microbial and swelling effects on pore size distribution in humous soil samples

1
IMF [4]. The simulations were performed with MATLAB on a typical desktop PC (Pentium D processor, 2.78 GHz). Typical results are presented in Figs. 1 and 2. Calculations were made over the magnetic field range from 2.11 to 21.1 T. The results obtained for 1 H nuclei were analyzed and compared with real NMR experiments. We show that the field dependence of line width contains information about the geometry of porous medium. This work is financially supported by the Knut and Alice Wallenberg Foundation and the Swedish Research Council VR. [1] Fatkullin FN, Sov. Phys. JETP, 74, 833, 1992. [2] Song YQ. Concepts Magn Reson A 2003;18A(2):97. [3] Archipov VR, Romanova EE, Sagidullin IA, Skirda VD. Appl Magn Reson 2005;29:481. [4] Mayer C. Prog Nucl Magn Reson Spectrosc 2002;40:307. doi:10.1016/j.mri.2007.01.092 Microbial and swelling effects on pore size distribution in humous soil samples F. Jaeger a , E. Grohmann b , G. Schaumann a a Dept. Environmental Chemistry, University of Technology Berlin, Sekr. KF 3, Strasse des 17. Juni 135, D-10623 Berlin, Germany, [email protected], b Dept. Environmental Microbiology, Universi- ty of Technology Berlin, Sekr. FR 1-2, Franklinstrasse 28/29, D-10587 Berlin, Germany Top soil layers are influenced by strong changes in moisture, which affect sorption and transport processes for e.g. pollutants. 1 H nuclear magnetic resonance ( 1 H-NMR) relaxometry may be used as a method to determine water uptake characteristics of soils, gaining information about water distribution and mobility as well as pore size distribution. Recent NMR studies in humus soil samples revealed relevance of swelling and wetting processes of soil organic matter (SOM), as well as microbial influences on 1 H-NMR relaxometry. The objective of this investigation was to achieve first indications to which extent microbial activity and quantity of bacteria affect relaxation time distribution during rewetting of humus soil samples. We used a humus forest soil sample and added cellobiose to selected samples to enhance microbial activity (treated samples). Treated and untreated samples were moistened to 43% water content. At several points of time during 3 weeks, transverse relaxation time distributions were at 2 MHz (Maran 2, Resonance, UK). Microbial respiratory activity was determined with a Respirocond system (Nordgreen Innovation, Norway) detecting conductiv- ity changes of KOH solution caused by CO 2 absorption. Total cell counts were determined by DAPI staining (4V,6-diamidino-2-phenylindol) after bacterial extraction with a Na4O7P2 solution. The initial relaxation time distribution of all samples showed up to three peaks (see figure). During hydration, the number of peaks decreased, and the peaks revealed significant movement towards lower relaxation times. Microbial respiratory activities were highest after 1–3 days of hydration, with values 2–15 times higher in the treated as compared to the untreated samples. Total cell counts increased in all samples from 1 to 5 109 cells per gram. We assume that changes in the pore size distribution and in spin relaxation mechanism are responsible for the shifts in the relaxation time distribution. This can be due to wetting and swelling of SOM and increasing numbers of paramagnetic centres on surfaces and in the bulk phase. In addition, production and release of extracellular polymeric substances and bacterial association within biofilm may form new pore systems and reduce interparticular pore diameter. doi:10.1016/j.mri.2007.01.093 Noise reduction in magnetic resonance images by Wavelet transforms: an application to the study of capillary water absorption in sedimentary rocks T. Schillaci a , R. Barraco a , M. Brai a , G. Raso a , V. Bortolotti b , M. Gombia b , P. Fantazzini b a Dip. di Fisica e Tecnologie Relative, Universita ` di Palermo, Italy, [email protected], b Dip. DICMA and Dip. di Fisica, Universita ` di Bologna, Italy Magnetic resonance imaging (MRI) is a powerful technique to study capillary water absorption kinetics in sedimentary rocks [1]. However, the noise in the images can limit the correct identification and the quantitative measurement of the average height reached by the wetting front inside the porous material where imbibition occurs. Therefore, denoising methods can be applied to improve the image quality for a more accurate analysis, without the disadvantages of longer acquisition times [2]. This study attempts to improve the signal-to-noise ratio of the images acquired by MRI on a sedimentary rock (Pietra di Lecce) using a wavelet- based thresholding technique. The idea is to average some slightly different discrete wavelet transforms (DWT), called e -decimated DWT, to define the stationary wavelet transforms (SWT). We denote by f = g+e the measurements of the image g corrupted by an additive zero-mean white Gaussian noise field e with standard deviation r. Applying the SWT to the MRI images, nonsignificant wavelet coefficients below a preset threshold value are discarded as noise, and the image is reconstructed from the remaining significant coefficients. The algorithm for the SWT denoising of the image f was implemented by the wavelet toolbox of the MATLAB software. In order to evaluate the improvement of the image quality, we compute the mean-to-standard deviation ratio (MSR) and the contrast-to-noise ratio (CNR), defined by MSR ¼ l d r d ; CNR ¼ jl d l u j ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0:5 r 2 d þ r 2 u q where l d and r d are the mean value and the standard deviation computed in the desired region of interest (DROI) and l u and r u are computed in that undesired ROI (UROI) such as a window or background. Fig. 1 shows the result of the wavelet denoising application. Fig. 1 The DROI and UROI used to compute the MSR and CNR indexes are highlighted in the figure on the left, while on the middle and on the right, the denoised images obtained using the Bior 3.7 and Haar wavelet functions, respectively, are shown. Abstracts / Magnetic Resonance Imaging 25 (2007) 544 – 591 581

