Food Research International 66 (2014) 132–139

Contents lists available at ScienceDirect

Food Research International

j ourna l homepage: www.e lsev ie r .com/ locate / foodres

Multi-scale characterization of pasta during cooking using microscopyand real-time magnetic resonance imaging

Diana Bernin a,1, Thomas Steglich b,c,d,⁎,1, Magnus Röding e, Annelie Moldin f, Daniel Topgaard g, Maud Langton d

a Swedish NMR Centre, University of Gothenburg, Göteborg, Swedenb SIK— The Swedish Institute for Food and Biotechnology, Göteborg, Swedenc Department of Chemical and Biological Engineering, Chalmers University of Technology, Göteborg, Swedend Department of Food Science, Swedish University of Agricultural Sciences (SLU), Uppsala, Swedene Department of Mathematical Sciences, Chalmers University of Technology, Göteborg, Swedenf Lantmännen Cerealia, Järna, Swedeng Department of Chemistry, Lund University, Lund, Sweden

⁎ Corresponding author at: SIK — The Swedish InstituGöteborg, Sweden. Tel.: +46 10 516 67 19.

E-mail address: Thomas.Steglich@sik.se (T. Steglich).1 These authors contributed equally.

http://dx.doi.org/10.1016/j.foodres.2014.09.0070963-9969/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 July 2014Accepted 8 September 2014Available online 16 September 2014

Keywords:SpaghettiSaltMRIGlutenStarchReal-time

Macroscopic properties of pasta, such as the texture, are formed during cooking by a complex interplay ofwater and heat with the structuring agents starch and gluten. The impact of the starch-to-gluten ratio onmicrostructure and water distribution in pasta was analyzed by a multi-scale approach combining magneticresonance imaging (MRI) and light microscopy. The cooking process and thus the water distribution wasmonitored non-invasively using 1H MRI in real-time with a temporal resolution of 45 s. Our MRI set-up allowedfollowing the water ingress by imaging the reduction of the uncooked core. The water ingress rate was neitherdependent on pasta composition nor on the presence of salt in the cooking media (0.7% NaCl). Starch-richsamples showed a more homogeneous water distribution in the gelatinized zone, which was mirrored in amore homogeneous microstructure. In contrast, gluten-rich samples showed both a heterogeneous waterdistribution and microstructure. Thus, the gluten content affected local water content in the gelatinized zonebut not the water ingress.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Most consumers determine the quality of cooked pasta based onappearance and texture properties (Marchylo, Dexter, & Malcolmson,2004). The texture is influenced by the two main components ofpasta, starch and gluten. Both components alter the microstructureduring the cooking as starch granules swell and disintegrate while glu-ten polymerizes (Resmini & Pagani, 1983). High protein content and acertain protein composition have been shown to correlate with desiredtexture parameters such as high firmness and low stickiness (Cubadda,Carcea, Marconi, & Trivisonno, 2007; Marchylo et al., 2004).

Together with starch and gluten content, the ingress and distribu-tion of water are important factors in defining the texture of pasta aswell (Horigane et al., 2006). An earlier study suggested that higherprotein content results in a slower water ingress into the spaghettitowards the center (Grzybowski & Donnelly, 1977), while otherscould not observe any differences (Cubadda et al., 2007).

te for Food and Biotechnology,

1H magnetic resonance imaging (MRI) is a non-invasive methodthat spatially resolves the amount and dynamics of water- andmacromolecule-protons. 1H MRI has been applied to monitor thewater ingress and distribution in pasta and noodle samples at differentcooking stages (Bonomi et al., 2012; Horigane et al., 2006; Kojima,Horigane, Nakajima, Yoshida, & Nagasawa, 2004; Lai & Hwang, 2004;McCarthy, Gonzalez, &McCarthy, 2002). In the aforementioned studies,pasta or noodle samples have been cooked for a definite period oftime and then removed from the water prior to MRI measurements.However, a more detailed analysis can be obtained from real-timemeasurements, i.e. acquiring MR images during the cooking. Mohoričet al. (2004) applied this concept to study the slow cooking processof rice kernels. We adapted the aforementioned method to monitorthe water ingress and the changes in microstructure in modelspaghetti throughout the cooking in real-time with a temporal resolu-tion of 45 s.

