Fig. 1

Agreement of the proposed 3D framework with traditional, clinical-grade 2D staining techniques. (A) Procedural workflow for conventional 2D analysis (upper row) and 3D imaging of immunofluorescence-stained, optically cleared breast cancer specimens (lower row). After tissue embedding in 3% agarose, a vibratome was used to obtain 200-µm-thick slices. Paired slices (n = 2) from the same patient were subjected to (1) paraffin embedding followed by traditional 2D hematoxylin and eosin (H&E) staining and IHC (n = 1) and (2) 3D immunofluorescence staining (n = 1). PD-L1 is labeled in green color in immunofluorescence images, whereas nuclei and cell membranes were counterstained with SYTO-16 (blue color) and DiD (red color), respectively. (B) Illustrative example of immunofluorescence images obtained from case #7 (percentage of tumor-infiltration immune cells = 10%): 3D reconstruction (upper row) and ortho-slice visualization (lower row, left side) showing the region of interest (lower row, right side); scale bar = 2000 μm (upper row and lower row, left side), scale bar = 200 μm (lower row, right side). (C) Comparison of real H&E (upper row, left side) and real IHC (upper row, right side) images versus pseudo H&E (lower row, left side) and pseudo IHC (lower row, right side) images obtained from the conversion of immunofluorescence images; scale bar = 200 μm. (D) Cross-sectional images obtained from case #2 (percentage of tumor-infiltration immune cells = 90%) were characterized by a uniform immunofluorescence staining in the 3D space; scale bar = 1000 μm. (left), scale bar = 200 μm (right). (E) Uniform PD-L1 labeling in the top and bottom layers of the sample obtained from case #2; scale bar = 200 μm