An angular spectrum algorithm is presented for fast, near-field diffraction tomographic imaging with diffuse photon density waves in highly scattering media. A general relation in K space is derived that connects the spatial variations of the optical properties of heterogeneities to the spatial spectra of the measured scattered diffuse photon density waves. The theory is verified experimentally for situations when boundary effects can be neglected. We further describe how to reconstruct absorption and scattering properties simultaneously, and how to incorporate boundary conditions into this angular spectrum algorithm for a turbid medium of finite size (e.g., the slab medium). Limitations and potential improvements of the near-field diffraction tomography are also discussed.