Probabilistic contrast measure example.
(a) Panel a shows a simplistic two tissue discrimination. On the left the probability distribution for the thin cuticle tissue over a 16 bit greyvalue range is shown. The probability distribution on the right shows the greyvalues of small muscle tissue type. The highlighted section indicates the overlap between the two distributions and therefore the discrimination error when applying a blanket threshold – in this example the threshold would be at around 0.25 on the x-axis. The area of the highlighted region gives the probabilistic tissue contrast (PTC) as a value between 0 and 1. (b) Panel b shows the same comparison of tissue probability distributions with all other measured tissue types included. The small muscle tissue type is contrasted against all other tissue types to determine its overall discriminability. (c) Panel c shows the same distribution as in panel b, zoomed-in into the region of interest which shows the pairwise overlap between the probability distributions of the small muscle tissue type and all other tissue types. (d) Panel d shows the cumulative overlap between the small muscle tissue type probability distribution and all others, giving the overall PTC for the small muscle tissue type. In this case, the value is close to 1, giving a low degree of discriminability.