- Open Access
Direct comparison of MRI and X-ray CT technologies for 3D imaging of root systems in soil: potential and challenges for root trait quantification
- Ralf Metzner†1Email author,
- Anja Eggert†2,
- Dagmar van Dusschoten1,
- Daniel Pflugfelder1,
- Stefan Gerth2,
- Ulrich Schurr1,
- Norman Uhlmann2 and
- Siegfried Jahnke1
© Metzner et al.; licensee BioMed Central. 2015
- Received: 28 November 2014
- Accepted: 23 February 2015
- Published: 11 March 2015
Roots are vital to plants for soil exploration and uptake of water and nutrients. Root performance is critical for growth and yield of plants, in particular when resources are limited. Since roots develop in strong interaction with the soil matrix, tools are required that can visualize and quantify root growth in opaque soil at best in 3D. Two modalities that are suited for such investigations are X-ray Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). Due to the different physical principles they are based on, these modalities have their specific potentials and challenges for root phenotyping. We compared the two methods by imaging the same root systems grown in 3 different pot sizes with inner diameters of 34 mm, 56 mm or 81 mm.
Both methods successfully visualized roots of two weeks old bean plants in all three pot sizes. Similar root images and almost the same root length were obtained for roots grown in the small pot, while more root details showed up in the CT images compared to MRI. For the medium sized pot, MRI showed more roots and higher root lengths whereas at some spots thin roots were only found by CT and the high water content apparently affected CT more than MRI. For the large pot, MRI detected much more roots including some laterals than CT.
Both techniques performed equally well for pots with small diameters which are best suited to monitor root development of seedlings. To investigate specific root details or finely graduated root diameters of thin roots, CT was advantageous as it provided the higher spatial resolution. For larger pot diameters, MRI delivered higher fractions of the root systems than CT, most likely because of the strong root-to-soil contrast achievable by MRI. Since complementary information can be gathered with CT and MRI, a combination of the two modalities could open a whole range of additional possibilities like analysis of root system traits in different soil structures or under varying soil moisture.
- X-ray Computed Tomography (CT)
- Magnetic Resonance Imaging (MRI)
- Root system architecture
- Common bean (Phaseolus vulgaris L.) 3D imaging
- Roots in soil
Roots are vital for higher plants to gather water and nutrients and thus critical for performance and productivity . Considering the impact on crop yields, more knowledge is needed about root system development belowground and interaction with the surrounding soil [2-4]. Approaches to investigate roots and their development reach from excavation and measurement of root system traits with manual or optical methods [5-7], mini-rhizotrons  or trenches dug into the ground , transparent artificial growth media [8,9] to soil based 2D growth monitoring in rhizoboxes [1,10,11]. All these are valid for answering specific questions concerning root development; however, none of these is able to follow 3D root development in soil where a multitude of biotic and abiotic interaction takes place. Quantitative knowledge of 3D root traits would help to achieve better mechanistic understanding of root architecture as it develops in a soil environment. Non-invasive approaches which have been used already to investigate plant roots in soil are magnetic resonance imaging (MRI) [12-15], neutron computed tomography [16,17] and X-ray computed tomography (CT) [3,18-21]. Neutron tomography requires access to a nuclear reactor or a high energy particle accelerator whereas MRI and CT on the other hand, though still expensive, are becoming available to a steadily growing number of plant biologists.
CT employs an X-ray beam passing through the sample which absorbs part of it thereby reducing the intensity of the beam. This process is called attenuation. The sample is rotated between an X-ray source and a detector, recording series of 2D projections of the object from which a 3D volume dataset can be reconstructed. X-ray attenuation is mainly determined by material properties, in particular electron-density. Thereby the internal structure of the sample becomes visible by contrast according to density and atomic number of the elements . Further information about the basics of CT can be found in various articles [23,24] or text books . Recent applications in plant sciences include lateral root development  or root elongation rates . A number of other applications covering a range of plant species and root traits are reviewed in . A major problem is the often very similar attenuation of roots and some structures in the soil such as water filled pores . This leads to low contrast hindering simple and straightforward segmentation of the roots from the soil background. Different approaches were tried to tackle this problem, such as adjusting soil humidity to optimize contrast  or sophisticated software tools for data analysis . Together with scanners that can produce high resolution scans in short time  these approaches may lead to a wider use of CT in plant sciences.