Upload: f-jaeger

Post on 21-Jun-2016

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Microbial and swelling effects on pore size distribution in humous soil samples

IMF [4]. The simulations were performed with MATLAB on a typical

desktop PC (PentiumD processor, 2.78GHz). Typical results are presented in

Figs. 1 and 2.

Calculations were made over the magnetic field range from 2.11 to 21.1 T.

The results obtained for 1H nuclei were analyzed and compared with real

NMR experiments. We show that the field dependence of line width

contains information about the geometry of porous medium.

This work is financially supported by the Knut and Alice Wallenberg

Foundation and the Swedish Research Council VR.

[1] Fatkullin FN, Sov. Phys. JETP, 74, 833, 1992.

[2] Song YQ. Concepts Magn Reson A 2003;18A(2):97.

[3] Archipov VR, Romanova EE, Sagidullin IA, Skirda VD. Appl Magn

Reson 2005;29:481.

[4] Mayer C. Prog Nucl Magn Reson Spectrosc 2002;40:307.

doi:10.1016/j.mri.2007.01.092

Microbial and swelling effects on pore size distribution in humous soil

samples

F. Jaegera, E. Grohmannb, G. Schaumanna

aDept. Environmental Chemistry, University of Technology Berlin,

Sekr. KF 3, Strasse des 17. Juni 135, D-10623 Berlin, Germany,

[email protected], bDept. Environmental Microbiology, Universi-

ty of Technology Berlin, Sekr. FR 1-2, Franklinstrasse 28/29, D-10587

Berlin, Germany

Top soil layers are influenced by strong changes in moisture, which affect

sorption and transport processes for e.g. pollutants. 1H nuclear magnetic

resonance (1H-NMR) relaxometry may be used as a method to determine

water uptake characteristics of soils, gaining information about water

distribution and mobility as well as pore size distribution. Recent NMR

studies in humus soil samples revealed relevance of swelling and wetting

processes of soil organic matter (SOM), as well as microbial influences on1H-NMR relaxometry.