This study uses a multi-scale approach investigating themicrostruc-tural changes in pasta throughout the cooking dependent on (i) the rawmaterials used and (ii) the presence of salt in the cooking water. Thewater ingress was monitored non-invasively with an accuracy ofabout hundred micrometers in real-time using MRI and was confirmedby polarized light microscopy. MR parameter maps were correlatedwith light microscopy images on the micrometer scale to characterize

133D. Bernin et al. / Food Research International 66 (2014) 132–139

the extent of rawmaterial transformation and thus the microstructuralchanges in pasta.

2. Material and methods

2.1. Material

Durum semolina (carbohydrate content 77% (w/w), protein 15%(w/w) of dry matter) was supplied by Lantmännen Cerealia (Malmö,Sweden). Starch (carbohydrate 99% (w/w), protein 0% (w/w) of drymatter) and gluten powder (carbohydrate 8% (w/w), protein 86%(w/w) of dry matter) were supplied by Lantmännen Reppe (Lidköping,Sweden). Starch and gluten were derived from soft wheat. Sodiumchloride (NaCl) was purchased in a local supermarket. From here onall concentrations given below are in % (w/w) if not noted differently.

2.2. Pasta production

Spaghetti samples were produced on laboratory-scale with a widerange of starch-to-gluten ratios. The reference sample D100 was madeof 100% durum semolina and was comparable in composition to a stan-dard commercial pasta product. For the starch-rich pasta termedS40D60, 40% of the semolina was replaced equally with starch powder.Gluten-rich pasta was produced by replacing 20% (G20D80) and 40%(G40D60) semolina with gluten powder. The carbohydrate and proteincontents of the samples were estimated from known compositions andare listed in Table 1.

Batches of 500 g were processed in a lab-scale pasta machine(Edelweiss TR/75C, Italy; machine mixes and extrudes the dough).Water was added to the well-blended dry ingredients to achieve amoisture content of about 33%, while taking the moisture contents ofsemolina and the powders into account. The ingredients were mixedfor 15 min and the dough was extruded through a Teflon-coated die(spaghetti form, 1.5 mm diameter). Spaghetti was hung on racksand placed in a combi steamer oven (CCM, Rational, Germany). Theextruded spaghetti rested up to 20 min in ambient conditions dueto the handling during and after extrusion. The drying program ofZweifel, Handschin, Escher, and Conde-Petit (2003) was adapted asthe oven allowed only limited humidity control. Spaghetti was initiallydried for 30min at 40 °C and then for 120min at 50 °C. The temperaturewas step-wise increased to 90 °C within 30 min, kept at this tempera-ture for 30 min, reduced to 50 °C again and kept at 50 °C for 120 min.The relative humidity was kept at the 100% setting of the oven untilthe last drying step of 120 min where it was reduced to 50%. Spaghettiwas stored at room temperature.

2.3. Cooking conditions

Spaghetti samples were cooked for the analyses described inSections 2.4–2.6 either in distilled water (for MRI experiments ultra-pure water, MilliQ, Millipore was used) or distilled water containing0.7% NaCl (w/v). The salt concentration corresponds to a level common-ly chosen for sensory evaluations (Delcour et al., 2000).

Table 1Carbohydrate and protein content (% w/w of dry matter) as well as diameter of dried, uncooke

S40D60 D100

40% starch + 60% semolina 100% semolina

Carbohydrate 89.5 82.0Protein 08.3 13.9Dry diameter [mm] 1.50 ± 0.05 1.60 ± 0.05

2.4. Macroscopic properties

Single strands of spaghetti (25 ± 2 mm in length) were placed inglass tubes containing 10mL boiling water. The samples were removedafter 7.6min and 13.1min (corresponding to the time of the eighth andfifteenth image of the MRI series), blotted and weighed. Additionally,the length of the samples was measured before and after cookingusing a gauge with an accuracy of 0.02 mm. The weight increase as ameasure for water absorption as well as the length increase weredetermined as the mass and length ratio between the cooked and thedry sample, (Wi/Wi0) and (Li/Li0), respectively.