MRI is based on the magnetic moment of atomic nuclei like 1H (protons) which are highly abundant in living tissues particularly in water molecules. The magnetic moment can be manipulated using strong magnetic fields and radio frequency fields to produce 3D datasets of samples. The magnetic fields require a substrate with low ferro-magnetic particle content for high quality images. A range of contrast parameters can be exploited, highlighting differences within the sample such as density of the protons or their physical and chemical micro-environment. This can be exploited to produce a strong difference between ‘root water signal’ and ‘soil water signal’ which provides a very high contrast between roots and soil background [13,14]. The basic principles of MRI and its use in biomedical sciences are described in detail in several textbooks [30,31] or review articles of plant biology [12,32-34]. Research applications to plant roots range from phytopathology , across storage root internal structures  to combined studies with positron emission tomography for structure function relations . Also water mobility in roots and soil has been shown to be detectable with MRI [36,37].
Both CT and MRI are able to image root systems in soil. Their fundamental different physical principles give each of them a unique perspective but also different challenges for application in the plant sciences. Therefore, a direct comparison of both techniques using the same samples and analyzing for the same root system may help to decide which method is better suited for a specific research question. A whole range of instrument parameters can affect root imaging with either method but also sample parameters such as soil type [14,38] or moisture . Pot sizes, on the other hand has rarely been considered despite its importance for root system and plant development  making it an important parameter of experimental design. Since pot size may also affect root imaging and data analysis depending on the applied method, we focus in our CT-MRI comparison approach on this parameter using common bean (Phaseolus vulgaris) as a model plant. We evaluated both techniques for imaging roots in soil, segmentation of roots from the soil background and extracting root length as a basic parameter of root system architecture (RSA). For comparison the roots were finally excavated, washed and scanned using WinRHIZO as a widely used standard technique.
We compared reconstructed root systems of bean plants in pots of three different sizes imaged by both CT and MRI. The small pots had an inner diameter (I.D.) of 34 mm and a height of 200 mm which could be used to study early growth stage in crop plants. Pots of this size or smaller were also used in other high resolution CT studies [3,19]. The medium pots with an I.D. of 56 mm and a height of 200 mm would allow plant growth for a few weeks or even a whole growth cycle of small plants. The large pots had I.D. of 81 mm and were 300 mm high, thereby suited to cultivate plants for a longer time and even plants with large storage roots like sugar beet . Here, we focused on the potentials and challenges for image quality and extracting root length as a major trait of RSA [1,11] to compare the two imaging modalities.
CT imaging of roots is based on high spatial resolution and a good segmentation procedure
Small pots (I.D. 34 mm)
Medium pots (I.D. 56)
Large pots (I.D. 81)
Total root length [mm]
Percent root length of WinRHIZO [%]
MRI imaging of roots is based on high root to soil contrast
The MRI image (Figure 1c) showed a similar 3D structure of the root system as segmented from the CT data (Figure 1b). Due to high contrast between roots and soil background, root segmentation from an MRI image is not necessary which means that Figure 1c shows the raw data with only a noise cut-off and no further processing. While the MRI scan took about as long as the CT scan, it must be noted that for the CT scan it took another two hours to reach the segmented image shown in Figure 1b. Though the root images from CT and MRI appear rather similar, there are some distinct differences due to fundamental differences in the imaging and segmentation principles. Roots which appear to be very thin in CT often look much thicker in MRI. This is caused by the much coarser spatial resolution of MRI (Figure 1c) with a voxel size of 330 × 330 × 1000 μm3, so any root detected is shown with voxels of this size whereas in the CT image (Figure 1b) the voxel size was 56 × 56 × 56 μm3. As already stated, in MRI the root detection is based on contrast against soil background which in this study was about two orders of magnitude, rather than on geometrical recognition as in CT. That means that roots even thinner than voxel resolution can be recognized as roots when only a (detectable) small fraction of the actual volume is comprised of root tissue. Therefore parts of the roots displayed in Figure 1c as one voxel thick were in reality thinner due to this sub-voxel resolution. We estimated that the thinnest roots detected with MRI were about 250 μm in diameter. The root length found with MRI was still only 70% of the total root length found with WinRHIZO after harvest (Table 1). Upon taking only the length of roots in the WinRHIZO classes thicker than 400 μm into account this fraction was 96%. Therefore, similar to CT, roots above this diameter appear to be largely detected, and in bean plants the primary, basal and hypocotyl borne roots are typically all above this threshold , the major root structure can be visualized in 3D. The high contrast between roots and soil background in MRI results from the fact that the signal decay from soil water is much faster than that from roots, so with the appropriate measurement settings only the latter is recorded. Since the signal decay of the soil water depends on soil properties, the measurement settings need to be adjusted adequately to the specific substrate. For MRI imaging of roots in soil the choice of substrate is important also because, due to the strong magnetic fields required for the measurements, disturbances of these fields have adverse effects on image quality. Ferromagnetic particles are thus most critical and their presence in the substrate for MRI should be minimized. This can either be done by selecting soils with naturally low content of ferromagnetic particles  or remove them from the sieved substrate with a simple arrangement of permanent magnets  as was used in this study. Paramagnetic ions such as Mn2+ described to be also problematic by some authors [50,51] did not appear to have any strong effect in our studies as demonstrated by the good image quality in the chosen soil (Figure 1c). The removal process for ferromagnetic particles led to a sieved and homogenized soil, and this substrate with good MRI properties was also used for CT. In contrast to pure MRI studies on roots [15,35] a higher fraction of coarse sand was used here, which may cause problems for plant cultivation because of low water retention capacity. However, pretests had shown that this allowed better segmentation results for CT due to higher contrast between roots and substrate, whereas MRI image quality was only marginally affected. Homogenized substrates appears to be advantageous for CT as it provides a rather homogenous background with small pores optimal for segmentation, which may be the reason why this type of substrate was also used in many CT studies targeting finer roots [28,52,53]. The need of homogenized substrates seems to be a common limitation for current MRI or CT applications in plant studies. Compared to natural soil from the field, which would be the ultimate condition to grow e.g. crop plants in, measure and characterize root traits, sieved soil comprises loss of texture and aggregate structure. Nevertheless, both methods are in principle able to investigate roots in soil cores collected in the field, as has been shown for CT  and also in a first MRI experiment in our lab (data not shown).
