Focused Ion Beam Ablation Tomography
A Novel 3D Imaging Method for Biological Samples
- Fig. 1: Human Osteoblast cells showing differential susceptibility with successive sweeps of the ion beam.
- Fig. 2: Scheme of the FIBAT technique, including image selection, ablation vector calculation, image reconstruction and volume rendering.
- Fig. 3: Ablation vector acquisitions. (A) Pixel values squared to enhance contrast. (B) Smoothed with median filter of radius 4. (C) Automatic detection of moving edges (highlighted with red lines). (D) Ablation parameters obtained from the first two images in A, i.e. distance of downward movement of each edge pixels (colored in green) from the first image to the second.
- Fig. 4: A selection of images through an entire cyanobacteria. The columns can be clearly seen, each is tracked and analyzed using the ImagJ plugin.
- Fig. 5: A series of images of T3T cells with the corresponding FIB Susceptibility plot. Red shows high sesceptibility, blue shows more resistive material.
We explore the potential of a novel imaging method to exploit differential susceptibility of biological tissue to erosion by Focused Ion Beam (FIB) milling. The contents of a given cell (nucleus, Golgi apparatus, vacuoles) have different physical compositions and will ablate differentially under the ion beam. Sequential fine slices of frozen cells have been ablated, after which slice-by-slice image analysis of the extracted ablation vectors is back-converted to a FIB susceptibility parameter.Cryogenic-FIB-SEM
In a scanning electron microscope only surface features of a sample are visible. To gain information on features below the surface in regions of interest it is possible to use a beam composed of accelerated gallium ions, a Focused Ion Beam (FIB), to strip away layers of material. If performed at very low temperatures (-150°C) the technique is known as cryogenic-FIB-SEM (Cryo-FIB-SEM) and can be applied to a range of ‘high water content' samples, including cells and tissues. During sample-milling, columns of material often remain, an artifact referred to as curtaining , which are often considered undesirable as they detract from clear images. Curtaining in FIB preparation arises from: (i) re-deposition of ablated material (ii) the difference in resistance to the path of the ions as they pass through the sample (iii) where the sample has a non-uniform surface structure, e.g. holes. Approaches to minimize the phenomenon have previously been the subject of papers [2, 3], including the deposition of organometallic precursors. Beyond the issue of curtaining, the imaging of biological samples under cryogenic conditions has additional problems that make it challenging. Assuming that curtaining can be successfully overcome to yield a flat planed surface, the contrast between the components of the material (i.e. cells) is low, particularly in the case of non-field emission source instruments or the latest detectors. Even with the use of backscatter detectors the contrast between C, N, P or other ‘low' atomic number elements can be poor, especially without the possibility of adding EM stains.
One route to enhance contrast is to thermally etch the surface of the sample, by raising the sample temperature to promote water sublimation at the cut face.
This step can be powerful in the case of individual sites of interest, however, is impractical after each cut, if the aim of the experiment is 3D tomography. Having noticed that some areas of biological cells were removed more rapidly by the beam than others (fig. 1) our idea was to use this difference in susceptibility, often perceived as a weakness of Cryo-FIB-SEM, to obtain information about the internal structure. Through use of the differential milling approach, we hope to offer a novel way to probe the contents of a sample without the need for repeated thermal etching or to work at the highest resolution or sensitivity.
Ablation Vector Extraction
By imaging the cut face of the sample between each sweep of the ion beam features created by the differential ablation rate within the sample were plotted and the material removed per sweep calculated as an ablation vector (fig. 2). By pooling these values for a 3D data set it was hoped that internal structures of differing susceptibility to the ion beam would be revealed and could be related to known structures in the cells. Several cell types were successfully frozen for cryo-SEM (cardiac, fibroblast, retinal epithelial cells). Protocols were adapted to include platinum shadowing and tungsten deposition to enhance grounding and minimize artifacts. To aid with image processing, a software analysis plugin was developed for ImageJ, an open-source image analysis package, to extract and reconstruct ablation susceptibility values as 3D datasets.
This process involves image selection, ablation vector extraction, image slice reconstruction and volume rendering (fig. 2). Ablation vectors record the quantity of ablation caused by FIB milling, which is the distance of edge movements in consecutive images. In a digital image, an edge consists of a sequence of continuous pixels. Therefore, each element in the vector corresponds to the distance moved by an edge pixel from one image to the next, the distance of which is measured in pixels. First, images are aligned (fig. 3A), each pixel value is squared to enhance contrast (fig. 3B), then a median filter is applied to remove noise in the images (fig. 3C), from which the advancing edge in the image needs to be detected. As FIB mills the current slice the position of the cut face is plotted against the "cliff" formed by the next slice. Following the progression along each vertical column of pixels then subtracting the y-coordinates of the same edge pixel in two consecutive images produces the ablation susceptibility parameter (fig. 3D).
Cultured cells reacted heterogeneously with delicate features appearing obvious to the human eye during milling. However, these proved difficult subjects for automated image analysis as the algorithms were designed based on certain assumptions. We found it difficult in early work to optimize an automated algorithm that works for all sequences, therefore a supervised semi-automated approach was chosen where the milled features were delineated manually using a touch-pad before an automatic algorithm extracted susceptibility vectors and reconstructed the data set in 3D. By calculating an erosion rate along each column of the samples segmentation between "hard" and "soft" areas was thus obtained by differential susceptibility between these regions. The method was applied to other samples of various material properties including Diatoms. These creatures proved to be excellent imaging subjects as they are composed of biological material which incorporates a hard silicate structure with a soft cell inside. We were able to see a difference in the ion beam susceptibility and to reconstruct details of the Diatom bodies including internal anatomy. Gradually stripping away layers of the sample away allowed us to build a picture in terms of its resistance to the ion beam (fig. 4), when the algorithm was applied to 3T3 cells the result was a color coded susceptibility plot (fig. 5).
This investigation explored the potential of a novel tomographic imaging method to reveal 3D information from biological samples. The potential of Focused Ion Beam Ablation Microscopy (FIBAT) has been demonstrated we believe that the FIBAT method, used in combination with confocal microscopy or other 3D imaging modality, can be used to realize the 3D tomography of cells and supply a novel contrast mechanism which may provide information on internal nano-structures within samples. A number of technical considerations have been identified and work is ongoing to improve the technique. Further development of the methodology should allow this novel contrast mechanism to be exploited in cryo-SEM imaging of a variety of samples.
 Drobne D. et al.: Microscopy Research and Technique 70, 895 (2007)
 Stokes D. Hayles M.: SPIE Conference Proceedings 7378G1 (2009)
 Hayles M. et al.: Journal of Microscopy 226, 263 (2007)
 The authors gratefully acknowledge funding from the Bridging the Gaps scheme at the University of Nottingham and their home institutions.
Dr. Christopher DJ Parmenter, FRMS (Corresponding author via e-mail request)
Dr. Kevin Webb
The Nottingham Nanotechnology and Nanoscience Centre
University of Nottingham