Quantitative Backscattered Electron Imaging of Bone Using a Thermionic or a Field Emission Electron Source

Abstract

Quantitative backscattered electron imaging is an established method to map mineral content distributions in bone and to determine the bone mineralization density distribution (BMDD). The method we applied was initially validated for a scanning electron microscope (SEM) equipped with a tungsten hairpin cathode (thermionic electron emission) under strongly defined settings of SEM parameters. For several reasons, it would be interesting to migrate the technique to a SEM with a field emission electron source (FE-SEM), which, however, would require to work with different SEM parameter settings as have been validated for DSM 962. The FE-SEM has a much better spatial resolution based on an electron source size in the order of several 100 nanometers, corresponding to an about [Formula: see text] to [Formula: see text] times smaller source area compared to thermionic sources. In the present work, we compare BMDD between these two types of instruments in order to further validate the methodology. We show that a transition to higher pixel resolution (1.76, 0.88, and 0.57 μm) results in shifts of the BMDD peak and BMDD width to higher values. Further the inter-device reproducibility of the mean calcium content shows a difference of up to 1 wt% Ca, while the technical variance of each device can be reduced to [Formula: see text] wt% Ca. Bearing in mind that shifts in calcium levels due to diseases, e.g., high turnover osteoporosis, are often in the range of 1 wt% Ca, both the bone samples of the patients as well as the control samples have to be measured on the same SEM device. Therefore, we also constructed new reference BMDD curves for adults to be used for FE-SEM data comparison.

Keywords: Adult human bone; Bone mineralization density distribution; Quantitative backscattered electron imaging.

Fig. 1

Representative images of the two types of cathodes discussed in the text. a Light microscopy image of a typical tungsten hairpin cathode used in the DSM 962. The red ($r=55\mu \text{m}$) and yellow ($r=136\mu \text{m}$) circles give an estimate of the source radius. b SEM image of the tip of a field emission cathode used in the SEM SUPRA 40 device. The green circle with $r=512\text{nm}$ indicates the tip radius

Fig. 2

Calibration procedure for quantitative measurements. a The main standard that is used for calibration. The light area corresponds to aluminum, the dark ones to carbon. b Corresponding gray-level histogram. Brightness and contrast of the detector are adjusted that the gray levels corresponding to carbon and aluminum are centered at 25 and 225, respectively

Fig. 3

GLs measured for 5 different minerals: Kernite (${\text{Na}}_2[{\text{B}}_4{\text{O}}_6({\text{OH}}_2)]·3{\text{H}}_2\text{O}$), Magnesite (${\text{MgCO}}_3$), Magnesium Fluoride (${\text{MgF}}_2$), Periclase (MgO), and Dolomite ($\text{CaMg}{[{\text{CO}}_3]}_2$) with two SEMs: a Zeiss DSM 962 equipped with a tungsten cathode and a Zeiss field emission SEM SUPRA 40. The mean Z number of the minerals was obtained by a compositional EDX analysis of the samples. GLs were measured with one nominal magnification of $\times$ 50 for the DSM962 and for three different nominal magnifications (× 65, × 130 and × 200) for the SEM SUPRA 40. The solid gray line denotes the calibration line: brightness and contrast of the detector are set such that pure carbon (Z = 6) and aluminum (Z = 13) have a gray level of 25 and 225, respectively. The thickness of the line corresponds to a $\pm 2\text{GL}$ uncertainty due to instrument calibration and carbon coating

Fig. 4

Repeated measurement of one sample on the same device. a BMDD from an iliac crest biopsy sample measured with the DSM 962 in the year 2012 and 2017, respectively. b The same sample measured on the SEM SUPRA 40 in 2018 and 2020, respectively. Additionally, CaPeak, CaMean, and CaWidth are given for the two curves (in wt% Ca as well as in GLs). The measured parameters do not differ more than 2 GLs

Fig. 5

Typical images obtained from measurements of the same ROI with different nominal magnifications and corresponding BMDD curves. a × 65 with pixel resolution 1.76 μm, b × 130 with pixel resolution 0.88 μm, c × 200 with pixel resolution 0.57 μm. All images show the same area of $587\times 440\mu {\text{m}}^2$ and have be obtained with the SEM SUPRA 40. d Corresponding BMDDs

Fig. 6

Images and corresponding BMDD curves of the same surface (covering a region of $500\times 500\mu {\text{m}}^2$) measured with the same resolution (1.76 μm/pixel) with both devices. a Image obtained with the DSM 962 and b with the SEM SUPRA 40. The same features (bone packet size and shape) can be observed in both images shown. c The BMDD curves show that there is a significant shift to higher mineralizations and broadening of the curve for the SEM SUPRA 40 compared to the DSM 962

Fig. 7

Adult trabecular and cortical references and corresponding BMDD parameters obtained from the SEM SUPRA 40. a Adult trabecular and cortical reference BMDDs. The insets show the single BMDDs used for averaging to obtain the corresponding reference BMDD. b The obtained BMDD parameters for the presented single BMDD curves are shown as mean ± SD and median [25%; 75%], respectively. Median and quartiles are obtained using the python numpy library