For small pots both modalities recovered a comparable fraction of a root system with more details visible in the CT
When comparing the root images provided by both techniques, it is obvious that CT (Figure1b) with its higher spatial resolution showed many of the thinner roots with a higher level of detail than MRI (Figure 1c). For studies targeting different root classes in plants such as bean  or maize  this feature may be advantageous for identification and tracking of lateral roots of each class which are difficult to follow with MRI. Also thinner root systems in general may be investigated with CT as shown for Arabidopsis roots  which can be hardly achieved with MRI.
Visual comparison between root systems in 3D can be quite useful for qualitative description of treatment effects but, in addition to that, a quantitative evaluation of trait development is required. For the root system in the small pots shown in Figure 1, a root length of 2.8 m with CT and 2.9 m with MRI on the following day was found (Table 1) fitting well with the slightly longer roots in the lower part of the pot shown with MRI (Figure 1c) compared to CT (Figure 1b). Considering this, the fraction of the WinRHIZO-detected roots found by CT and MRI (69% and 73% respectively; Table 1) was almost identical, i.e. for such a (small) pot there was no obvious difference between MRI and CT in acquiring root lengths. Another experiment with a similar plant yielded longer total root length for all methods (Additional file 1), but the fraction of the roots found by WinRHIZO of CT (61%) and MRI (71%) was similar to that in Figure 1.
For medium sized pots, CT showed more details but MRI revealed a larger fraction of the root system
On the other hand, many roots detected with MRI also appeared to be successfully segmented with CT (Figure 2c) with two significant exceptions: on the left hand side and the bottom ~20 mm of the pot (Figure 2c) almost no roots could be successfully segmented from CT data, while MRI showed an abundance of roots. The lack of segmentation of roots from the CT image at these different positions can be taken as an example to explain different challenges frequently met when extracting roots from CT measurements. High water content can be an issue for segmentation as attenuation of roots and water-filled pores may overlap [28,42]. Gravity will inevitably lead to higher water content at the bottom, thereby degrading CT root segmentation particularly at the bottom of the pot; as dry soil also may increase artefacts from air filled pores in segmentation, a compromise may be to keep the soil in the pot at the soil-specific field capacity . However, especially in higher pots needed to study the development of long roots or in sandy soils, the problem of water gradients in the pot still remains leading to segmentation problems towards the bottom of the pot. Additionally, water uptake by plants would also alter water distribution in a pot. It is therefore advisable to measure these effects for the chosen substrate-pot combination, water contents, CT-scanner and imaging system as a very basic control. Since the lowest part of a pot is the most likely to cause segmentation problems with CT, Figure 2b showed no roots in this area whereas a large number of roots was monitored at the bottom of the pot with MRI (Figure 2a). The MRI image also showed more unconnected voxels in this area with grey values similar to the roots compared to the upper part. We observed such unconnected voxels to be an artefact in MRI in the soil mixture used here when the water content surpassed about 10% to 15% (v/v). Since these voxels are not connected to the roots or to each other they were not counted in the quantification of root length unless they were directly adjacent to the roots. This also points to a high water content at the bottom of the pot as the reason for failed CT root segmentation in this volume. However, we occluded roots found by MRI in this volume from root length quantification in order to compare similar volumes for both CT and MRI measurements.
