Figure 2

Testing embedding materials and optimising projection number. (A–D) Standard comparison samples made from embedding a piece of plastic-insulated wire containing seven drawn copper pieces in air, agar, wax and epoxy. These were subjected to an ‘over-projected’ ~40-minute scan (24,001 projections, 80 ms exposure). Comparing the (A–D) first and (A^#,B^#,C^#,D^#) last (24,001^st) projection of each scan showed little deviation in air, wax and epoxy samples, confirming their stability during X-ray exposure. However, during agar scanning (B), expanding air bubbles (yellow arrows) shunted the wire piece. (A*,B*,C*,D*) Tomographic reconstructions were compromised in the (B*) agar-embedded sample by the sample being shifted by these expanding air bubbles. Difference maps between full-projection tomograms and tomograms reconstructed from 6,000 (A⁶⁰⁰⁰,B⁶⁰⁰⁰,C⁶⁰⁰⁰,D⁶⁰⁰⁰) 2,000 (A²⁰⁰⁰,B²⁰⁰⁰,C²⁰⁰⁰,D²⁰⁰⁰) and 500 (A⁵⁰⁰,B⁵⁰⁰,C⁵⁰⁰,D⁵⁰⁰) projection subsets. (E–H) 6,001-projection scans of a piece of unstained spinal cord embedded in each medium (propagation distance 20 mm) show that ‘air embedding’ does not adequately stabilise spinal cord samples. (I) X-ray transmission, assessed by comparing ten 100 × 100 pixel ‘background’ (embedding material only) regions in images of each embedding material to a flat-field image. Wax and epoxy absorbed ~30% of transmitted X-rays, with ~10% more absorbed by agar (mean ± standard error of the mean). (J) The peak signal to noise ratio (PSNR) measured in decibels [db] confirms that increasing projections reduces background noise, with ~85% saturation reached with 6,000 projections for wax. (K) The root mean square (RMS) contrast shows that increasing projections improves contrast, but this quickly saturates beyond ~2,000 projections.