2.5. Real-time magnetic resonance imaging

MRI experiments were carried out on an 11.7 T Avance II spectrom-eter (Bruker, Germany) operating at a 1H resonance frequency of500 MHz. The magnet was fitted with a Bruker MIC-5 microimagingprobe giving a maximum gradient strength of 3 T m−1 in three orthog-onal directions. A 5mm radiofrequency coil designed to operate at tem-peratures up to 200 °C (EVT, Bruker, Germany) was used for excitationand detection of the 1H signal. The temperature inside the magnet wascontrolled by airflow. The heater close to the NMR tubewas set to 378 Kwhich yielded a water temperature of 90 °C in the tube after 1 min andreached the set temperature of 99 °C after 3 min. The tube (5mmouterdiameter) was filled 2 cm in height (~400 μL liquid). A glass sphere(3 mm in diameter) was placed at the bottom of the tube to avoidsuperheating of the cooking water. A 1 cm lengthwise piece of drypasta was placed 1.5 cm above the bottom of the tube and kept inplace with a plastic tube (inner diameter equal to the dry pasta's outerdiameter) that was pulled over the top of the pasta piece (furtherdetails are depicted in Fig. 1). The prepared tube was placed into thespectrometer at the cooking time zero. Then, the probe was tunedand matched and the signal was put on-resonance. The proceduretook about 1 min before the first experiment was initialized. 1H sliceselective RARE (rapid acquisition with relaxation enhancement)experiments with a duration of 31 s (plus 16 s for storing data betweeneach experiment) were repeated 15 times giving a total cooking time of13.5 min. Three slices with a thickness of 1 mm and a gap of 1 mmwere acquired using a CPMG sequence (Carr & Purcell, 1954;Meiboom & Gill, 1958) generating 8 echoes, which were spaced by amultiple of 3.66 ms. The field-of-view was set to 15 mm × 5 mm (96frequency- and 32 phase-encoded points), which gave an in-plane res-olution of 0.156 ∙ 0.156 mm2. One scan was acquired and the repetitiontime was set to 0.9 s to minimize T1-weighted signal arising from thepasta, while the cooking water was heavily T1-weighted and thus hadlower signal intensity (Mohorič et al., 2004). The last (16th) experimentlasted 4 min and allowed to estimate additionally T1 from three repeat-ed experiments with different repetition times tRep, being 0.9, 1.6 and5.0 s, respectively.

The acquired complex data was converted to a sequence of 2D imagesby Fourier transformation andmagnitude calculation. The image intensityI(t) as a function of echo time t is given by I(t)= I0exp(−t / T2), whereI0 is the signal intensity at t=0. T1 was estimated using I(tRep)= I0T1[1− exp(−tRep / T1)] where I0T1 is the non-weighted signal intensity. Foreach volume element in the image, the values of I0, I0T1, T1 and T2 wereestimated by regressing the equations above onto the experimental

d spaghetti.

G20D80 G40D60

20% gluten + 80% semolina 40% gluten + 60% semolina

67.1 52.928.6 43.71.65 ± 0.10 1.75 ± 0.10

Fig. 1.A sketch (left) of the set-up of the NMR tube during pasta cooking inside the spectrometer. The pasta piece is held in placewith a plastic tube. Illustrative description (right) explainsthe creation of masks used for radius estimation of the uncooked core. For radius estimation, the reader is referred to the text.

134 D. Bernin et al. / Food Research International 66 (2014) 132–139

data. The 16th experiment was used to estimate the extent of T1-weighting of I0 by calculating the corrected initial signal intensity I0Cas follows I0 / (1 − exp(−tRep / T1)) for tRep = 0.9 s. Magnitude noisewas treated as described in Steglich et al. (2014) and a thresholdvalue was used to avoid inaccurate MR parameter estimations. Thethreshold was calculated by creating a histogram of a large number ofnoise values for all runs, which was then fit to a Rayleigh distribution(Sijbers & den Dekker, 2004) and the value at probability P = 0.999was set as the threshold value.

The inner radius was estimated by creating a mask using 30% of themaximal I0 as a cutoff value (Fig. 1). The signal of the cookingwater wascut using 60% of the maximal signal intensity at the first echo time as acutoff value, as depicted in Fig. 1. The spaghetti-shaped pasta wasassumed to be circular and thus the inner radius was estimated by tak-ing the square root of the total amount of pixels of the created maskmultiplied by 0.1562 mm2/π. The inner radii were estimated for twoout of three slices. The data of the slice closest to the plastic tube holderwas discarded because of a deviating trend, which might have beencaused by temperature differences or air bubbles. S40D60 was difficultto evaluate as small air bubbles were found at the surface after cooking,which created magnetic susceptibility artifacts.

All processing and fitting were done using Matlab (Mathworks,USA). Based on the set of estimated values of T2, a probability distribu-tion was constructed using the built-in kernel density estimatorksdensity with a Gaussian kernel (Bowman & Azzalini, 1997).

2.6. Light Microscopy

The area of the central uncooked core was determined by polarizedlight microscopy (Del Nobile et al., 2003). Single spaghetti strands werecooked for fixed times, drained and immediately frozen to−20 °C. Thesamples were then cut to a thickness of 10 μm using a cryostat (Leica,Austria) and slices were analyzed without staining.

