Figure 1 | Scientific Reports

Figure 1

From: Automated feature quantification of Lipiodol as imaging biomarker to predict therapeutic efficacy of conventional transarterial chemoembolization of liver cancer

Figure 1

Segmentation and registration workflow. (a) Tumors were semi-automatically segmented on all three imaging time points (baseline MRI, 24 h CT, follow-up MRI). Compartment masks on MRI (viable/necrotic tumors masks) were produced using quantitative European Association for the Study of the Liver (qEASL), while compartment masks on 24 h CT (low/mid/high density Lipiodol masks) were generated using Hounsfield Unit thresholds. (b) Each of the two MRI tumor segmentation masks were diffeomorphically registered to the 24 h CT tumor mask. These two registration procedures (represented by arrows of differing shades of gray) each generated a separate transformation matrix. (c) This transformation matrix was used as a template to simultaneously register the corresponding images and compartment masks to 24 h CT.

Back to article page