High-contrast differentiation resolution 3D imaging of rodent brain by X-ray computed microtomography

Journal of Instrumentation OPEN ACCESS High-contrast differentiation resolution 3D imaging of rodent brain by X-ray computed microtomography To cite this article: T. Zikmund et al 2018 JINST 13 C02039 View the article online for updates and enhancements. Related content Coronal in vivo forward-imaging of rat brain morphology with an ultra-small optical coherence tomography fiber probe Yijing Xie, Tim Bonin, Susanne Löffler et al. - S-values for a tomographic brain model Tsi-chian Chao and X George Xu - Three-dimensional digital mouse brain atlas David Purger, Todd McNutt, Pragathi Achanta et al. - This content was downloaded from IP address 78.45.128.252 on 03/03/2018 at 10:51 2018 JINST 13 C02039 Published by IOP Publishing for Sissa Medialab Received:December 5, 2017 Revised:January 31, 2018 Accepted:January 31, 2018 Published:February 22, 2018 International workshop on Imaging II 4–8 September 2017 Varenna, Italy High-contrast differentiation resolution 3D imaging of rodent brain by X-ray computed microtomography T. Zikmund,aM. Novotná,aM. Kavková,aM. Tesařová,aM. Kaucká,b,cB. Szarowská,c I. Adameyko,b,cE. Hrubá,dM. Buchtová,dE. Dražanová,eZ. Starčukeand J. Kaisera,1 aCentral European Institute of Technology, Brno University of Technology, CEITEC - BUT, Purkyňova 123, 612 00 Brno, Czech Republic bDepartment of Physiology and Pharmacology, Karolinska Institutet, Nanna Svartz väg 2, Stockholm 17177, Sweden cDepartment of Molecular Neurosciences, Center for Brain Research, Medical University of Vienna, Spitalgasse 4, Vienna, Austria 1090 dLaboratory of Molecular Morphogenesis, Institute of Animal Physiology and Genetics AS CR, v. v. i., Brno, Czech Republic eInstitute of Scientific Instruments of the CAS, v. v. i., Brno, Czech Republic E-mail: [email protected] Abstract: The biomedically focused brain research is largely performed on laboratory mice considering a high homology between the human and mouse genomes. A brain has an intricate and highly complex geometrical structure that is hard to display and analyse using only 2D methods. Applying some fast and efficient methods of brain visualization in 3D will be crucial for the neurobiology in the future. A post-mortem analysis of experimental animals’ brains usually involves techniques such as magnetic resonance and computed tomography. These techniques are employed to visualize abnormalities in the brains’ morphology or reparation processes. The X-ray computed microtomography (micro CT) plays an important role in the 3D imaging of internal structures of a large variety of soft and hard tissues. This non-destructive technique is applied in biological studies because the lab-based CT devices enable to obtain a several-micrometer resolution. However, this technique is always used along with some visualization methods, which are based on the tissue staining and thus differentiate soft tissues in biological samples. Here, a modified chemical contrasting protocol of tissues for a micro CT usage is introduced as the best tool for ex vivo 3D imaging of a post-mortem mouse brain. This way, the micro CT provides a high spatial resolution 1Corresponding author. c 2018 The Author(s). Published by IOP Publishing Ltd on behalf of Sissa Medialab. Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. https://doi.org/10.1088/1748-0221/13/02/C02039 2018 JINST 13 C02039 of the brain microscopic anatomy together with a high tissue differentiation contrast enabling to identify more anatomical details in the brain. As the micro CT allows a consequent reconstruction of the brain structures into a coherent 3D model, some small morphological changes can be given into context of their mutual spatial relationships. Keywords: Computerized Tomography (CT) and Computed Radiography (CR); MRI (whole body, cardiovascular, breast, others), MR-angiography (MRA) 2018 JINST 13 C02039 Contents 1 Introduction 1 2 Materials and methods 2 2.1 Sample preparation for micro CT measurement 2 2.2 Micro CT measurement and data processing 2 2.3 MRI measurement 3 3 Results and discussion 3 4 Conclusion 9 1 Introduction Computed tomography is widely applied in the human medicine especially to monitor bone injuries, to diagnose head, lung and chest conditions or to detect cancer. In the last decades, this nondestructive imaging technique has been evolved and therefore is able to provide high-resolution ex vivo analyses of biological samples. The micro CT brings unique possibilities concerning the imaging of complex biological systems, such as the following cases: a description of a mouse knee cartilage development [1]; a formation of 3D models of mouse chondrocranium, an analysis of the ossification onset and morphological changes in various transgenic mice [2–4]; a