Sections of samples cooked for 7.6 min and 13.1 min were addition-ally studied bywide-field bright lightmicroscopy to visualize themicro-structure. The samples were stainedwith Light Green for 30min aswellaswith iodine in a Lugol's solution (2:1) for 30 s. In the resulting images,

proteins were stained in green, starch granules in blue/violet, amylosein blue and amylopectin in brown.

All samples were analyzed with an Eclipse Ni-U microscopeequipped with a Digital Sight DS-Fi2 camera and processed withthe software NIS-Elements BR (all Nikon, Japan). The software function‘n-points’ circle was used to manually determine the diameter of theuncooked core.

3. Results and discussion

The macroscopic water absorption and microstructure of gluten-and starch-rich spaghetti samples were analyzed for two or threecooking time points while the cooking process wasmonitored through-out the cooking in real-time with a temporal resolution of 45 s usingMRI.

We first discuss the results with regards to the rawmaterial compo-sition followed by the impact of salt.

3.1. Macroscopic properties

The analyzed samples varied in initial dry diameter (1.5–1.7 mm),which has been shown to influence the water absorption rate indepen-dent of the raw material used (Cafieri, Mastromatteo, Chillo, & DelNobile, 2010; Ogawa, Kobayashi, & Adachi, 2011). We normalizedtherefore the cooking times with the initial surface area of the samplesas proposed by Ogawa et al. (2011). Surface area strictly correlates withdiameter2, thus the presented unit for normalized cooking times is[time ∙ diameter−2].

Due to the limited amounts of available samples, weight and lengthincrease were determined for two cooking times only. Other studieshave shown that if there are differences in water absorption ratebetween samples, they are persistent throughout the cooking process(Cafieri et al., 2010; Ogawa & Adachi, 2013a). For both times measured,the starch-rich S40D60 seemed to absorb more water than the othersamples (Fig. 2). The reference and the two gluten-rich pasta samplesdid not show any difference. The same behavior was observed forthe length increase, with S40D60 increasing more than the three othersamples. Thus, only the sample with the lowest protein content of

Fig. 2. a) Relativeweight and b) relative length increase during the cooking indistilled (full symbols) and saltedwater (open symbols). Diameter normalizedwith the initial diameter; timenormalized with initial surface area (correlating with initial diameter2). Each symbol represents a separate measurement. c) Longitudinal slices showing proton density images for D100representing the length and diameter increase throughout the cooking.

135D. Bernin et al. / Food Research International 66 (2014) 132–139

8.3% deviated in water absorption rate, while the samples with a widerange of protein contents of 13.9% to 43.7% did not differ. The resultsmatch observations by Sozer and Kaya (2008) and Del Nobile, Baiano,Conte, and Mocci (2005). They found no influence of protein contenton water absorption rate for samples with a protein content of 11.7%to 15.5%. This indicates that there might be a threshold value for theprotein content above which the water absorption rate might be inde-pendent of the protein content.

The presence of salt in the water influenced neither the amount ofabsorbedwater nor the length increase. This is in agreementwith recentfindings of Ogawa and Adachi (2013a) who report no influence at asimilar NaCl concentration (0.5%). However, higher salt levels (5% hasbeen reported) decrease the amount of absorbed water (Ogawa &Adachi, 2013a; Sozer & Kaya, 2008).

3.2. Real-time magnetic resonance imaging

The chosen MRI set-up allowed studying the cooking process ofpasta with a temporal resolution of 45 s. I0 and T2 maps for a completecooking series are shown as an example for D100 cooked in distilled

Fig. 3. Time series of MRI measurement during the cooking exemplarily shown for D100 (timc) polarized light micrographs are shown for selected times. Note: The T1-weighted I0 and T2sample strands.

and salted water (Fig. 3). The two slices (only middle slice shown)were consistent and showed a similar cooking progress. The outercontour of the pastawas visible in the firstmaps (~2.1 min), which con-firmed the water ingress and the beginning of starch gelatinization dueto the cooking process. The gray region in the middle, which decreasedwith increasing the cooking time, represent the ungelatinized core ofthe pasta, as shown by polarized light micrographs (Fig. 3c). At thispoint, we want to emphasize that throughout the cooking, the pastaincreased significantly in radius and length (see 3.1). Because of thelength increase, the MR parameter maps reflect different parts of thepasta depending on cooking time.