The objective of this investigation was to achieve first indications to which

extent microbial activity and quantity of bacteria affect relaxation time

distribution during rewetting of humus soil samples. We used a humus

forest soil sample and added cellobiose to selected samples to enhance

microbial activity (treated samples). Treated and untreated samples were

moistened to 43% water content. At several points of time during 3 weeks,

transverse relaxation time distributions were at 2 MHz (Maran 2,

Resonance, UK). Microbial respiratory activity was determined with a

Respirocond system (Nordgreen Innovation, Norway) detecting conductiv-

ity changes of KOH solution caused by CO2 absorption. Total cell counts

were determined by DAPI staining (4V,6-diamidino-2-phenylindol) after

bacterial extraction with a Na4O7P2 solution.

The initial relaxation time distribution of all samples showed up to three

peaks (see figure). During hydration, the number of peaks decreased, and

the peaks revealed significant movement towards lower relaxation times.

Microbial respiratory activities were highest after 1–3 days of hydration,

with values 2–15 times higher in the treated as compared to the untreated

samples. Total cell counts increased in all samples from 1 to 5 � 109 cells

per gram.

We assume that changes in the pore size distribution and in spin relaxation

mechanism are responsible for the shifts in the relaxation time distribution.

This can be due to wetting and swelling of SOM and increasing numbers of

paramagnetic centres on surfaces and in the bulk phase. In addition,

production and release of extracellular polymeric substances and bacterial

association within biofilm may form new pore systems and reduce

interparticular pore diameter.

doi:10.1016/j.mri.2007.01.093

Noise reduction in magnetic resonance images by Wavelet transforms:

an application to the study of capillary water absorption in

sedimentary rocks

T. Schillacia, R. Barracoa, M. Braia, G. Rasoa, V. Bortolottib, M. Gombiab,

P. Fantazzinib

aDip. di Fisica e Tecnologie Relative, Universita di Palermo, Italy,

[email protected], bDip. DICMA and Dip. di Fisica, Universita di

Bologna, Italy

Magnetic resonance imaging (MRI) is a powerful technique to study

capillary water absorption kinetics in sedimentary rocks [1]. However, the

noise in the images can limit the correct identification and the quantitative

measurement of the average height reached by the wetting front inside the

porous material where imbibition occurs. Therefore, denoising methods can

be applied to improve the image quality for a more accurate analysis,

without the disadvantages of longer acquisition times [2].

This study attempts to improve the signal-to-noise ratio of the images

acquired by MRI on a sedimentary rock (Pietra di Lecce) using a wavelet-

based thresholding technique. The idea is to average some slightly

different discrete wavelet transforms (DWT), called e-decimated DWT,

to define the stationary wavelet transforms (SWT). We denote by f =g+ethe measurements of the image g corrupted by an additive zero-mean

white Gaussian noise field e with standard deviation r. Applying the SWT

to the MRI images, nonsignificant wavelet coefficients below a preset

threshold value are discarded as noise, and the image is reconstructed from

the remaining significant coefficients. The algorithm for the SWT

denoising of the image f was implemented by the wavelet toolbox of

the MATLAB software.

In order to evaluate the improvement of the image quality, we compute the

mean-to-standard deviation ratio (MSR) and the contrast-to-noise ratio

(CNR), defined by

MSR ¼ ld

rd

; CNR ¼ jld � lujffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi0:5 r2

d þ r2u

� �q

where ld and rd are the mean value and the standard deviation computed in

the desired region of interest (DROI) and lu and ru are computed in that

undesired ROI (UROI) such as a window or background. Fig. 1 shows the

result of the wavelet denoising application.

Fig. 1 The DROI and UROI used to compute the MSR and CNR indexes

are highlighted in the figure on the left, while on the middle and on the

right, the denoised images obtained using the Bior 3.7 and Haar wavelet

functions, respectively, are shown.

Abstracts / Magnetic Resonance Imaging 25 (2007) 544 – 591 581