For the left side of the medium pot there was no hint in the MRI image (Figure 2a) of high water content. The failure in root segmentation by CT in this region might be therefore different from what has been discussed before. Another problem in CT root segmentation can be an imprecise removal of the pot material that accidentally removed root segments from the image during the image processing pipeline. If part of a major root is removed, all the attached roots will also be lost for root length calculation if the resulting gap is too large. For MRI this was not an issue since PVC, similar to most other common plastic materials, gave no detectable signal at the applied measurement settings and therefore did not show up on the images. Nevertheless, the strong difference in root length extracted from the medium pot by CT or MRI (2.8 m and 4.6 m respectively), was unlikely based exclusively on this effect. With reference to the WinRHIZO data, MRI found 95% of the roots which, considering that the lowest part of the pot was not used for the MRI root length calculation, advises root length to be somewhat overestimated here by MRI. Possibly, some signal caused by high water content in the soil adjacent to the roots could not be excluded from root length quantification. Therefore, also for MRI it seems to be advisable to investigate effects of high soil water content, probably close to or above field capacity, for the specific plant-soil system on root detection and root length quantification as a basic control.
For the large pots, MRI showed a larger fraction of the root system and more details compared to CT
Both CT and MRI produced high quality 3D images of root systems in vivo as demonstrated here for bean plants grown in soil-filled pots of three different sizes (34, 56 and 81 mm in diameter). For comparison, root lengths of the same plants were calculated from data obtained in consecutive measurement with both modalities. The obtained length of the investigated root systems and image details were distinctly different between CT and MRI and, in particular dependent on the pot size. Compared to CT, MRI was less affected by a lower spatial resolution associated with larger pot sizes and appeared to be better suited for imaging large root systems, e.g. at later developmental stages. On the other hand, CT gave almost identical results as MRI for small pots with respect to root length but provided more detailed information such as finely graduated root diameter estimation, in particular of small roots, due to higher spatial resolution. The possibility to visualize soil structures is a further intriguing feature of CT while MRI opens the possibility of measuring the water content of roots or soils. Considering the time needed to analyze a full root system, the scan time for both modalities was comparable but segmentation of roots from CT images took much longer than the scans while MRI data could be used for quantification immediately. Segmentation of the CT data, however, may be performed separately while scanning on other plants continues. According to WinRHIZO data, however a considerable portion of the root system present could not be detected by either MRI or CT even in the small pots. Most likely these are roots thinner than 400 μm. Altogether combining the two modalities on the same samples would be intriguing as, with direct root and water content detection by MRI and soil structure imaging by CT, many questions regarding root-soil interactions can be addressed. Interesting sub-volumes identified with MRI in larger pots could then be targeted and investigated in detail with high resolution CT. As demonstrated here, both methods have their specific pros and cons, but are excellent tools to non-destructively investigate properties of roots growing in soil.
Common bean plants (Phaseolus vulgaris L. cv. ‘Shiny Fardenlosa’) were grown in a growth chamber from seed in a mixture of homogenized agricultural topsoil and coarse sand (1:9; v/v). A mix with less sand and the same soil was already shown to be suited for MRI application . The agricultural soil, characterized as a gleyic cambisol, was collected by removing the top 30 cm from an agricultural field (Kaldenkirchen, Germany) and air dried. Subsequently, the soil was powdered and homogenized in a drum hoop mixer (J. Engelsmann, Ludwigshafen, Germany), sieved to 2 mm and freed of stronger ferromagnetic particles by moving it in a thin layer on a conveyor belt through a magnetic field provided by rare earth magnets (NdFeB N42, 1.3 T; Webcraft GmbH, Gottmardingen, Germany). Coarse quartz sand (grain size 0.71 to 1.4 mm; Quartzwerke Witterschlick, Alfter, Germany) was similarly freed of ferromagnetic particles. The ready mixture was filled into PVC tubes of three different sizes with (a) an inner diameter of 34 mm, a height of 200 mm, a volume of 0.182 l, (b) and I.D. of 56 mm, a height of 200 mm, a volume of 0.49 l or (c) an inner diameter (I.D.) of 81 mm, a height of 300 mm and a volume of 1.5 l. All had holes drilled into their bottom caps for drainage and aeration covered with nylon mesh (grid size 200 μm) to prevent loss of substrate and roots growing out.
The pots were watered to above container capacity and, after excess water had drained away, seeds were laid down in holes 2 cm deep and covered with soil. After germination the pots were watered automatically once per day with tap water. The growth chamber was set to 16 h light/ 8 h dark and 20°C/16°C, respectively, while relative humidity was kept constant at 60 ± 3%. Lighting was provided by 5 × 400 W HPI and 5 × 400 W SON-T lamps (both Philips, Hamburg, Germany) which alternated every 2 hours with 5 min overlap giving PAR intensity between 350 and 450 μmol m−2 s−1 at canopy level.
For the root detection we used the CT setup at the Fraunhofer Development Center X-ray Technology (EZRT) . The CT setup was used with a FXE 225.99 X-ray tube and a Perkin Elmer XRD 1620 detector, which is a flat panel type and operates at a frame rate of 2 images s−1 in 14-bits full frame mode (2048 × 2048 pixels).