I0mapsmay be used as a proxy for thewater concentration (Steglichet al., 2014) if the signal has relaxed back to the equilibrium during therepetition time, i.e. the signal is not T1-weighted. To estimate the degreeof T1-weighting of the I0 maps, T1 was measured at the end of eachcooking series (Fig. 4a). The estimated T1 maps varied between thesamples and S40D60 had on average the largest T1 values. As a result,I0C (corrected for T1) and I0 differed in particular for S40D60 implyingT1-weighting in contrast to gluten-rich pasta (Fig. 4b + c). The I0Cmaps indicate that S40D60 absorbed the most water, which agrees

es show average time point per image). a) T1-weighted I0, b) T2 maps. For comparison,maps and the polarized light micrographs are derived from separate, individually cooked

Fig. 4.MRparametermaps representing a) T1, b) T1-weighted initial signal intensity (T1-w I0) aswell as c) corrected initial signal intensity (I0C). Spaghettiwas cooked for 15min indistilledor salted water.

136 D. Bernin et al. / Food Research International 66 (2014) 132–139

with thewater absorption resultsmeasured byweight increase. In gener-al, all pasta samples showed more or less T1-weighted I0 values close tothewater interface where I0C represent the largest values. The uncorrect-ed, T1-weighted I0 parametermaps, althoughnot reflecting the truewaterconcentration, allowed nevertheless comparing the temporal changesthroughout the cooking with and without salt in a qualitative manner.

Fig. 5a + b shows T1-weighted I0 and T2 maps for selected times forall pasta samples cooked in distilled or saltedwater. The cooking processproceeded in the same pace for all pasta samples independently on thepresence of salt in the cooking water at the observed temporal andspatial resolution. I0 values tend to decrease close to the ungelatinizedcore, which is true for all model pasta. The distribution of I0 valuesis rather homogeneous throughout the cooked regions, in particularfor the S40D60 and D100. For the gluten-rich samples G20D80 andG40D80, the distribution of I0 values seemed to bemore heterogeneous.This indicates that the polymerization of gluten in combination withwater absorption occurs not as homogeneous in high protein pastacompared to pasta containing more starch.

The T2 maps are characterized by larger T2 values (yellow) close tothewater–pasta interface and shorter T2 (red) towards the ungelatinizedcore. The smaller T2 values arose from lowwater content, whichwas alsoindicated by lower I0 values. This trend was found for all model pasta.No obvious difference could be found comparing the MR maps (I0 andT2) obtained from pasta cooked in distilled and in salted water. This isconsistent with Kojima, Horigane, Yoshida, Nagata, and Nagasawa(2001) who reported no effect of a 0.69% NaCl solution on T2 of cookedwheat gel samples (compared to cooking in distilled water).

Visualizing the values of the T2 maps in histograms allowed for asemi-quantitative comparison of thepasta samples.We chose smoothed

histograms that present probability density estimates (PDEs) to illus-trate the differences (Fig. 5c) occurring during the cooking. PDEs of T2are presented for the same cooking times as the T2 maps shown inFig. 5b. For each cooking time, the PDEs did not vary much with saltconcentrationwhile differenceswere found between the pasta samples.A difference in the amount of gluten in G20D80 and G40D60 yielded analmost identical PDE characterized by a narrow distribution with theshortest T2. PDEs for D100 were located on the gluten-rich PDEs orbetween gluten-rich and S40D60 PDEs. The width of the PDEs seemedto broaden with increasing starch content and the PDE of S40D60yielded the broadest distribution with the longest T2 values. This is ingood agreementwith the T1 and I0Cmapsproving that S40D60 absorbedlarger amounts of water in contrast to gluten-rich pasta. Increasing T2values with decreasing protein content was also reported for variousstages of cooking in noodles (Kojima et al., 2004).

3.3. Water absorption kinetics

The decreasing radius of the ungelatinized core over cooking timecan be seen as a measure for water absorption kinetics. The estimatedinner radius from T1-weighted I0 maps represents the average of twoslices as a function of cooking time (Fig. 5d). Deviations within thesame sample were small until the inner radius had reached about 30%of the initial radius. At this point in time, the standard deviationsincreased as the uncooked core is represented only by few voxelsresulting in a large relative error. Additionally, the samples showed asomewhat heterogeneous inner structure with some small, randomlydistributed air inclusions. This complicated the evaluation (a representa-tive air inclusion can be seen in the polarized light micrograph for

Fig. 5. MR parameter maps presenting a) T1-weighted I0 and b) T2 for selected times during the cooking in distilled and in salted water. c) Probability density estimates of T2 maps.d) Decrease of ungelatinized, inner core over time during the cooking as a measure for the water ingress rate. Diamonds, determined by MRI; circles, determined by polarized lightmicroscopy (PLM); full symbols, distilled water; open symbols, salted water. Error bars represent standard deviation. The horizontal line at 30% of the initial diameter is an aid to theeye. Values below this threshold are regarded as uncertain as the pasta samples were not homogeneous and showed especially in the core some air inclusions.

