Label-Free Macroscopic Fluorescence Lifetime Imaging of Brain Tumors

Abstract

Advanced stage glioma is the most aggressive form of malignant brain tumors with a short survival time. Real-time pathology assisted, or image guided surgical procedures that eliminate tumors promise to improve the clinical outcome and prolong the lives of patients. Our work is focused on the development of a rapid and sensitive assay for intraoperative diagnostics of glioma and identification of optical markers essential for differentiation between tumors and healthy brain tissues. We utilized fluorescence lifetime imaging (FLIM) of endogenous fluorophores related to metabolism of the glioma from freshly excised brains tissues. Macroscopic time-resolved fluorescence images of three intracranial animal glioma models and surgical samples of patients' glioblastoma together with the white matter have been collected. Several established and new algorithms were applied to identify the imaging markers of the tumors. We found that fluorescence lifetime parameters characteristic of the glioma provided background for differentiation between the tumors and intact brain tissues. All three rat tumor models demonstrated substantial differences between the malignant and normal tissue. Similarly, tumors from patients demonstrated statistically significant differences from the peritumoral white matter without infiltration. While the data and the analysis presented in this paper are preliminary and further investigation with a larger number of samples is required, the proposed approach based on the macroscopic FLIM has a high potential for diagnostics of glioma and evaluation of the surgical margins of gliomas.

Keywords: FLIM; autofluorescence; fluorescence lifetime imaging; glioblastoma; image processing; rat glioma model.

Figure 1

Histopathology of glioma models glioma C6, glioblastoma 101.8 and anaplastic astrocytoma 10-17-2 and normal rat brain. H&E-staining. (A) Initial magnification X7. (B) Initial magnification X40. Enlarged regions are indicated by the yellow squares on the lower-magnification panel. Bars are applicable to all images in the row.

Figure 2

Macro-FLIM of glioma models and normal rat brain. (A) Representative autofluorescence time-resolved images of glioblastoma C6, glioblastoma 101.8, anaplastic astrocytoma 10-17-2 and rat brain without tumor. Pseudo RGB image is built from three time channels where the red color reflects fast emitting species, green medium, and blue long lived components; PCA images in the pseudo RGB images: red, green and blue colors correspond to a first, second, and third principal component; LU images reflect three selected classes: grey matter, white matter, and the background. The tumors are marked with a white line. Scale bar: 10 mm, applicable to all images. (B) Quantification of τ_m and a₁/a₂ ratio in (1) glioblastoma C6 (2), glioblastoma 101.8 (3), anaplastic astrocytoma 10-17-2 and normal brain. Scatter dot plot displays the measurements for individual animals (dots) and the median, minimum and maximum (horizontal lines). τ_m is the mean fluorescence lifetime. a₁/a₂ is the ratio of relative contributions of short and long components.

Figure 3

Macro-FLIM of patients’ glioblastoma and the peritumoral white matter with and without infiltration by tumor cells. (A) Representative autofluorescence time-resolved images. Scale bar: 3 mm. (B) Histopathology of samples shown in (A). H&E-staining. Scale bar: 100 µm. (C) Quantification of τ_m and a₁/a₂ ratio in tumors and the peritumoral white matter. Scatter dot plot displays the measurements for individual samples (dots) and the median, minimum and maximum (horizontal lines). The values for the images in (A) are marked in red. τ_m is the mean fluorescence lifetime. a₁/a₂ is the ratio of relative contributions of short and long components.

Figure 4

Image processing of 3D datasets from FLIM imaging of a peritumoral white matter with infiltration (TDWM), tumor and a white matter without tumor infiltration. (A) Three band pseudo RGB of the excised samples. (B) The image was processed Linear Unmixing with no supervision. Linear Unmixing of the datasets demonstrate a significant visual difference between the infiltrated white mater and the white matter. Scale bar: 3 mm.