UPDATE: Using tangents, see e.g. this nice answer.
\documentclass[tikz,border=3.14mm]{standalone}
\usetikzlibrary{calc}
\begin{document}
\begin{tikzpicture}[thick,
dot/.style = {
draw,
fill = black,
circle,
inner sep = 0pt,
minimum size = 2pt
}]
\draw[smooth cycle, tension=0.6] plot coordinates{(-1,-.4) (1.1,-.6) (.9,.5) (-1.1,.5)} node at (1,.6) {$\Omega$};
\draw[blue] (-1.1,.5) coordinate(start) .. controls (-.5,-.2) and (.5,0) ..
coordinate[dot,pos=0.3,black] (A) coordinate[dot,pos=0.5,black] (B)
node[black,anchor=north] {$x_0$} coordinate[dot,pos=0.7,black] (C) coordinate[pos=0.92,black] (D) (1.1,-.6)
coordinate(end);
% see https://tex.stackexchange.com/a/76226/121799
\pgfmathsetmacro{\rsmall}{0.6}
\pgfmathsetmacro{\rbig}{0.75}
\node [draw=gray,circle,minimum size=\rsmall*1cm] (cA) at (A){};
\node [draw=gray,circle,minimum size=\rbig*1cm] (cB) at (B){};
\node[anchor=south,gray] at ([yshift=10]B) {$\gamma(\bar{D})$};
\node [draw=gray,circle,minimum size=\rsmall*1cm] (cC) at (C){};
\node[blue,anchor=south west] at (D) {$S$};
\begin{scope}[overlay] % see https://tex.stackexchange.com/a/76226/121799
\coordinate (cAB) at (barycentric cs:B=-\rsmall,A=\rbig);
\coordinate (cCB) at (barycentric cs:B=-\rsmall,C=\rbig);
\end{scope}
\foreach \Y in {1,2}
{\foreach \X in {A,C}
{\path (tangent cs:node=c\X,point={(c\X B)},solution=\Y) coordinate(pB\X-\Y)
(tangent cs:node=cB,point={(c\X B)},solution=\Y) coordinate(p\X B-\Y);}
\path (tangent cs:node=cA,point={(start)},solution=\Y) coordinate(pstartA-\Y)
(tangent cs:node=cC,point={(end)},solution=\Y) coordinate(pendC-\Y); }
\draw[red,dashed] plot[smooth] coordinates {(start) (pstartA-2) (pBA-2) (pAB-2) (pCB-1) (pBC-1)
(pendC-1) (end)};
\draw[red,dashed] plot[smooth] coordinates {(start) (pstartA-1) (pBA-1) (pAB-1) (pCB-2) (pBC-2)
(pendC-2) (end)};
\end{tikzpicture}
\end{document}

OLDER ANSWER: You started drawing this with Bezier curves (which I may not necessarily have done), so perhaps the best way is to continue this strategy. Yet I think it helps to make the circles nodes such that you can easily access the points on their peripheries.
\documentclass[tikz,border=3.14mm]{standalone}
\begin{document}
\begin{tikzpicture}[thick,
dot/.style = {
draw,
fill = black,
circle,
inner sep = 0pt,
minimum size = 2pt
}]
\draw[smooth cycle, tension=0.6] plot coordinates{(-1,-.4) (1.1,-.6) (.9,.5) (-1.1,.5)} node at (1,.6) {$\Omega$};
\draw[blue] (-1.1,.5) .. controls (-.5,-.2) and (.5,0) ..
coordinate[dot,pos=0.3,black] (A) coordinate[dot,pos=0.5,black] (B)
node[black,anchor=north] {$x_0$} coordinate[dot,pos=0.7,black] (C) coordinate[pos=0.92,black] (D) (1.1,-.6);
\node [draw=gray,circle,minimum size=0.6cm] (cA) at (A){};
\node [draw=gray,circle,minimum size=0.75cm] (cB) at (B){};
\node[anchor=south,gray] at ([yshift=10]B) {$\gamma(\bar{D})$};
\node [draw=gray,circle,minimum size=0.6cm] (cC) at (C){};
\node[blue,anchor=south west] at (D) {$S$};
\draw[red,dashed] (-1.1,0.5) to[out=-20,in=190] (cA.110)
to[out=10,in=170] (cB.90) to[out=-10,in=170] (cC.80) to[out=-10,in=150] (1.1,-.6);
\draw[red,dashed] (-1.1,0.5) to[out=-50,in=150] (cA.250)
to[out=-20,in=180] (cB.270) to[out=00,in=190] (cC.280) to[out=10,in=180] (1.1,-.6);
\end{tikzpicture}
\end{document}

\begin{document}missing,graphicblueis not a color defined in your MWE and you might want to reword the question such that it becomes clearer. – Nov 27 '18 at 14:49