137D. Bernin et al. / Food Research International 66 (2014) 132–139

5.2min in Fig. 3). The inner radii determined byMRI agreedwell with theradii determined by polarized light microscopy. Again, the uncertaintyincreased when the radius was below 30% of the initial value. Still, thegeneral pattern of a linear decrease at an early cooking stage and

accelerated decrease at a later cooking stage is consistent with datareported by Del Nobile, Buonocore, Panizza, and Gambacorta (2003).

The data indicated that neither saltedwater (0.7% NaCl) nor the pro-tein content of the samples influence the water ingress rate (Fig. 5d).

138 D. Bernin et al. / Food Research International 66 (2014) 132–139

Others have seen little to no influence of protein content as well(Cubadda et al., 2007; Del Nobile et al., 2005; Kojima et al., 2004); atleast as far as high temperature drying is concerned. Differences mayapply at low-temperature drying (Cubadda et al., 2007; Resmini &Pagani, 1983). In addition to the temperature, the behavior of thewater absorption might be influenced as well by other factors duringdrying such as drying time or relative humidity.

Although the water ingress rate was independent on the proteincontent, S40D60 absorbedmorewater in total. Hence, starch gelatiniza-tion defines the water ingress rate, whereas in gelatinized zones thelocal water content is influenced by the protein content. Ogawa andAdachi (2013b) also showed that starch gelatinization regulates thewater ingress into the uncooked pasta regions.

3.4. Light Microscopy

All samples showed gradual changes inmicrostructure from the sur-face to the core of the spaghetti (Fig. 6). Furthermore, samples cooked in

Fig. 6. Representative parts of spaghetti cross-sections cooked for 7:36min and 13:06min in a)solution. Protein appeared green, starch blue/violet. Small unstained areas show fiber structurethe references to color in this figure legend, the reader is referred to the web version of this ar

distilled water showed at 7.6 min a distinct outer zone of strongly swol-len, partly disintegrated and fused starch granules. Some amyloseleaching was visible (can be seen in the image as dark blue), beingmost pronounced in S40D60. The degree of starch granule swellingdecreased towards the core; but to different extents. In S40D60 starchgranules were visibly swollen in the whole cross-section in contrastto the other samples, which showed a more heterogeneous swellingpattern. While some starch granules did swell in the core zone almostto the same extent as S40D60, other granules appeared unswollenand embedded in the protein matrix. It seems as if higher gluten con-centrations and thus denser gluten network partly hindered starchgranule swelling, but not starch gelatinizing (as presented in Fig. 5d).At 13.1 min, the aforementioned microstructural changes progressedtowards the core. S40D60 and D100 showed partly an open structureat the surface due to disintegrated starch granules. G20D80 andG40D60, conversely, still showed a continuous proteinmatrix, althoughstarch granules disintegrated as well. Again, a denser protein networkmight limit the swelling capabilities of starch. Alternatively, themedian

distilled and b) saltedwater. The sectionswere stainedwith Light Green and Lugol's iodines; large unstained areas show air inclusions entrapped in thematrix. (For interpretation ofticle.)

139D. Bernin et al. / Food Research International 66 (2014) 132–139

distance between starch granules increases with increasing proteincontent, which makes it more unlikely that neighboring granules canswell to such an extent that they fuse.

The microstructure of samples cooked in salted water did not differmuch from samples cooked in distilled water. However, the proteinmatrix seemed to be more intact in the region close to the surfacecompared to samples cooked in distilledwater. This might be explainedby the protein stabilizing effect of NaCl due to increasing hydrophobicinteractions between gluten units (Peressini, Sensidoni, Pollini, &Cindio, 2000; Sozer & Kaya, 2008).

The micrographs and T2 maps were in good agreement. T2 mapsshowed the highest T2 values in the surface zone and this zone increasedwith cooking time, which can be related to the zone of strongly swollenand partly disintegrated starch granules. S40D60 showed the largest T2values and had an open structure at 13.1min. This open structure at thesurface could explain that S40D60 had the highest water concentrationat the surface (Fig. 4c), which correlated with the large T2 values.Furthermore, the stronger starch granule swelling of S40D60 through-out the cross-sections compared to the other samples matched higherT2 values found in the core. Finally, the more homogeneous granuleswelling in S40D60 agreed with the narrower PDE distribution, whilethe broader PDE distribution of G20D80 and G40D60 might be due totheir more heterogeneous swelling (Fig. 5c).

