Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms
3D quantitative elemental mapping using simultaneous proton induced X-ray emission tomography and scanning transmission ion microscopy tomography
Introduction
The combination of simultaneous on/off axis scanning transmission ion microscopy tomography (STIM-T) and proton induced X-ray emission tomography (PIXE-T) is a new method for micron-resolution quantitative 3D multi-elemental analysis. The equipment and software were installed and an experiment was performed to test the feasibility of this method in order to perform 3D micron-resolution analysis. Restrictions on the feasibility include time for analysis and limits of the accuracy of sample reconstruction especially as the sample is damaged during analysis.
Tomography has considerable advantages over 2D analysis. 2D analysis gives depth-averaged information while tomographic analysis allows the accurate localisation of trace elements in microstructures without the need for sectioning therefore potentially damaging and exposing sub-components to sources of contamination. Fine structures can be determined that can be easily masked by 2D analysis. Tomographic analysis does not only have applications to biomedical research but all fields where the composition of microstructures are important although this work is focused on biological samples.
STIM gives the ability of mapping samples by the measurement of proton energy loss passing through the sample. As the protons traverse the sample they lose energy, the amount of which is related to the electron density; with knowledge of the matrix (bulk) composition, the areal density can be determined. By measuring the energy loss and X-rays produced, trace element concentrations can be deduced. Rutherford scattering, where the energy and quantity of scattered protons are measured, can provide the bulk matrix composition, generally C, N and O for biological specimens which are difficult to detect by PIXE especially with the low efficiency at low energies of our detector.
To further reduce the time required STIM-T is performed simultaneously with PIXE-T by using the on/off-axis configuration [1]. On-axis protons (taking a straight path through the sample) are scattered through a thin foil and detected by an offset detector. This also allows PIXE-T and STIM-T data sets to be mapped accurately, thus density correlates with elemental masses. This is the first time tomography has been performed using this configuration. Switching between the low and high currents greatly increases the time to perform the experiment. The beam would require optimising and the sample removed during this process.
The scattering foil allows protons to be measured that have travelled in a straight path through the sample using a detector in an off-axis position. Off-axis STIM measures protons that have scattered from a straight path therefore are not representative of the areal density along the path. This may also affect the number of protons detected, which is required to normalise to the charge. On-axis measurements are therefore more accurate but cannot be used alongside PIXE-analysis since PIXE requires a high current resulting in a prohibitive STIM count rate and detector damage.
X-ray production can be much more accurately determined by locating where in the sample the X-rays are generated. The correlation of elements can be more localised, in particular for structures within the sample such as the cell nucleus, without interference from extra-cellular material and the membrane, for example. Delicate and inhomogeneous samples can be analysed without sectioning, which may introduce contamination or structural damage. This is a great advantage when biological samples are to be analysed.
While ion beam tomography has been developed over a period of over 20 years [2], the complexities in reconstruction, the computing power required, and microprobe technology has hindered advances. Tomography experiments are still rare and the problems encountered have limited tomography measurements of cells to only a couple of experiments. In 1999, Michelet and Moretto [3] demonstrated impressive PIXE-T of a cultured human ovarian adenocarcinoma cell. However, with a basic reconstruction method only a limited number of major elements were identified due to the relatively inefficient experimental set-up. STIM-T of cartilage has been performed [4] and the data reconstructed using an iterative tomographic reconstruction method developed in Melbourne by Sakellariou [5], the Discrete Image Space Reconstruction Algorithm (DISRA). One major factor affecting the accuracy of PIXE data is the charge (number of protons) per pixel. To solve this problem, here the PIXE data were normalised to the STIM counts since there is no way of measuring the charge per pixel directly,
Due to the ease of preparation, relatively high levels of trace elements it was decided to perform a tomography measurement of hair. The levels of trace elements in hair are affected by the preparation method and performing tomography on hair may help resolve some issues regarding hair analysis without risk of contamination. PIXE-T and STIM-T of a section from a strand of hair was performed.
Section snippets
Tomographic reconstruction
The tomographic reconstruction is complicated by the non-linear physics involved in PIXE and STIM. 2D analysis only gives depth-averaged information; the non-linear energy loss of protons and the non-homogeneous distribution of trace elements mean that it is not possible to accurately determine the relation between X-ray yield and elemental mass in inhomogeneous samples. Hence the accuracy of 2D analysis, especially with intermediate to thick samples where energy loss may significantly affect
Experimental set-up
The analysis took part at the University of Surrey Ion Beam Centre. Simon [6] has described the facilities in depth. The tomographic equipment was designed for use in the microbeam line although this could be transferred to a new nanobeam line currently under construction. The microbeam can produce a beam brightness (or current density) of 1 pA μm−2 mrad−2 MeV−1 for 2 MeV protons although this has been observed to be much higher [6].