noninvasive observation of a human embryo [5]. The principle of the X-ray micro CT imaging is based on taking series of 2D projection radiographs from different angles and their subsequent processing. In each projection, the information about radiographic density is recorded. A mathematical process called tomographic reconstruction forms a 3D matrix which represents a map of volume density. The spatial resolution of the system is determined by the geometry of a cone beam. The achievable resolution is down to 1 µm for lab-based machines. This method has proven to be an effective tool for imaging of native bone tissues [6,7], since the hydroxyapatite is dense enough to be easily detected. In the case of biological samples and soft tissues ex vivo, the micro CT imaging requires an application of some X-ray absorbing contrast agents, e. g. phosphotungstic acid (PTA), iodine or osmium [8]. If the contrast is sufficient, it is possible to segment different structures within the 3D tomographic data by using an appropriate software. The major advantage of the CT method is the ability to image a bone, soft tissues and blood vessels at the same time. In some cases, it is advantageous to combine the CT method with some other techniques for in vivo imaging to visualize better various soft tissues. Especially in the case of cancer diagnosis, the computed tomography is combined with the positron emission tomography (PET/CT) [9]. –1– 2018 JINST 13 C02039 In biomedical research, the rodent brain attracts a lot of attention because of some human-related pathologies that can be modelled in transgenic animals and also because of the high homology between the human and mouse genome, which causes numerous fundamental neurobiological questions [10]. To take advantage of the information gained from mouse brain research, it is necessary to systematically collect the phenotype information at all biological levels. Conventionally used 2D analysis methods — histology, immunochemistry or transmission electron microscopy — show a high resolution within the plane in which the tissue is sectioned. However, these methods are destructive and two dimensional in their nature. In the case of post-mortem 3D imaging of mouse brain, the micro CT is not the only method that could be applied. The magnetic resonance imaging (MRI) could be used for a similar purpose as well. In the past, the micro CT was used as an accessory technique to MRI for the imaging of the mouse brain. The micro CT was employed for the imaging of the skull [6,7] or for the visualization of vascular system of mouse brain filled with radio-opaque silicone rubber Microfil [11–17]. With the recent progress in development of chemical contrasting protocols for brain tissue [6,18–22] or application of micro CT imaging as a tool for location of cerebral ischemia [22,23] or brain tumors [21, 24], micro CT imaging is becoming a vital research tool for mouse brain imaging in general. In this article, we will focus on the mouse brain 3D imaging with a high-resolution laboratory micro CT system combined with an optimization of sample staining. We use different staining protocols based on iodine solution and phosphotungstic acid. The benefits of such approaches will be specified. We will summarize the comparison of the different staining procedures and we will also compare the gained data with the corresponding magnetic resonance data. Furthermore, the power of presented method will be demonstrated by a number of 3D models of selected brain structures with specifically complex 3D geometries that we have reconstructed and analyzed. 2 Materials and methods 2.1 Sample preparation for micro CT measurement An adult mouse was sacrificed with isoflurane overdose. The brain was dissected from skull and collected into ice-cold PBS. Subsequently, the sample was fixed in freshly prepared 4% paraformaldehyde (PFA). The sample was dehydrated in ethanol grade (12 h) and stained in 1% iodine solution in 90% methanol. After 24 hours in the solution, the sample was rinsed in ethanol rehydration series to end up in sterile distilled water. For the sake of micro CT measurement, the brain was embedded in 1% agarose gel and placed in a plastic tube. All animal (mouse) concerned work had been approved and permitted by the Ethical Committee on Animal Experiments (Norra Djurförsöksetiska Nämd, ethical permit N226/15 and N5/14) and was conducted according to The Swedish Animal Agency’s Provisions and Guidelines for Animal Experimentation recommendations. 