In this paper

a new approach to reconstruction of transverse cross sections of arteries from biplane orthogonal projections is proposed. A weighting distance is defined as measure of resemblance between adjacent slices. The measure of resemblance is optimized by modifying minimum flow approach. The connected constraint of cross sections and the random number are then introduced to obtain most connected cross sections. Experimental results show that 2D slice reconstruction which is of the most resemblance to the guide model and the most connected cross section is obtained under the projection data fidelity constraint using this method.