A special problem of integral geometry (Q2377501)

In this paper the author considers a problem of integral geometry that can be viewed as a generalization of typical inverse problems arising in \(X\)-ray tomography. Let \(G\subset \mathbb R^n\), \(n>1\), be a bounded domain of diameter \(d\), containing a system of disjointed domains \(\{G_i\}\), \(i=1,2,\dots,p\), such that \(G_0=\bigcup_{i=1}^nG_i\) is dense in \(\overline{G}\). Let \(\Omega\) be the unit sphere of \(\mathbb R^n\), \(B=\{v\in \mathbb R^n,\:0<r_1<|v|<r_2,\:r_1<1<r_2\}\), and let \(g(x,y,\omega)\) be a function defined on \(G\times G\times B\) such that, for any fixed \((x,\omega)\in G\times \Omega\), \(g(x,x+t\omega,\omega)\) is piecewise continuous in \(t\), for \(t\in (-d,d)\). The problem is the determination of \(\partial G_0\setminus\partial G\) from the knowledge of the integrals \(H(x,\omega)=\int_{-d}^dg(x,x+t\omega,\omega)dt\), for all \((x,\omega)\in G\times\Omega\). In the case where \(g\) depends only on \(y\), and \(n=2\), then the usual Radon transform of \(g(y)\) is obtained, and uniqueness theorems are known not only for discontinuity surfaces of the integrand, but also for the function \(g(y)\) itself. Here the author takes a very general point of view, and, though the task is consequently reduced just to the determination of the discontinuity surfaces of the integrand, the approach seems to be useful in medical tomography, where the knowledge of contact boundaries is a very important information for the reconstruction of the structure to be investigated. Uniqueness conditions for the solution to the given problem are discussed through methods which are also convenient for numerical algorithms