Histological Hyperspectral Glioblastoma Dataset (HistologyHSI-GB)

Abstract

Hyperspectral (HS) imaging (HSI) technology combines the main features of two existing technologies: imaging and spectroscopy. This allows to analyse simultaneously the morphological and chemical attributes of the objects captured by a HS camera. In recent years, the use of HSI provides valuable insights into the interaction between light and biological tissues, and makes it possible to detect patterns, cells, or biomarkers, thus, being able to identify diseases. This work presents the HistologyHSI-GB dataset, which contains 469 HS images from 13 patients diagnosed with brain tumours, specifically glioblastoma. The slides were stained with haematoxylin and eosin (H&E) and captured using a microscope at 20× power magnification. Skilled histopathologists diagnosed the slides and provided image-level annotations. The dataset was acquired using custom HSI instrumentation, consisting of a microscope equipped with an HS camera covering the spectral range from 400 to 1000 nm.

Fig. 1

Graphical abstract of the methodology followed. (a) Resection procedure. (b) Macroscopic annotations of the GB locations. (c) HS data capture using a microscopic HS system. (d) ROI selection. (e) Dataset summary.

Fig. 2

Microscopic HS system. (a) HS camera. (b) Halogen light source. (c) Positioning joystick. (d) XY linear stage.

Fig. 3

Capture process to obtain focused HS cubes. (a) Example of a histology slide with tumour and non-tumour annotations. The yellow square identifies a ROI where the HS image was captured. (b) Synthetic RGB image where the Yλ frame employed to focus the sample is marked in yellow. Examples of (c) focused and (d) unfocused Yλ frames.

Fig. 4

Effect of calibration in the spectral signatures. (a) Grayscale image (generated by averaging each spectral band) and selecting pixels corresponding to different materials. (b) WR and DR spectral signatures. (c) Uncalibrated spectral signatures from the selected pixels. (d) Calibrated spectral signatures from the selected pixels. Colours in (c,d) correspond to selected pixels in (a).

Fig. 5

Examples of the uncalibrated and calibrated HS images. (a,b) grayscale representation generated by averaging all spectral bands of the uncalibrated and calibrated HS images, respectively. (c,d) synthetic RGB image of the uncalibrated and calibrated HS images, respectively, generated using a model of human eye spectral response.

Fig. 6

Human eye spectral response to light where different colour line represents the normal probability distribution function modelling each channel.

Fig. 7

Examples of HS images from the HistologyHSI-GB dataset showing the synthetic RGB images and different spectral bands after calibration for tumour and non-tumour samples.

Fig. 8

Graphical representation of the HistologyHSI-GB dataset structure.

Fig. 9

Spectral and spatial calibration targets. (a) Certified WCT-2065 polymer. (b) 0.01 mm microscope slide reticule.

Fig. 10

SNR of the microscopic HS system: (a) over the spectral range for the central pixel of the push-broom frame and (b) its spatial distribution for different wavelengths (blue: 403 nm, orange: 448 nm, yellow: 655 nm, purple: 894 nm and green: 997 nm).

Fig. 11

Spectral characterization of the microscopic HS system. (a) Manufactured certified spectral signature of the WCT-2065 polymer (black line) and spectral signature captured by the microscopic HS system (red line). (b) Pearson Correlation Coefficient between WCT-2065 and the measured HS peak values.

Fig. 12

Pixel size validation using a micrometre ruler. (a) Profile of Yλ frame extracted from the Yλ frame and (b) its first derivative where red crosses are local minima peaks and green crosses local maxima peaks.