Parameters of CT measurements
Tube voltage [kV]
Tube current [μA]
Exposure time [ms]
Cu 1 mm
Cu 1 mm
Number of projections
Voxel size [μm]
FOV in horizontal plane [mm]
FOV vertical [mm]
Measurement time [min]
The scans were made with continuous sample rotation (fly-by) over 360°. For improved handling and decreased calculation times the root structure was analyzed after a 2 × 2 binning with a voxel sampling of 56 μm for pot size 34 mm, 68 μm for pot size 56 mm and 99 μm for pot size 81 mm. The effective dose to the sample during the scan was measured with an additional pot in the center of which a Dose Rate Meter 6150 AD with Scintillator Probe from automess (Automation und Messtechnik GmbH, Ladenburg, Germany) was buried in the same soil and subjected to the same X-ray exposure as the plant samples described above. This measurement gave a value of 274 mGy.
Segmentation of roots from CT images
For the root segmentation we used the Volume Player Plus (VPP) software of the EZRT and the Modular Algorithms for Volume Images (MAVI) software package (Fraunhofer ITWM Kaiserslautern, Germany). In the first step the pot material was removed by manually fitting an ellipse to the inner pot wall in the first slice and removing everything outside this ellipse in all slices of the vertical stack. In cases where thin roots are directly abutting the pot wall or the pot is not perfectly rounded this may cause the loss of root voxels. Afterwards the images were binarised using a range of different gray values. First we determined for every data set which gray value area contains the roots. This area is quite wide due to the different thickness of the roots. Big roots show higher gray values than small ones. According to this we limited the gray value area by defining gray value ranges for small, middle and big roots based on empirical knowledge. In summary we cover the whole gray value area containing roots, but split this area in three parts linked to root thickness by hysteresis binarization. As a result we receive three partial volumes to work with. In the following steps every partial volume is cleaned from remaining non-root parts (described later on). At the end the three volumes containing small, medium and big roots were combined to one volume again and a closing algorithm is used to generate a closed architecture. The closing algorithm is dilation followed by erosion of the objects and was used with a closing radius of three times the respective voxel size. With the MAVI software the remaining non-root parts in the binarised image like stones, soil and air or water filled pores were removed using the labeling function and the object filter. The labeling function assigns to every connected object in the image a unique value. For defining the connectivity of the objects a neighborhood system is used. The information to the different labels is stored in the image history. With this history an object filter can separate them working with object size and sphericity. Small and roundish labels are defined by their sphericity as non-root objects and removed from the image. A sphere would have a value of 1 and a line one of 0. The threshold was set to 0.3. Further information about the image processing can be found in the textbook written by Ohser and Schladitz . The whole procedure for segmenting roots took 1 h for one dataset, in case of the 34 mm pots 2 and for 56 mm pots 3 datasets were acquired. Both pots were of the same height but for the small pots MRI showed that they had not reached the bottom. In case of 81 mm pots 4 datasets were necessary to cover the whole pot, whereas only one could be used for segmentation.
Parameters of MRI measurements
Voxel size in horizontal plane [μm]
Voxel size vertical [μm]
Repetition time T R [s]
Echo time T E [ms]
Number of averages
FOV in horizontal plane [mm]
FOV vertical [mm]
Measurement time [min]
Segmentation of roots from MRI images
Segmentation of roots from MRI images was achieved using a simple threshold on the MRI signal. The threshold was set to four times the noise level of the MRI data. For the medium sized pot, the excessive soil water was visible in the MRI images. For this data set the threshold was manually increased to remove the soil voxels.
Prior to root length calculation, the MRI images were co-registered to the CT data by automated image registration using the program FSL  and interpolated onto the same grid. As some parts of the CT images could not be segmented, this co-registration step assures that the images of both modalities contain the same part of the root system. Furthermore, effects due to different voxel sizes are avoided by using the same data grid. Root length calculation was done as described by , with some adaptations to be able to process not only MRI images but also segmented root masks obtained from the CT data. First, to be able to process the data with our developed software tools the images were downsampled by an integer factor, resulting in isotropic voxel sizes for the individual plant between 300-400 μm. The root system architecture was then extracted by calculating the cheapest path  from each segmented root voxel to the manually selected shoot. In Schulz et al. , the cost of a path element was determined by the geometrical length and the MRI image intensity at the respective position. For the CT data, the actual image intensity is not a useful measure to describe the root structure, thus we only worked with the segmented root mask. As the ‘image value’ for the segmented mask was either 0 or 1, all paths had equal costs and multiple parallel traces in one root segment did occur during the root extraction. These were avoided by additionally weighting the path costs with the distance transform of the root mask, thus favoring paths that run in the center of the segmented mask. The resulting root trees then needed pruning: paths shorter than 3 mm were removed, as well as paths bridging more than 3.5 mm of non-segmented voxels. After visual inspection of the resulting root system architecture, the total length was finally calculated from it.