4. Conclusions

The presented real-time MRI approach facilitates a method to mon-itor the cooking process of pasta.Microscopy, although time-consumingand invasive, is complementary to MRI and enabled a detailed charac-terization of themicrostructure on themicrometer to millimeter lengthscales.

Remarkably, the pace of thewater ingresswas neither dependent onthe pasta composition nor the addition of salt (0.7%NaCl) in the cookingwater although the micrographs revealed significant differences in themicrostructure. Taken MR parameter maps and micrographs together,an increase in gluten content leads to a more heterogeneous waterdistribution andmicrostructure. The results suggest that above a certainprotein content, the microstructure, water uptake and swelling as wellas water distribution are similar. Thus, the gluten content affects thelocal water content in the cooked (gelatinized) zone but not the wateringress.

Acknowledgments

The project is a part of the VINNExcellence Centre SuMoBiomaterials(Supermolecular Biomaterials — Structure dynamics and properties)and is financed by Lantmännen Research Foundation (Project number396). Daniel Topgaard is financed by the Swedish Research Council VR(grant numbers 2009-6794 and 2011-4334).

References

Bonomi, F., D'Egidio, M. G., Iametti, S., Marengo, M., Marti, A., Pagani, M.A., et al. (2012).Structure–quality relationship in commercial pasta: A molecular glimpse. FoodChemistry, 135, 348–355. http://dx.doi.org/10.1016/j.foodchem.2012.05.026.

Bowman, A. W., & Azzalini, A. (1997). Applied smoothing techniques for data analysis: Thekernel approach with s-plus illustrations (1st ed.). Oxford: Oxford University Press.

Cafieri, S., Mastromatteo, M., Chillo, S., & Del Nobile, M.A. (2010). Modeling the mechan-ical properties of pasta cooked at different times. Journal of Food Engineering, 100(2),336–342. http://dx.doi.org/10.1016/j.jfoodeng.2010.04.019.

Carr, H. Y., & Purcell, E. M. (1954). Effects of diffusion on free precession in nuclearmagnetic resonance experiments. Physical Review, 94(3), 630–638. http://dx.doi.org/10.1103/PhysRev.94.630.

Cubadda, R. E., Carcea, M., Marconi, E., & Trivisonno, M. C. (2007). Influence of glutenproteins and drying temperature on the cooking quality of durum wheat pasta.Cereal Chemistry, 84(1), 48–55. http://dx.doi.org/10.1094/CCHEM-84-1-0048.

Del Nobile, M.A., Baiano, A., Conte, A., & Mocci, G. (2005). Influence of protein content onspaghetti cooking quality. Journal of Cereal Science, 41(3), 347–356. http://dx.doi.org/10.1016/j.jcs.2004.12.003.

Del Nobile, M.A., Buonocore, G. G., Panizza, A., & Gambacorta, G. (2003). Modeling thespaghetti hydration kinetics during cooking and overcooking. Journal of FoodScience, 68(4), 1316–1323. http://dx.doi.org/10.1111/j.1365-2621.2003.tb09645.x.

Del Nobile, M.A., Fares, C., Conte, A., Martinez-Navarrete, N., Mocci, G., & La Notte, E.(2003). Influence of raw materials and processing conditions on spaghetti hydrationkinetic during cooking and overcooking. Cereal Chemistry, 80(5), 601–607. http://dx.doi.org/10.1094/CCHEM.2003.80.5.601.

Delcour, J. A., Vansteelandt, J., Hythier, M. -C., Abécassis, J., Sindic, M., & Deroanne, C.(2000). Fractionation and reconstitution experiments provide insight into the roleof gluten and starch interactions in pasta quality. Journal of Agricultural and FoodChemistry, 48(9), 3767–3773. http://dx.doi.org/10.1021/jf991051m.

Grzybowski, R. A., & Donnelly, B. J. (1977). Starch gelatinization in cooked spaghetti.Journal of Food Science, 42(5), 1304–1305.

Horigane, A. K., Naito, S., Kurimoto, M., Irie, K., Yamada, M., Motoi, H., et al. (2006).Moisture distribution and diffusion in cooked spaghetti studied by NMR imagingand diffusion model. Cereal Chemistry, 83(3), 235–242. http://dx.doi.org/10.1094/CC-83-0235.