2.2 Micro CT measurement and data processing Micro CT measurement was performed with a GE Phoenix v|tome|x L 240 (GE Sensing & Inspection Technologies GmbH, Germany), equipped with a nanofocus X-ray tube with maximum power of 180 kV/15 W. The data were acquired using a high contrast flat panel detector DXR250 with 2048 px ×2048 px, 200 µm×200 µm pixel size. The micro CT scan was carried out in an airconditioned cabinet (21◦C) at 60 kV acceleration voltage and 200 µA tube current. Exposure –2– 2018 JINST 13 C02039 time was 900 ms and 3 images were averaged for reducing the noise. Two different measurement procedures were performed. The first one was used for the evaluation of staining protocol. The brains contrasted by various techniques were scanned with a voxel resolution of 18 µm. After the evaluation, the second measurement procedure was applied for one brain with a voxel resolution of 6.5 µm. 2200 projections were taken over 360◦in this case. A tomographic reconstruction was realized by software GE phoenix datos|x 2.0 (GE Sensing & Inspection Technologies GmbH, Germany). Reconstructed slices were imported into a VG Studio MAX 3.1 (Volume Graphics GmbH, Germany). In this software, the 3D data were aligned so that orthogonal slices matched the horizontal (xy), coronal (yz) and sagittal plane (xz) resulting in series of 1700 coronal, 1090 sagittal and 680 horizontal tomographic sections of the brain. These data were compared with an anatomical atlas, and 3D models of anatomical structures were obtained by a segmentation based on the global thresholding. 2.3 MRI measurement The post-mortem mouse brain within the skull was measured by a high-resolution MRI scanner Bruker Avance 9.4 T (Bruker Biospin MRI, Ettlingen, Germany). The resolution of the obtained data was 0.027×0.027 mm (matrix 512×512 pixels). T1-weighted anatomical images of 25 parallel 2D slices were taken using the FLASH sequence with slice thickness 0.5 mm, interslice distance 0.5 mm, repetition time (TR) 461.3 ms, effective echo time (TE) 6.1 ms, 50 averages, flip angle 35.0◦, echo spacing 6.1 ms. The total measurement time was 9 hours and 25 images were obtained. 3 Results and discussion For a comprehensive visualization of the mouse brain structures done by a conventional micro CT, a chemical contrasting step is usually required. The differentiation of soft tissues by an imaging method delivers the biologically-relevant information only if the contrasting is sufficient. Our priority in this research was achieving the ultimate differential contrast enabling the identification of all major cell types or the tissue modes. Inorganic iodine and phosphotungstic acid (PTA) are the most broadly used agents in the field of post-mortem tissue contrasting for the following X-ray investigation [25]. Previously, it was suggested to use alcohol solutions containing these compounds to enable a rapid diffusion into the sample [8]. If iodine is compared to PTA, PTA appears as a larger molecule with much slower tissue penetration rates [8]. Ten different staining protocols were tested to visualize mouse brain structures by micro CT imaging (table 1). Samples were stained in PTA, iodine or combination of both of solutions for various periods of time. In order to compare the contrast among these samples, we analyzed all of them in the same condition. The voxel resolution of obtained CT data was 18 µm, which turned out to be enough to compare the quality of different contrasting protocols (figure 1). In summary, the staining in iodine exhibited more clear contours of fibres than in the case of PTA-stained samples. Fixation of samples in 4% PFA demonstrated higher contrast to tissues fixed in 10% PFA. The best resolution was obtained in 1% iodine in 90% methanol solution. As this staining showed the most contrastive and sharp edges of anatomical structures at coronal brain sections, we selected the sample stained with this protocol for further analysis. –3– 2018 JINST 13 C02039 Figure 1. Comparison of the coronal sections of samples following different staining protocols. (A) 1% PTA in 90% MeOH for 21 days, (B) 1% PTA in 90% MeOH for 14 days, (C) 1% PTA in 100% MeOH for 16 days, (D) 1% PTA for 10 days + 1% iodine in 90% MeOH for additional 24h, (E) 1% iodine in 100% EtOH for 24h, (F) 1% iodine in 100% EtOH for 48h, (G) 1% iodine in 100% EtOH for 7 days, (H) 1% iodine in 90% MeOH for 24h, (I) 1% iodine in 100% EtOH overnight, (J) 1% iodine in 100% EtOH for 24h. –4– 2018 JINST 13 C02039 Table 1. List of different processing and staining protocols for brain visualization in micro CT. Pre-fixation Post-fixation Dehydration Staining Time A4% PFA 4% PFA (24h) EtOH / 12h 1% PTA in 90% MeOH 21 days B/ 4% PFA (24h) EtOH / 12h 1% PTA in 90% MeOH 14 days C10% PFA 10% PFA (24h) EtOH / 12h 1% PTA in 100% MeOH 16 days D4% PFA 4% PFA (24h) EtOH / 12h 1% PTA + 1% iodine in 90% MeOH 10 days PTA/24h iodine E4% PFA 4% PFA (24h) EtOH / 12h 1% iodine in 100% EtOH 24h F4% PFA 4% PFA (24h) EtOH / 12h 1% iodine in 100% EtOH 48h G4% PFA 4% PFA (24h) EtOH / 12h 1% iodine in 100% EtOH 7 days H4% PFA 4% PFA (24h) EtOH / 12h 1% iodine in 90% MeOH 24h I10% PFA 10% PFA (12h) EtOH / 2h 1% iodine in 100% EtOH overnight J10% PFA 10% PFA (12h) EtOH / 2h 1% iodine in 100% EtOH 24h Four coronal sections (figure 2) from different parts of the mouse brain demonstrate the extent of resolution quality of contrasting protocol as captured by a micro CT scan. Identification of the Figure 2. Visualization of the internal brain structures as shown on CT images in the coronal sections. (A) ACO anterior commissure, CC corpus callosum, CP caudoputamen, LOT lateral olfactory tract, NDB nucleus of diagonal band, VL lateral ventricle. (B) CC corpus callosum, CP caudoputamen, FX fornix, INT internal capsule, OPT optic tract, SM stria medullaris (thalamus), V3 third ventricle, VL lateral ventricle. (C) CC corpus callosum, CP caudoputamen, DG dentate gyrus, FX fornix, INT internal capsule, MTT mamillothalmic tract, OPN olivary pretectal nucleus, V3 third ventricle. (D) ALV alveus, CC corpus callosum, CPD cerebral crus, DG dentate gyrus, MM medial mammillary nucleus, PC posterior commissure. –5– 2018 JINST 13 C02039 described structures in the coronal sections was based on the annotation from Allen Mouse Brain Atlas [26]. For instance, corpus callosum, cerebral peduncle and anterior commissure appeared among the most contrasted and easily identifiable structures. Furthermore, the smaller and finer structures such as mamillothalmic tract, column of the fornix or caudoputamen were successfully recognized as well. Because of the grey matter’s high contrast, the ventricles (e.g. lateral ventricle and third ventricle) were also evident. Although both micro CT and MRI belong to the category of methods providing 3D imaging of the internal structures, the character of the output data including the resolution quality is rather different. The geometry of the data element called voxel has a cubic shape (the same dimensions in all three axes) in the CT and a rectangular cuboid shape in the MRI. The MRI data voxel is defined by the pixel size and slice thickness, which is typically significantly larger than pixel dimensions, i.e. the lateral and axial resolution are different. Isotropic 3D imaging is feasible by both methods, however the length of such experiments needed for achieving a good signal-to-noise ratio is limiting. Consequently, the brain structures were analyzed in detail on 1900 coronal CT slices (with voxel size 6.5µm×6.5µm×6.5µm) and 25 coronal MRI slices (with voxel size 27 µm×27 µm×500 µm). From this point of view, CT data provided enhanced opportunities to study the arbitrary cross-sections within one dataset and also demonstrated an accurate 3D modelling of internal structures. The voxel values are defined by greyscale values depending on different properties of the material. CT greyscale values are defined by the X-ray absorption properties in relation to the atomic number of the sample material and to the accelerated voltage of the X-ray source [27, 28]. Based on this fact, brighter tissues refer to denser materials. In the case of anatomical T1 weighted MRI, the amount of the obtained signal depends on the time needed for realigning proton spins in main magnetic field (realigned to Boltzmann equilibrium) [29,30]. For instance, the fat realigns its spins quickly, and therefore it appears bright on a T1weighted image [29,30]. Water realigns much slower, and therefore it has a lower signal and is represented by darker values. As for the measurement time, it depends on the particular method. Considering ex vivo approach plus the highest possible resolution of both systems, the typical brain measurement took approx. 9 hours for the MRI and 2 hours for the CT. To compare corresponding MRI and micro CT data (figure 3, table 2), we selected four pairs of corresponding coronal slices of the adult mouse brain. Corpus callosum, anterior commissure, medial mammillary nucleus, posterior commissure and cerebral peduncle were clearly recognizable on both micro CT and MRI images. All the mentioned structures demonstrated well-defined edges in micro CT data. Lateral ventricle, lateral olfactory bulb, optic tract, stria medullaris, internal capsule and cerebral peduncle were also detectable in both micro CT and MRI data. However, unlike the micro CT images, MRI pictures did not display a clear identification of above mentioned structures. The structures such as myelinated fibres in caudoputamen, third ventricle, column of the fornix, dentate gyrus, mamillothalmic tract and other locations were clearly distinguished in the micro CT. Moreover, the compartmentation of dorsal hippocampal commissure from corpus callosum and alveus was possible only in micro CT images. As the next step, we selected several structures for segmentation to demonstrate possible outputs from micro CT data (figure 4, figure 5, figure 6). Tissue segmentation and construction of accurate –6–