pandas to_latex() escapes mathmode

pandas to_latex() methode seems to be a convenient way to convert bigger tabulars into into latex code. However, writing math mode, the $ seems to get escaped.

An example:

import pandas as pd
import numpy as np

table=np.asarray([['$x^2$',3]])
df=pd.DataFrame(table[:,1], columns=['value'], index=table[:,0])
print(df.to_latex(encoding='utf-8'))

output:

\begin{tabular}{ll}
\toprule
{} & value \\
\midrule
\$x\textasciicircum2\$ &     3 \\
\bottomrule
\end{tabular}

Is there a way to prevent this from happening?