRI (enkel de beginpuls en het uiteindelijke signaal zijn gegeven, het gaat om een spin en gradiëntsequentie). Vul ze aan en geef het schema in de k-ruimte.

3. Oefening dynamisch programmeren: gradiënt en doelfunctie gegeven. Bereken het kortste pad, waarvoor kan dit algoritme gebruikt worden, wat betekenen de verschillende termen van de doelfunctie, wanneer kan men gebruik maken van dynamisch programmeren?

Oude examenvragen

Kijk op de site van VTK.

18 juni 2008 (9u00)

1)MRI a) (fig. gegeven) Waarom is rand tussen water en vet donker?

b) (andere fig. gegeven) Waarom is die kolom op de figuur verschoven tov werkelijkheid? (bevat andere stof, pc weet dat niet, dus laat dit magnetisch moment overeenkomen met een iets andere locatie)

(bijvragen over artefacten, diffusie, MRA…)

2)CT Oef over dual CT a) Teken X-ray spectrum door verschillende posities (na filter1, na mens, na filter2…) (laagfrequente valt meer en meer weg)

b) Trek 2 detectorbeelden C en D van elkaar af om uiteindelijk bot niet meer te zien op beeld… (2 onbekenden: diktes bot en weefsel. Je bekomt stelsel van 2 vgln, waaruit onbekenden gehaald kunnen worden)

3)Oef met mutual information Werk oef uit met zwarte en witte vakjes (zoals die van oefenzitting, maar met ander aantal vakjes). (bijvragen over: markers, hoe CT op CT leggen? (=Som van kwadratische verschillen), als we geen gigantische matrix willen? (Grijswaarde-transfo en zo kleinere matrix bekomen))

Zowel prof als assistent zijn heel vriendelijk.