Kojima, T. I., Horigane, A. K., Nakajima, H., Yoshida, M., & Nagasawa, A. (2004). T2 map,moisture distribution, and texture of boiled Japanese noodles prepared from differenttypes of flour. Cereal Chemistry, 81(6), 746–751. http://dx.doi.org/10.1094/CCHEM.2004.81.6.746.

Kojima, T. I., Horigane, A. K., Yoshida, M., Nagata, T., & Nagasawa, A. (2001). Change in thestatus of water in Japanese noodles during and after boiling observed by NMR microimaging. Journal of Food Science, 66(9), 1361–1365. http://dx.doi.org/10.1111/j.1365-2621.2001.tb15215.x.

Lai, H. -M., & Hwang, S. -C. (2004). Water status of cooked white salted noodles evaluatedby MRI. Food Research International, 37(10), 957–966. http://dx.doi.org/10.1016/j.foodres.2004.06.008.

Marchylo, B.A., Dexter, J. E., & Malcolmson, L. J. (2004). Improving the texture of pasta. InD. Kilcast (Ed.), Texture in food, volume 2: Solid foods (pp. 475–500). CRC Press: BocaRaton.

McCarthy, K. L., Gonzalez, J. J., & McCarthy, M. J. (2002). Change inmoisture distribution inlasagna pasta post cooking. Journal of Food Science, 67(5), 1785–1789. http://dx.doi.org/10.1111/j.1365-2621.2002.tb08723.x.

Meiboom, S., & Gill, D. (1958). Modified spin-echo method for measuring nuclear relaxa-tion times. The Review of Scientific Instruments, 29, 688–691. http://dx.doi.org/10.1063/1.1716296.

Mohorič, A., Vergeldt, F. J., Gerkema, E., de Jager, A., van Duynhoven, J. P.M., van Dalen, G.,et al. (2004). Magnetic resonance imaging of single rice kernels during cooking.Journal of Magnetic Resonance, 171(1), 157–162. http://dx.doi.org/10.1016/j.jmr.2004.08.013.

Ogawa, T., & Adachi, S. (2013a). Effect of salts on the water sorption kinetics of driedpasta. Bioscience, Biotechnology, and Biochemistry, 77(2), 249–252. http://dx.doi.org/10.1271/bbb.120639.

Ogawa, T., & Adachi, S. (2013b). Measurement of moisture profiles in pasta duringrehydration based on image processing. Food and Bioprocess Technology, 7(5),1465–1471. http://dx.doi.org/10.1007/s11947-013-1156-y.

Ogawa, T., Kobayashi, T., & Adachi, S. (2011). Water sorption kinetics of spaghetti atdifferent temperatures. Food and Bioproducts Processing, 89(2), 135–141. http://dx.doi.org/10.1016/j.fbp.2010.04.004.

Peressini, D., Sensidoni, A., Pollini, C. M., & Cindio, B. (2000). Rheology of wheat doughs forfresh pasta production: Influence of semolina-flour blends and salt content. Journal ofTexture Studies, 31(2), 163–182. http://dx.doi.org/10.1111/j.1745-4603.2000.tb01415.x.

Resmini, P., & Pagani, M.A. (1983). Ultrastructure studies of pasta. A review. FoodMicrostructure, 2(1), 1–12.

Sijbers, J., & den Dekker, A. J. (2004). Maximum likelihood estimation of signal amplitudeand noise variance from MR data. Magnetic Resonance in Medicine, 51(3), 586–594.http://dx.doi.org/10.1002/mrm.10728.

Sozer, N., & Kaya, A. (2008). The effect of cooking water composition on textural andcooking properties of spaghetti. International Journal of Food Properties, 11(2),351–362. http://dx.doi.org/10.1080/10942910701409260.

Steglich, T., Bernin, D., Röding, M., Nydén, M., Moldin, A., Topgaard, D., et al. (2014).Microstructure and water distribution of commercial pasta studied by microscopyand 3D magnetic resonance imaging. Food Research International, 62, 644–652.http://dx.doi.org/10.1016/j.foodres.2014.04.004.

Zweifel, C., Handschin, S., Escher, F., & Conde-Petit, B. (2003). Influence of high-temperature drying on structural and textural properties of durum wheat pasta.Cereal Chemistry, 80(2), 159–167. http://dx.doi.org/10.1094/CCHEM.2003.80.2.159.