Transportable hyperspectral imaging setup based on fast, high-density spectral scanning for in situ quantitative biochemical mapping of fresh tissue biopsies

Abstract

Significance: Histopathological examination of surgical biopsies, such as in glioma and glioblastoma resection, is hindered in current clinical practice by the long time required for the laboratory analysis and pathological screening, typically taking several days or even weeks to be completed.

Aim: We propose here a transportable, high-density, spectral scanning-based hyperspectral imaging (HSI) setup, named HyperProbe1, that can provide in situ, fast biochemical analysis, and mapping of fresh surgical tissue samples, right after excision, and without the need for fixing, staining nor compromising the integrity of the tissue properties.

Approach: HyperProbe1 is based on spectral scanning via supercontinuum laser illumination filtered with acousto-optic tunable filters. Such methodology allows the user to select any number and type of wavelength bands in the visible and near-infrared range between 510 and 900 nm (up to a maximum of 79) and to reconstruct 3D hypercubes composed of high-resolution (4 to $5\mu m$ ), widefield images ( $0.9\times 0.9{\mathrm{mm}}^2$ ) of the surgical samples, where each pixel is associated with a complete spectrum.

Results: The HyperProbe1 setup is here presented and characterized. The system is applied to 11 fresh surgical biopsies of glioma from routine patients, including different grades of tumor classification. Quantitative analysis of the composition of the tissue is performed via fast spectral unmixing to reconstruct the mapping of major biomarkers, such as oxy-( ${\mathrm{HbO}}_2$ ) and deoxyhemoglobin (HHb), as well as cytochrome-c-oxidase (CCO). We also provided a preliminary attempt to infer tumor classification based on differences in composition in the samples, suggesting the possibility of using lipid content and differential CCO concentrations to distinguish between lower and higher-grade gliomas.

Conclusions: A proof of concept of the performances of HyperProbe1 for quantitative, biochemical mapping of surgical biopsies is demonstrated, paving the way for improving current post-surgical, histopathological practice via non-destructive, in situ streamlined screening of fresh tissue samples in a matter of minutes after excision.

Keywords: biomedical optics; biophotonics; digital histopathology; hyperspectral imaging; neurosurgery.

Fig. 1

(a) Schematics of the HyperProbe1 with all its components. (b) Picture of the spectral illumination side of HyperProbe1, including the SCL source, the AOTF, and the controlling devices. (c) Picture of the imaging and detection side of HyperProbe1, depicting the illumination output, the amplitude stabilizers, the speckle reducer, and the camera. (d) Result of the imaging tests on the USAF1951 target with HyperProbe1 at 650 nm. (e) The smallest resolved line pair (group 6, element 6) is highlighted in blue and its corresponding line profile that shows minimum FWHM separation ($4.38\mu \mathrm{m}$). (f) Example of a sample of excised glioma tissue obtained from the surgical biopsies.

Fig. 2

(a) Absorption coefficients for ${\mathrm{HbO}}_2$, HHb, lipids and water, and scattering coefficient of generic brain tissue in the visible and NIR range ($150\mathrm{g}/\mathrm{L}$ concentration of Hb in the blood and blood volume content in generic brain tissue equal to 5% are assumed).^, (b) Molar extinction coefficients of ${\mathrm{HbO}}_2$, HHb, oxCCO, and redCCO in the visible and NIR range.^,

Fig. 3

(a) Simulated fluence rates within the 3D cerebral biopsy model at different wavelengths (the plots are in a logarithmic scale and the absorbing layer was excluded from the plots to maintain a visible contrast). The distribution of the fluence showed a noticeably higher penetration of the light at the NIR wavelengths. (b) Simulated mean photon pathlength within the 3D biopsy model, as a function of wavelength. (c) Partial pathlength distributions of the photons simulated within the 3D biopsy model at various wavelengths (as previously, the absorbing layer was not included in this figure due to a negligible number of photons passing through it).

Fig. 4

(a) Intercomparison of averaged reflectance spectra over the entire imaged FOVs of each biopsy sample. (b) Comparison between average reflectance spectra grouping LGG (WHO 2, 3) against HGG samples (WHO 4), with resulting $p$-values from statistical analysis on differences between each wavelength. (c) Processed spectral image from HyperProbe1, at 560 nm, of HGG (WHO grade 4) biopsy sample S2, with highlighted, selected ROIs in the FOV in which average reflectance spectra were calculated. (d) Example of average reflectance spectra in the corresponding ROIs of the biopsy sample S2. (e) Example of average attenuation spectra in different ROIs of biopsy sample S2, with the portion within the NIR range 780 to 900 nm enlarged. (f) Example of the second derivative of the average attenuation spectra in different ROIs of biopsy sample S2.

Fig. 5

Inferred HbT and diffCCO concentration maps of HGG S4 FOV#1 (a) and LGG grade S10 (b) samples fitting the whole measured wavelength spectrum (510 to 900 nm), and a model fitting of the observed attenuation for the marked pixel in the HbT image is shown in the third column for both biopsies. (c) Histogram showing probability density distribution of inferred lipid volumetric content of each pixel across different LGG (displayed in green) and HGG (displayed in red) grade samples. (d) Distribution means of the lipid content (reported on $x$-axis) and diffCCO concentrations (reported on $y$-axis) suggest that lipid mean content could be able to distinguish the grading of all samples, whereas no apparent separation is visible for the inferred mean diffCCO concentrations.

Fig. 6

Inferred HbT and diffCCO concentration maps of HGG S4 FOV#1 (a) and LGG S10 (b) samples fitting exclusively the NIR range (740 to 900 nm), and a model fitting of the observed attenuation for the marked pixel in the HbT image is shown in the third column for both biopsies. (c) Histogram showing probability density distribution of inferred diffCCO concentrations of each pixel across different LGG (displayed in green) and HGG (displayed in red) grade samples. (d) Distribution means of diffCCO and ${\mathrm{HbO}}_2$ concentrations suggest that these parameters could be able together to distinguish lower and higher grade samples of glioma tissue, with diffCCO being the most accurate, across all the investigated samples.

Fig. 7

Histogram showing probability density distribution of inferred oxCCO (a) and redCCO (b) concentrations of each pixel across different LGG (displayed in green) and HGG (displayed in red) grade samples. Neither the inferred oxCCO nor redCCO density distributions using the NIR range (740 to 900 nm) suggest being able to differentiate tumor grading, contrary to the diffCCO mean concentrations.