Figure 6
Differences in deep learning model predictions based on which final time point is used as input. Left: the different mean liposome concentrations throughout the tumour. The red points are the measured accumulations calculated from imaging and the black lines are the predictions of the network. Each line goes through exactly two known points (the point marked with an ‘x’ at time and concentration zero, and the dot with matching colour to the line). Note that the prediction using the first time point (at 10 minutes) is omitted from the graph as it has a high maximum which makes the rest of the graph difficult to interpret. Right: the mean intratumoral IFP as a function of the time point used for predictions (with colour matched to the colour on the left). Notice that, with the exception of the first time point at 10 minutes, the mean intratumoral pressure remains relatively steady while changing the known data.
