foods

Article Weakening Pin Bone Attachment in Fillets Using High-Intensity Focused Ultrasound

Martin H. Skjelvareid 1,* ID , Svein Kristian Stormo 1, Kristín Anna Þórarinsdóttir 2 and Karsten Heia 1

1 Department of Industry, Nofima AS, P.O. Box 6122, 9291 Tromsø, Norway; svein.stormo@nofima.no (S.K.S.); karsten.heia@nofima.no (K.H.) 2 Marel , Austurhraun 9, 210 Gardabaer, Iceland; [email protected] * Correspondence: mhs@nofima.no; Tel.: +47-77-62-92-14

Received: 22 August 2017; Accepted: 8 September 2017; Published: 18 September 2017

Abstract: High Intensity Focused Ultrasound (HIFU) can be used for the localized heating of biological tissue through the conversion of sound waves into heat. Although originally developed for human medicine, HIFU may also be used to weaken the attachment of pin bones in fish fillets to enable easier removal of such bones. This was shown in the present study, where a series of experiments were performed on HIFU phantoms and fillets of cod and salmon. In thin objects such as fish fillets, the heat is mainly dissipated at the surfaces. However, bones inside the fillet absorb ultrasound energy more efficiently than the surrounding tissue, resulting in a “self-focusing” heating of the bones. Salmon skin was found to effectively block the ultrasound, resulting in a significantly lower heating effect in fillets with skin. Cod skin partly blocked the ultrasound, but only to a small degree, enabling HIFU treatment through the skin. The treatment of fillets to reduce the pin bone attachment yielded an average reduction in the required pulling force by 50% in cod fillets with skin, with little muscle denaturation, and 72% in skinned fillets, with significant muscle denaturation. Salmon fillets were treated from the muscle side of the fillet to circumvent the need for penetration through skin. The treatment resulted in a 30% reduction in the peak pulling force and 10% reduction in the total pulling work, with a slight denaturation of the fillet surface.

Keywords: fish fillet; pin bones; HIFU; high intensity focused ultrasound; cod; salmon

1. Introduction Fish fillets are made by cutting away the muscle on each side of the central backbone of the fish and trimming away the fins. In many fish species, so-called pin bones remain in the flesh after filleting and have to be removed to produce the desired boneless product. In most cases, pin bone removal is still a manual process, making it costly and time-consuming. In this paper, we consider the pin bones of two commercially important species, Atlantic cod (Gadus Morhua) and Atlantic salmon (Salmo salar), and study high-intensity focused ultrasound as a possible technology for weakening the attachment of the pin bones to the surrounding muscle. This could potentially enable automated pin bone removal and thereby improve fish processing efficiency. Fish muscle consists of a number of segments, called myomeres, separated by layers of connective tissue (CT), called myosepta. The smaller muscle segments converge at a thick layer of CT called the horizontal septum, dividing the muscles on the back of the fish (epaxial muscles) from the muscles on the belly side (hypaxial muscles). The horizontal septum can further be described as an array of tendons connecting the backbone to the skin [1]. In cod, the pin bones are embedded in this lattice of tendons in the horizontal septum, whereas in salmon, the pin bones are embedded in the epaxial muscles [2,3]. From a fish processing perspective, this means that for cod, the pin bones are positioned

Foods 2017, 6, 82; doi:10.3390/foods6090082 www.mdpi.com/journal/foods Foods 2017, 6, 82 2 of 15

Foods 2017, 6, 82 2 of 15 between the loin and belly parts of the fillet, while for salmon, the pin bones are positioned within the loin part. This is illustrated in Figure1. The typical number of pin bones per fillet is approximately 17 positioned within the loin part. This is illustrated in Figure 1. The typical number of pin bones per for cod [2] and 30 for salmon [4,5]. is approximately 17 for cod [2] and 30 for salmon [4,5].

