40 Chapter 2 Figure 3. A single axial slice is shown (columns left to right) for the T1-weighted image and the ASL image in their native space before registration and for the results of the four registration methods in the standard space: regT1, regASLrigid, regASLaffine, and regASLdct of three trigonocephaly subjects (rows). In the standard space, a CBF image is shown with borders of the WM from the template (with a threshold at 50%) shown in red. Finally, the pseudo-CBF images of the three trigonocephaly patients are shown. Table 2. Odds ratios for receiving a higher rating (i.e., the odds are calculated as P(rating>k)/ P(rating≤k) Odds Ratio 2.5% 97.5% trigonocephaly 2.611 0.698 9.954 regASLrigid 0.332 0.143 0.770 regASLaffine 2.608 1.113 6.249 regASLdct 17.508 6.659 49.378 The resulting estimated probabilities for an MRI of a patient to receive a particular rating are visualized in Figure 4 and reported in Table 3. The highest probabilities of a good rating were estimated for regASLdct (0.38, 95% CI 0.19–0.60 for rating 4 and 0.56, 95% CI 0.38–0.72 for rating 3). RegASLaffine had the highest probability of receiving a rating of 3 (0.64, 95% CI 0.49–0.76), and a one in four chance of receiving a rating of 2 (95% CI0.12–0.42). For regT1 and regASLrigid ratings 2 and 3 were most likely (regT1 rating 3: 0.48, 95% CI 0.29–0.66, rating 2: 0.43, 95% CI 0.24–0.61; regASLrigid rating 3: 0.26, 95% CI 0.12–0.44, rating 2: 0.59, 95% CI 0.44–0.72).
RkJQdWJsaXNoZXIy MTk4NDMw