This study was supported by the German Federal Ministry for Education and Science (BMBF) within the German-Plant-Phenotyping Network (project identification number 031A053). The authors would like to thank Esther Breuer for help with plant cultivation and doing the WinRHIZO analysis, Johannes Kochs for technical support in designing and building hardware for MRI and Vitali Weizel for technical support with the CT instrument.
- Gregory PJ. Plant roots. Oxford, UK: Blackwell Publishing Ltd; 2006.View ArticleGoogle Scholar
- Gregory PJ, Bengough AG, Grinev D, Schmidt S, Thomas WTB, Wojciechowski T, et al. Root phenomics of crops: opportunities and challenges. Funct Plant Biol. 2009;36:922–9.View ArticleGoogle Scholar
- Mooney SJ, Pridmore TP, Helliwell J, Bennett MJ. Developing X-ray computed tomography to non-invasively image 3-D root systems architecture in soil. Plant Soil. 2012;352:1–22.View ArticleGoogle Scholar
- Giehl RFH, von Wirén N. Root nutrient foraging. Plant Physiol. 2014;166:509–17.View ArticlePubMedGoogle Scholar
- Trachsel S, Kaeppler SM, Brown KM, Lynch JP. Shovelomics: high throughput phenotyping of maize (Zea mays L.) root architecture in the field. Plant Soil. 2011;341:75–87.View ArticleGoogle Scholar
- Lynch JP. Root phenes for enhanced soil exploration and phosphorus acquisition: tools for future crops. Plant Physiol. 2011;156:1041–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Eshel A. Plant roots: the hidden half. 4th ed. Boca Raton, Florida, USA: CRC Press; 2013.Google Scholar
- Iyer-Pascuzzi AS, Symonova O, Mileyko Y, Hao YL, Belcher H, Harer J, et al. Imaging and analysis platform for automatic phenotyping and trait ranking of plant root systems. Plant Physiol. 2010;152:1148–57.View ArticlePubMed CentralPubMedGoogle Scholar
- Downie H, Holden N, Otten W, Spiers AJ, Valentine TA, Dupuy LX. Transparent soil for imaging the rhizosphere. PLoS One. 2012;7:e44276.View ArticlePubMed CentralPubMedGoogle Scholar
- Nagel KA, Kastenholz B, Jahnke S, Van Dusschoten D, Aach T, Muhlich M, et al. Temperature responses of roots: impact on growth, root system architecture and implications for phenotyping. Funct Plant Biol. 2009;36:947–59.View ArticleGoogle Scholar
- Neumann G, George TS, Plassard C. Strategies and methods for studying the rhizosphere-the plant science toolbox. Plant Soil. 2009;321:431–56.View ArticleGoogle Scholar
- Blümler P, Windt CW, Van Dusschoten D. Magnetic resonance in plants. Nova Acta Leopoldina. 2009;357:17–30.Google Scholar
- Jahnke S, Menzel MI, van Dusschoten D, Roeb GW, Bühler J, Minwuyelet S, et al. Combined MRI-PET dissects dynamic changes in plant structures and functions. Plant J. 2009;59:634–44.View ArticlePubMedGoogle Scholar
- Rogers HH, Bottomley PA. In situ Nuclear Magnetic Resonance imaging of roots - Influence of soil type, ferromagnetic particle content and soil water. Agron J. 1987;79:957–65.View ArticleGoogle Scholar
- Metzner R, van Dusschoten D, Bühler J, Schurr U, Jahnke S. Belowground plant development measured with magnetic resonance imaging (MRI): exploiting the potential for non-invasive trait quantification using sugar beet as a proxy. Front Plant Sci. 2014;5:469.View ArticlePubMed CentralPubMedGoogle Scholar
- Tumlinson LG, Liu HY, Silk WK, Hopmans JW. Thermal neutron computed tomography of soil water and plant roots. Soil Sci Soc Am J. 2008;72:1234–42.View ArticleGoogle Scholar
- Leitner D, Felderer B, Vontobel P, Schnepf A. Recovering root system traits using image analysis exemplified by two-dimensional neutron radiography images of lupine. Plant Physiol. 2014;164:24–35.View ArticlePubMed CentralPubMedGoogle Scholar
- Jenneson PM, Gilboy WB, Morton EJ, Gregory PJ. An X-ray micro-tomography system optimised for the low-dose study of living organisms. Appl Radiat Isot. 2003;58:177–81.View ArticlePubMedGoogle Scholar
- Flavel RJ, Guppy CN, Tighe M, Watt M, McNeill A, Young IM. Non-destructive quantification of cereal roots in soil using high-resolution X-ray tomography. J Exp Bot. 2012;63:2503–11.View ArticlePubMedGoogle Scholar
