1 and 2) Magnetic resonance angiography (as in imaging blood vessels, arteries in this simulation). Prostate imaging reporting and data system, and then the same for liver. Both are simple algorithms for interpreting studies of those organs along with a standard lexicon and technique to do them.
3) k-space is where MRI is acquired, in the frequency domain. Elliptical centric means we acquire from the center and then outwards in a spiral. This is significant because k-space is structured such that long wavelength/low frequency data are near the center and small wavelength/high frequency data are at the edges. Because it takes a finite, non-negligible time to image all of k-space (on the order of 15-40 seconds in our simulations) then it stands to reason that we need high signal during that whole 15-40 second duration so that we get information spanning the whole frequency gamut (as in image contrast in the middle of a vessel as well as sharp edges as defined by the high frequency components).
.
4) [Gd] is concentration of gadolinium, the contrast agent we use for MRI. SI is MRI signal intensity. [ I ] is iodine concentration. R1 is R1 relaxivity, same with R2* relaxivity. Those are fundamental MRI physics properties relating to time for longitudinal and transverse relaxation.
5) Axes are labeled. Signal intensity versus concentration of gadolinium. Note non-linearity and dependence on blood concentration.
6) Spoiled gradient, the type of acquisition we use for MR angiography. The highlighted term is important because this is a "T1 weighted" acquisition (which means related to R1 properties) yet the gradient sequence has a dependence on R2* as well, so that if R2* goes up then signal goes down. Lower signal is not a good thing.
7) Ok, this one isn't labeled. X is seconds, 0 being when injection was started. Y axis is arbitrary units of flux first and then signal intensity afterwards. Internal scaling is correct but the units don't mean anything. Point is that a square wave at the injector turns into a different shape by the time it gets to where we model, the renal artery in this case. Also note recirculation on the last image: even after it goes through once the contrast still goes through again and again, albeit at very low concentration.
8) This time the axes are labeled again. Gadolinium concentration on top. Signal intensity that results after running through our fake MRI simulation on bottom. Point is that high concentrations (green curve, fastest rate of injection) result in lower signal intensity, and that even low concentrations result in nearly as much signal. This is due to the R1 non-linearity and R2* degradation (SPGR slide).
9) These are all things we stick in computed tomography scanners in order to measure resolution, with the bottom right curve a modulation transfer function.
10) My sine wave phantom. Low-high-low wavelength is artifact from aliasing. Think of it as the long sine wave from slide 11 top of image cut up in little chunks and then stuffed in a 512 x 512 x 100 voxel (3-D pixel) box, because that's what it is.
11) This is the MR simulation in MATLAB in which I've been working for two years now. Sine wave gets stuffed in box, box gets run through simulated MRI (adding noise, Fourier transforms, using the gadolinium concentration as an input function applied to the sine wave in the box). What comes out is another sine wave that is sampled along the green line such that white -> 1, black -> 0. That's the sine wave at the bottom now in different axes. That's then sampled over 10 adjacent wavelengths on each side to get a max and a min at each point. Max - min is computed which then gives us:
12) a modulation transfer function, like that which we saw with the CT resolution testing. That's modulation (white - black after going through the MRI wringer) on the y axis, where one sees that modulation drops to 0 at the resolution limit. In this case the resolution limit is around 2.8 mm, which is because we stop sampling at about 1.4 mm. Apply Nyquist's theorem and voila, 2.8 mm.
13) This explains what the modulation transfer function results means. These are really simple simulated vessels with a stenosis/narrow region in each, run through the MRA simulation with different rates of injection. Slow is top left, fastest bottom right. Fastest should show the worst resolution since we have less modulation at smaller wavelengths per the modulation transfer function.
14) This blown up version of the slowest and fastest injections shows this to be true subjectively: slow wins this race in terms of having edge detail/lack of blurring.
15) Everything in this slide should ring a bell as it's all recap. The other sections are just filler that I threw in to come closer to the required time for this presentation and aren't really worth going over as they're not original.
