Skip to main content
. 2016 Jan 5;24(2):306–317. doi: 10.1038/mt.2015.219

Figure 4.

Figure 4

Pathway of data transformation for quantitative distribution and infection void analysis. Data transformation pathway to allow for quantitative distribution analysis by generating binary data labels of infected and uninfected tumor tissue from SPECT/CT images following infection. (a) Axial SPCET/CT planar image through representative MPC11 tumor-bearing BALB/C mouse 1d post administration of IV VSV-Δ51-NIS. White dashed outline denotes tumor location. (b) Threshold of radiotracer activity is set to distinguish infected from uninfected space. Lower left region of activity is the bladder. (c) A label is applied to identify tumor from nontumor space, resulting in (d) a binary data set of infected (black within white tumor space) and uninfected tumor tissue (white). (e) Distributive distance transformation determines the Euclidean distance to the nearest infected voxel for every voxel within the tumor volume. A heat-map is used to visualize the nearest infected neighbor distances. (f) The 30 largest nonoverlapping spherical infection voids are identified from the nearest infected neighbor distances and mapped in gray. (g) Theoretical random infection is simulated within the same tumor volume so that the tumor has the same burden of infection as in d. (h) Distributive distance transformation is performed on theoretical random distribution as in e, and (i) infection voids are identified as in f. (j) The cumulative distribution of the nearest neighbor distances describes the distribution of uninfected and infected space and can be used for comparison of distribution to theoretical random distribution. Theo1–Theo5 are the cumulative distributions of five independent theoretical random simulations, demonstrating the repeatability of the random simulation. Exp is the cumulative distribution of the experimentally observed nearest infected neighbor distances. (k) The 30 largest nonoverlapping infection voids in experimentally observed (gray) and theoretical random simulated (red) infection distributions can be compared.