- Zhu JM, Ingram PA, Benfey PN, Elich T. From lab to field, new approaches to phenotyping root system architecture. Curr Opin Plant Biol. 2011;14:310–7.View ArticlePubMedGoogle Scholar
- Koebernick N, Weller U, Huber K, Schluter S, Vogel HJ, Jahn R, et al. In situ visualization and quantification of three-dimensional root system architecture and growth using X-Ray computed tomography. Vadose Zone J. 2014;13:8.View ArticleGoogle Scholar
- Plews AG, Atkinson A, McGrane S. Discriminating structural characteristics of starch extrudates through X-ray micro-tomography using a 3-D watershed algorithm. Int J Food Eng. 2009;5:11.View ArticleGoogle Scholar
- Hanke R, Fuchs T, Uhlmann N. X-ray based methods for non-destructive testing and material characterization. Nucl Instrum Meth A. 2008;591:14–8.View ArticleGoogle Scholar
- Salvo L, Cloetens P, Maire E, Zabler S, Blandin JJ, Buffiere JY, et al. X-ray micro-tomography an attractive characterisation technique in materials science. Nucl Instrum Meth B. 2003;200:273–86.View ArticleGoogle Scholar
- Kalender WA. Computed tomography: fundamentals, system technology, image quality, applications. 3 revth ed. Erlangen, Germany: Publicis Publ; 2011.Google Scholar
- Bao Y, Aggarwal P, Robbins NE, Sturrock CJ, Thompson MC, Tan HQ, et al. Plant roots use a patterning mechanism to position lateral root branches toward available water. Proc Natl Acad Sci U S A. 2014;111:9319–24.View ArticlePubMed CentralPubMedGoogle Scholar
- Tracy SR, Black CR, Roberts JA, Sturrock C, Mairhofer S, Craigon J, et al. Quantifying the impact of soil compaction on root system architecture in tomato (Solanum lycopersicum) by X-ray micro-computed tomography. Ann Bot (London). 2012;110:511–9.View ArticleGoogle Scholar
- Zappala S, Mairhofer S, Tracy S, Sturrock CJ, Bennett M, Pridmore T, et al. Quantifying the effect of soil moisture content on segmenting root system architecture in X-ray computed tomography images. Plant Soil. 2013;370:35–45.View ArticleGoogle Scholar
- Mairhofer S, Zappala S, Tracy S, Sturrock C, Bennett MJ, Mooney SJ, et al. Recovering complete plant root system architectures from soil via X-ray mu-Computed Tomography. Plant Methods. 2013;9:1–7.View ArticleGoogle Scholar
- Haacke EM, Brown RW, Thompson MR, Venkatesan R. Magnetic Resonance Imaging. New York, USA: John Wiley and Sons; 1999.Google Scholar
- Callaghan PT. Principles of nuclear magnetic resonance microscopy. New York, USA: Oxford University Press; 1993.Google Scholar
- Köckenberger W, De Panfilis C, Santoro D, Dahiya P, Rawsthorne S. High resolution NMR microscopy of plants and fungi. J Microsc. 2004;214:182–9.View ArticlePubMedGoogle Scholar
- Van As H, Scheenen T, Vergeldt FJ. MRI of intact plants. Photosynth Res. 2009;102:213–22.View ArticlePubMed CentralPubMedGoogle Scholar
- Borisjuk L, Rolletschek H, Neuberger T. Surveying the plant’s world by magnetic resonance imaging. Plant J. 2012;70:129–46.View ArticlePubMedGoogle Scholar
- Hillnhütter C, Sikora RA, Oerke EC, van Dusschoten D. Nuclear magnetic resonance: a tool for imaging belowground damage caused by Heterodera schachtii and Rhizoctonia solani on sugar beet. J Exp Bot. 2012;63:319–27.View ArticlePubMed CentralPubMedGoogle Scholar
- Gruwel MLH. In situ magnetic resonance imaging of plant roots. Vadose Zone J. 2014;13:3.View ArticleGoogle Scholar
- MacFall JS, Johnson GA. Plants, seeds, roots, and soils as applications of magnetic resonance microscopy. In: Encyclopedia of magnetic resonance. New York, USA: John Wiley & Sons, Ltd; 2012.Google Scholar
- Lontoc-Roy M, Dutilleul P, Prasher SO, Han LW, Brouillet T, Smith DL. Advances in the acquisition and analysis of CT scan data to isolate a crop root system from the soil medium and quantify root system complexity in 3-D space. Geoderma. 2006;137:231–41.View ArticleGoogle Scholar
- Poorter H, Bühler J, van Dusschoten D, Climent J, Postma JA. Pot size matters: a meta-analysis of the effects of rooting volume on plant growth. Funct Plant Biol. 2012;39:839–50.View ArticleGoogle Scholar
