Reconstruction based on adaptive group least angle regression for fluorescence molecular tomography

Abstract

Fluorescence molecular tomography can combine two-dimensional fluorescence imaging with anatomical information to reconstruct three-dimensional images of tumors. Reconstruction based on traditional regularization with tumor sparsity priors does not take into account that tumor cells form clusters, so it performs poorly when multiple light sources are used. Here we describe reconstruction based on an "adaptive group least angle regression elastic net" (AGLEN) method, in which local spatial structure correlation and group sparsity are integrated with elastic net regularization, followed by least angle regression. The AGLEN method works iteratively using the residual vector and a median smoothing strategy in order to adaptively obtain a robust local optimum. The method was verified using numerical simulations as well as imaging of mice bearing liver or melanoma tumors. AGLEN reconstruction performed better than state-of-the-art methods with different sizes of light sources at different distances from the sample and in the presence of Gaussian noise at 5-25%. In addition, AGLEN-based reconstruction accurately imaged tumor expression of cell death ligand-1, which can guide immunotherapy.

Fig. 1.

The procedures of processing and reconstruction for in vivo experiments data. (1) The fluorescence collection system and the CT imaging system distribute orthogonally. (2) Denoise fluorescence data, segment CT data based on a threshold, and map processed data into the FEM tetrahedron based on coordinate. (3) Reconstruct 3D FMT and fuse with CT.

Fig. 2.

Reconstructions of a digital mouse bearing a tumor and illuminated with a single light source (first simulation). Reconstructions were carried out based on the indicated algorithms. The white circles in the plane view demarcate the actual fluorescence source.

Fig. 3.

Reconstructions of a digital mouse bearing a tumor and illuminated with three light sources (second simulation). The white circles in the plane view demarcate the actual fluorescence source.

Fig. 4.

Reconstructions of a digital mouse bearing a tumor and illuminated with two light sources in the presence of Gaussian noise at 5, 15, or 25% (third simulation). The white circles in the plane view demarcate the actual fluorescence source.

Fig. 5.

Reconstructions of living mice bearing single liver tumors, based on the indicated methods. In the plane view, red areas represent the reconstruction; blue areas, the ground truth; and purple areas, the intersection between the two.

Fig. 6.

AGLEN-based reconstruction of living mice bearing a single melanoma tumor, based on fluorescence signal from a probe that binds to PD-L1 on the tumor surface. Animals were treated only with probe (“No block”) or first with antibody against PD-L1, followed by probe (“Block”). Fluorescence images were mapped onto anatomical images based on computed tomography (CT) and magnetic resonance imaging (MRI). (a) Representative FMT reconstructions of PD-L1 expression. (b-c) Average fluorescence intensity (AFI) and maximum fluorescence intensity (MFI) of the reconstruction. Results were normalized to the MFI of the “No block” reconstruction. (d) Tissue sections from two mice after staining with hematoxylin-eosin (left panel) or antibody against PD-L1 (right panel).