A rotating flange is attached to the rear port which holds the
Analysis
A strand of untreated hair was inserted into a capillary that in turn was inserted into a hypodermic needle using superglue. The insertion was performed by hand under a microscope. This was in turn inserted into the shaft of the motor. The capillaries used were of approximately 100 μm internal diameter so that when the hair was inserted it rotated around the centre of mass as close as possible to reduce artefacts in the reconstruction. This worked reasonably well although several samples broke
Data processing and normalisation
Event files produced by OMDAQ [8] were converted into the GeoPIXE [9] format to be read into DISRA and collated and converted into STIM and PIXE sinograms. The 2D maps were smoothed and centred around the central energy loss. Pixels containing no data were visible due mainly to the computer being occupied thus not acquiring data. To remove these the average of pixels either side were used to fill in gaps and smooth ‘spikes’ in the maps.
We have no charge/pixel measuring capability, so PIXE
Results
Fig. 1 shows the results of the tomographic analysis. The elements S, Cl, K, Ca, Fe and Zn were reconstructed to produce 3D quantitative maps. The average density determined by STIM-T is approximately 0.8 g/cm3 although this varied per slice.
Twelve slices were analysed. A total volume of approximately 230,000 μm3 of hair weighing 0.2 μg was analysed. The total energy loss of the first projection, assuming an approximate homogeneous density and matrix composition, supports this.
A computer phantom
On/off-axis STIM and charge normalisation
The major problem with the use of gold foil was the contamination of gold and copper from it. Although the collimator was increased in length and the collimator moved backwards the foil was still visible by the Si(Li) detector. Moreover, protons scattered by the hair hit the aluminium collimator and these aluminium X-rays could be detected; the expected X-ray attenuation of aluminium was not very pronounced, suggesting the aluminium X-rays were not being produced in the hair but from the
Conclusions
The potential of this technique is high. With an increase in beam current density available on completion of a new nanobeam facility, this will result in sub-micron resolution tomography on cells.
References (9)
- et al.
Nucl. Instr. and Meth. B
(1999) - et al.
Nucl. Instr. and Meth. B
(2001) - et al.
Nucl. Instr. and Meth. B
(2001) - et al.
Nucl. Instr. and Meth. B
(2004)
Cited by (8)
A dual-PIXE tomography setup for reconstruction of Germanium in ICF target
2017, Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and AtomsCitation Excerpt :PIXE-T is one suitable method to measure elements’ quantitative spatial distribution, which can provide depth profile information. So far, most of PIXE-T focus on biological sample, like Caenorhabditis elegans [9,10], flagellates [11], human cancer cells [12] and hair [13]. The sensitivity of PIXE-T is the same with PIXE.
An implementation of the NiftyRec medical imaging library for PIXE-tomography reconstruction
2017, Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and AtomsCitation Excerpt :So, an error that may occur in the quantification or localization of one element for instance would have an influence on all other elements. Despite these drawbacks, DISRA has remained the most complete reconstruction code designed for STIMT and PIXET and for this reason, has been used and/or modified by several research groups [20–22]. To overcome the first two difficulties, a major improvement of the code was proposed (JPIXET) by replacing FBP by a method widely used for SPECT imaging, MLEM (Maximum Likelihood Expectation Maximization) [23,24].
Reconstruction of Ge spatial distribution in ICF target using PIXE-T
2016, Fusion Engineering and DesignCitation Excerpt :In order to improve counting statistics for X-ray emission, both high beam intensity of more than 100 pA (for proton) and long collecting time of several hours are needed. However, mass loss and change in morphology are reported in Refs. [10,15,17]. Based on projections taken during PIXE-T and scanning transmission ion microscopy tomography (STIM-T) experiments, the target doesn’t have morphological change (like shrink) under several hours’ exposure to proton beam.
A comparison of quantitative reconstruction techniques for PIXE-tomography analysis applied to biological samples
2014, Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and AtomsCitation Excerpt :At Surrey PIXE-T analysis was performed on a strand of hair. It was pre-irradiated to induce morphological changes and the mass loss estimated by extrapolating the change in the RBS count rate for each element [11], which demonstrated changes in the weight fractions of C, O and S. With STIM-T pre- and post-analysis, this can be improved upon significantly as the morphological changes are imaged more clearly. RBS is required to determine the matrix composition in order to accurately determine the density and for accurate simulations.
GPU-accelerated reconstruction methods for Proton Induced X-Ray Emission Tomography
2014, Radiation Physics and ChemistryFast simulation of Proton Induced X-Ray Emission Tomography using CUDA
2013, Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and AtomsCitation Excerpt :Combined with Scanning Transmission Ion Microscopy-Tomography (STIM-T) and Rutherford Backscattering Spectroscopy, it can produce quantitative 3D elemental maps. Biological samples have been analysed [2,3] although the success has been limited for trace elements due to the significant counts required to reconstruct accurate tomograms, which is also hindered by the sample sensitivity to beam radiation damage [4]. Multiple detectors can be used to increase the geometric efficiency.