- Schmidt S, Bengough AG, Gregory PJ, Grinev DV, Otten W. Estimating root-soil contact from 3D X-ray microtomographs. Eur J Soil Sci. 2012;63:776–86.View ArticleGoogle Scholar
- Helliwell JR, Sturrock CJ, Grayling KM, Tracy SR, Flavel RJ, Young IM, et al. Applications of X-ray computed tomography for examining biophysical interactions and structural development in soil systems: a review. Eur J Soil Sci. 2013;64:279–97.View ArticleGoogle Scholar
- Kaestner A, Schneebeli M, Graf F. Visualizing three-dimensional root networks using computed tomography. Geoderma. 2006;136:459–69.View ArticleGoogle Scholar
- Mairhofer S, Zappala S, Tracy SR, Sturrock C, Bennett M, Mooney SJ, et al. RooTrak: automated recovery of three-dimensional plant root architecture in soil from X-ray microcomputed tomography images using visual tracking. Plant Physiol. 2012;158:561–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Tracy SR, Roberts JA, Black CR, McNeill A, Davidson R, Mooney SJ. The X-factor: visualizing undisturbed root architecture in soils using X-ray computed tomography. J Exp Bot. 2010;61:311–3.View ArticlePubMedGoogle Scholar
- Ohser J, Schladitz K. 3D images of materials structures: processing and analysis Weinheim. Germany: Wiley-VCH; 2009.View ArticleGoogle Scholar
- Rubio G, Lynch JP. Compensation among root classes in Phaseolus vulgaris L. Plant Soil. 2007;290:307–21.View ArticleGoogle Scholar
- Wildenschild D, Hopmans JW, Vaz CMP, Rivers ML, Rikard D, Christensen BSB. Using X-ray computed tomography in hydrology: systems, resolutions, and limitations. J Hydrol. 2002;267:285–97.View ArticleGoogle Scholar
- Perret JS, Al-Belushi ME, Deadman M. Non-destructive visualization and quantification of roots using computed tomography. Soil Biol Biochem. 2007;39:391–9.View ArticleGoogle Scholar
- Baveye PC, Laba M, Otten W, Bouckaert L, Dello Sterpaio P, Goswami RR, et al. Observer-dependent variability of the thresholding step in the quantitative analysis of soil images and X-ray microtomography data. Geoderma. 2010;157:51–63.View ArticleGoogle Scholar
- Bottomley PA, Rogers HH, Foster TH. NMR Imaging shows water distribution and transport in plant-root systems in situ. Proc Natl Acad Sci U S A. 1986;83:87–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Heeraman DA, Hopmans JW, Clausnitzer V. Three dimensional imaging of plant roots in situ with x-ray computed tomography. Plant Soil. 1997;189:167–79.Google Scholar
- Gregory PJ, Hutchison DJ, Read DB, Jenneson PM, Gilboy WB, Morton EJ. Non-invasive imaging of roots with high resolution X-ray micro-tomography. Plant Soil. 2003;255:351–9.View ArticleGoogle Scholar
- Hargreaves CE, Gregory PJ, Bengough AG. Measuring root traits in barley (Hordeum vulgare ssp vulgare and ssp spontaneum) seedlings using gel chambers, soil sacs and X-ray microtomography. Plant Soil. 2009;316:285–97.View ArticleGoogle Scholar
- Hochholdinger F, Tuberosa R. Genetic and genomic dissection of maize root development and architecture. Curr Opin Plant Biol. 2009;12:172–7.View ArticlePubMedGoogle Scholar
- Lucas M, Swarup R, Paponov IA, Swarup K, Casimiro I, Lake D, et al. SHORT-ROOT Regulates Primary, Lateral, and Adventitious Root Development in Arabidopsis. Plant Physiol. 2011;155:384–98.View ArticlePubMed CentralPubMedGoogle Scholar
- Jenkinson M, Smith S. A global optimisation method for robust affine registration of brain images. Med Image Anal. 2001;5:143–56.View ArticlePubMedGoogle Scholar
- Schulz H, Postma J, Dusschoten D, Scharr H, Behnke S. Plant root system analysis from MRI images. In: Csurka G, Kraus M, Laramee R, Richard P, Braz J, editors. Computer vision, imaging and computer graphics theory and application. Volume 359. Berlin Heidelberg: Springer; 2013. p. 411–25. Communications in Computer and Information Science.View ArticleGoogle Scholar
- Dijkstra EW. A note on two problems in connexion with graphs. Numer Math. 1959;1:269–71.View ArticleGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.