[en] The Oak Ridge National Laboratory (ORNL) computational phantoms have been the standard for assessing the radiation dose due to internal and external exposure over the past three decades. In these phantoms, the body surface and each organ are approximated by mathematical equations; hence, some of the organs are not necessarily realistic in their shape. Over the past two decades, these phantoms have been revised and updated: some of the missing internal organs have been added and the locations of the existing organs have been revised (e.g., thyroid). In the original phantom, only three elemental compositions were used to describe all body tissues. Recently, the compositions of the organs have been updated based on ICRP-89 standards. During the past decade, phantoms based on CT scans were developed for use in dose assessment. Although their shapes are realistic, some computational challenges are noted; including increased computational times and increased memory requirements. For good spatial resolution, more than several million voxels are used to represent the human body. Moreover, when CT scans are obtained, the subject is in a supine position with arms at the side. In some occupational exposure cases, it is necessary to evaluate the dose with the arms and legs in different positions. It will be very difficult and inefficient to reposition the voxels defining the arms and legs to simulate these exposure geometries. In this paper, a new approach for computational phantom development is presented. This approach utilizes the combination of a mathematical phantom and a voxelized phantom for the representation of the anatomy