Fig. 1. Head-mounted miniature two-photon microscope design. (a) Schematic design and assembly of FHIRM-TPM^[48], (i) core optical components and mechanical structure, (ii) photographs of a FHIRM-TPM device (compared with a fingertip) and mounted on a mouse head; (b) schematic design and assembly of FHIRM-TPM 2.0^[89], (i) schematic diagram of FHIRM-TPM 2.0 head-mounted assembly with an enlarged view of the Z-axis scanning module, (ii) photograph of a mouse wearing the FHIRM-TPM 2.0 head-mounted assembly, (iii) 2000-frame maximum-intensity projection of GCaMP6s-expressing mPFC neurons in cortical layer 2; (c) core features of a MINI2P microscope^[99], (i) schematic of key MINI2P components, (ii) μTlens quartet comprising four stacked piezo-membrane lenses, (iii) axial scanning range of MINI2P with quartet μTlens, (iv) tapered fiber bundle structure, (v) custom objectives for three distinct imaging scenarios

Fig. 2. Spatial-bandwidth product expansion methods for a fiber-scanning endomicroscope. (a) Composite cantilever design based on conventional DCF^[107], (i) schematic of the composite cantilever paired with a micro-objective, (ii) illustration of the evolution of the beam numerical aperture, from the DCF end-face (via a spacer fiber), forming a new focus near the exit face of the gradient-index rod lens, and finally reaching the focal spot in the sample (water-immersion environment), (iii) photograph of the distal section of the fabricated composite cantilever; (b) Kagome-lattice hollow-core fiber fused with a microsphere lens to form a sub-micron focus^[52], (i) scanning electron microscope image of a 30 μm-diameter silica microsphere fused to and sealing the hollow core, (ii) close-up view of the microsphere lens fused to the hollow core, (iii) without the microsphere lens, the hollow-core fiber has a mode-field diameter of 15 μm, (iv) the microsphere acts as a ball lens, focusing the exiting light into a spot approximately 1 μm in diameter

Fig. 3. Label-free in vivo metabolic imaging using two-photon NADH fluorescence lifetime endomicroscopy^[123]. (a) Metabolic dynamics of cultured A431 cancer cells during apoptosis induction, including square-root-scaled NADH intensity images (top), color-coded images of the free-NADH fraction αfree (middle), and intensity-weighted mean lifetime τavg images (bottom); (b) color-coded αfree images of subcutaneous tumor graft under STS treatment, including 4 control fields and 24 post-STS-injection fields (each column corresponds to one imaging site, from #1 the farthest to #6 the closest to the injection site), with the post-injection acquisition time indicated at the upper left corner of each image; (c) NADH intensity grayscale images (top) and color-coded images of the free-to-bound NADH ratio (Cfree/Cbound) (bottom) acquired at different time points in a mouse kidney ischemia-reperfusion model (p.i. denotes post-ischemia and p.r. denotes post-reperfusion)

Fig. 4. SHG endomicroscopy of cervical collagen fibers. (a)‒(e) SHG endomicroscopy imaging of mouse cervical tissue sections from non-pregnant (NP) and pregnant mice at gestation days 6, 12, 15, and 18, showing remarkable changes in cervical collagen morphology during pregnancy^[26]; (f)‒(j) trichrome-stained microscopic images of cervical tissue sections at the same gestational time points; (k) typical SHG endomicroscopy images of cervical tissue sections from normal pregnant mice at gestation day 15 (upper row) and RU486-induced preterm birth mouse models (lower row)^[133]; (l) representative SHG endomicroscopy images of sub-epithelial cervical collagen acquired directly from intact mouse cervices of normal pregnant mice at gestation day 15 (upper row) and RU486-treated preterm mouse models (lower row)^[133]

Fig. 5. MINI2P recording results in mouse VC and MEC^[99]. (a) Using FOV stitching, MINI2P cumulatively recorded more than 10,000 neurons across sequential sessions in VC, (i) the 5×5 FOV array (yellow squares) overlaid on the retinotopic map within a 4.6 mm cranial window, color indicates the sine of the angle between vertical (V) and horizontal (H) retinotopic gradients (red: +1; blue: -1), (ii) averaged image covering a 2.2 mm×2.2 mm FOV in imaging plane 2, (iii) two-dimensional projection of neurons at different depths (magenta: -100 μm; blue: -140 μm) within the imaging plane; (b) distribution of grid cells in MEC and adjacent parasubiculum (PAS), color-coded for imaging-plane depth, dashed line denotes the MEC-PAS border; (c) examples of neighboring grid cells with similar grid spacing and orientation but mixed phases, (i) magnified view of three adjacent grid cells, (ii) low correlation of calcium traces indicates minimal cross-talk, (iii) for each individual cell separately (first three columns) and overlaid (rightmost column): calcium events superimposed on mouse trajectory (top), spatial tuning maps (middle), and their autocorrelation analyses (bottom)