Fig. 5.

Box plot representing the dispersion between the layers thickness estimated with the proposed method M and the manual tracings performed by ${\mathcal{A}}_1$, compared to the inter- and intra-analysts variability, for the 26 pullbacks of the testing dataset. This corresponds to 10 images per pullback ($n=260$) for the method evaluation ($\text{M}-{\mathcal{A}}_1$), and 2 images per pullback ($n=52$) for inter- and intra-analysts variability. Percentiles are indicated by boxes ($25\mathrm{th}$ and $75\mathrm{th}$), inner lines ($50\mathrm{th}$), and error bars ($5\mathrm{th}$ and $95\mathrm{th}$)