Tomographic Diffraction Microscopy (TDM) enables quantitative, label-free three-dimensional imaging of transparent samples, but its performance is limited when applied to thick or structurally complex specimens. Depth-dependent signal degradation and asymmetric frequency sampling can compromise both image quality and quantitative accuracy. To address these challenges, a research team from IRIMAS (France-BioImaging Alsace node) has developed a dual-view TDM approach. By combining two opposite and complementary views of the same sample, this method improves the robustness, reliability and applicability of 3D TDM reconstructions.

Limitations of conventional transmission TDM

TDM relies on the diffraction of light as it propagates through materials with different optical properties. By recording the phase and amplitude of the diffracted light under multiple illumination angles, TDM reconstructs a 3D map of the sample’s refractive index and absorption without the need for fluorescent labelling.

However, in its conventional transmission configuration, TDM suffers from intrinsic limitations. As imaging depth increases, signal quality progressively degrades due to absorption, defocusing and multiple scattering. In parallel, incomplete and asymmetric sampling of spatial frequencies can lead to reconstruction artefacts, such as axial distortions or ambiguities between refractive index and absorption. These effects reduce the reliability of 3D reconstructions, particularly for thick, dense or structurally complex samples.

Dual-view TDM: principle and implementation

To overcome these limitations, the proposed dual-view TDM approach is based on the simultaneous acquisition of two tomographic datasets from opposite directions. Each view provides a full 3D reconstruction of the sample, but with different depth-dependent signal attenuation and complementary spatial frequency coverage.

In this configuration, structures that appear deep and poorly resolved in one view are closer to the surface and better resolved in the opposite view. The two reconstructed volumes are then spatially registered and combined using a dedicated fusion strategy that selects, for each axial plane, the view exhibiting the richest spatial frequency content. This process effectively compensates for depth-related signal loss and mitigates asymmetries in the Optical Transfer Function, resulting in improved axial resolution, reduced reconstruction artefacts and more homogeneous image quality throughout the volume. Both refractive index and absorption maps benefit from this enhanced robustness, leading to more faithful and reliable 3D reconstructions.

Applications in label-free imaging of complex samples

Dual-view TDM is particularly well suited for the label-free imaging of structurally complex samples, where conventional TDM may fail to provide reliable results. Typical applications include the study of samples such as diatoms and pollen grains, whose intricate surface features and internal compartments require homogeneous image quality across the entire sample. The method could also be relevant for small 3D cellular assemblies or spheroids, where depth-dependent signal degradation limits quantitative analysis.

By improving the robustness of refractive index and absorption reconstructions, dual-view TDM enables more accurate characterization of dense or absorbing regions, extending the applicability of TDM to a broader range of biological and biophysical studies without additional labelling or complex sample manipulation.

While dual-view Tomographic Diffraction Microscopy significantly improves the robustness of 3D TDM reconstructions, it does not fully eliminate the intrinsic limitations of transmission-only TDM, such as multiple scattering or anisotropic resolution in very complex samples. This approach nevertheless represents an important contribution in extending the applicability of label-free 3D imaging. This dual-view approach also paves the path towards 4Pi tomography, which aims at delivering isotropic-resolution imaging. Note that another French team at Institut Fresnel-Marseille works on a similar approach, called mirror-assisted tomography.

Readers interested in the technical details and experimental validation can consult the scientific article here.

Carlos Alberto Chacón Ávila, Nicolas Verrier, Matthieu Debailleul, Bruno Colicchio, and Olivier Haeberlé, “Dual-view tomographic diffraction microscopy” Opt. Express 33, 51444-51458 (2025)