Figure 3
From: Quality measures for fully automatic CT histogram-based fat estimation on a corpse sample
(a) Monte Carlo simulated CT image YA,r with components fat (dark grey), muscle (light grey), air (black). The true image parameters were: EF+ = −80 HU, EM+ = 50 HU, SF+ = SM+ = 70 HU, z+ = 0.3. (b) Grey-scale value histograms X (empirical), X(Θ) (fitted via CFES with α = 1), XF(Θ), XM(Θ) with: Thin line: measured, fat line: estimated, dotted: fat, dashed: muscle. (c and d) Scatterplots of mean true value μYA,r (τα+) vs. mean estimated value μYA,r (τα*) with z = 0.3. (c) The diagrams containing all of the synthetic images YA,r generated. (d) Diagram as in the upper right part only diagrams YA,r were considered with standard deviations S = SF+ = SM+ ≥ 70 HU. The parameter S = SF+ = SM+ labels the data-points. The weight exponent α = 1 was used in Sα.
