Help separate numbers from string of characters in a data.table

[KROD]

I would like to extract the meaningful numbers from the data imported as .txt.
Maybe separate based on number location in string for each column?

Image A: C1-Composite
Image B: C2-Composite

Pearson's Coefficient:
r=0.413

Overlap Coefficient:
r=0.6

r^2=k1xk2:
k1=1.082
k2=0.333

Using thresholds (thrA=58 and thrB=58)

Overlap Coefficient:
r=0.879

r^2=k1xk2:
k1=0.709
k2=1.09

Manders' Coefficients (original):
M1=1.0 (fraction of A overlapping B)
M2=0.999 (fraction of B overlapping A)

[woodward]

I don't understand the question. Please post your code and your data.