(a) (b)

Figure 1. Schematic drawing of fish fillets showing the position of the pin bones. (a) Cod fillet. (b) Salmon Figure 1. Schematic drawing of fish fillets showing the position of the pin bones. (a) Cod fillet. fillet. (b) Salmon fillet.

Salmon also differs from cod in terms of the chemical composition. It is a fatty fish species, meaningSalmon that also lipids differs are fromstored cod within in terms the of musculature, the chemical mainly composition. in the It CT is abetween fatty fish muscle species, segments meaning [6].that Thus, lipids arethe stored microstructure within the of musculature, the CT is different, mainly in which the CT may between influence muscle the segments strength [6 of]. Thus, bone athettachments microstructure [7]. of the CT is different, which may influence the strength of bone attachments [7]. These anatomical differences between the two species may partly explain why higher forces are needed to pull out the bones from cod compared to salmon [[2,4,5].2,4,5]. In both species, the strength of the pin bones’ attachment to surrounding tissue tissuess is reduced by post mortem ageing of the fillet.fillet. This mainly occurs though enzymatic activity in the muscle, which causes mild degradation of the CT CT [7]. [7]. Usually, the pin bones in cod filletsfillets are removedremoved by cutting out a strip containing the bones. In salmon, however,however, sincesince the the pin pin bones bones are are located located within within the the high-value high-value loin partloin ofpart the of fillet, the thefillet, bones the bonesare usually are usually pulled out.pulled This out. minimizes This minimizes weight loss weight from loss the f loinrom and the ensures loin and a visuallyensures aappealing visually appealingintact fillet. intact For salmon, fillet. For bone salmon, removal bone is normallyremoval is done normally after rigor done mortis after rigor when mortis bone attachments when bone attachmentsare partially are degraded partially [7], degraded as the risk [7], of as bone the breakagerisk of bone and breakage failure in and bone failure removal in bone is lower removal than inis pre-rigorlower than fish. in pre-rigor fish. The CT of fish fish is heat heat-labile-labile and denatures at lower temperatures than CT in products from land animals [8]. [8]. The major constituents of of the CT are proteoglycans and fibrous fibrous proteins, with collagen being thethe mostmost abundantabundant protein protein [7 [7].]. The The thermal thermal properties properties of collagenof collagen and and other other muscle muscle proteins proteins have havebeen studied been studied by differential by differential scanning scanning calorimetry. calorimetry. The thermal The transition thermal of transition cod collagen of cod appears collagen as a ◦ appearssmall peak as a in small the endothermic peak in the heatendothermic flow at approximately heat flow at approx 30 C andimately a stronger 30 °C and peak a atstronger approximately peak at ◦ approx40 C[9imately]. In comparison, 40 °C [9]. In the comparison, thermal transition the thermal of salmon transition collagen of salmon appears collagen as a single appears strong as a peak single in ◦ strongthe range peak of approximatelyin the range of 43–46 approxC[imately10]. These 43–46 findings °C [10]. indicate These findings that mild indicate heating mightthat mild weaken heating the mightattachments weaken of the pin attachments bones to the muscle.of pin bones However, to the heating muscle. the However, whole fillet heating would the result whole in fillet a very would poor productresult in and a very the heating poor product therefore and needs the heatingto be limited therefore to the needs area directly to be limited surrounding to the the area bone. directly The surrounding the bone. The main goal of the current work is to study whether high-intensity focused ultrasound can be used to perform such localized heating inside the fillet.

Foods 2017, 6, 82 3 of 15 main goal of the current work is to study whether high-intensity focused ultrasound can be used to perform such localized heating inside the fillet. High-intensity focused ultrasound is a technique initially developed for cancer treatment in humans [11]. An ultrasound transducer creates a focused sound field with a very high intensity and, when the sound is directed into biological tissue, viscous losses cause part of the ultrasonic energy to be converted into heat. Because the sound field can be focused into a very narrow beam, the technique enables localized heat treatment inside the tissue. The focal point of an HIFU transducer has an ultrasonic intensity in the order of 1000 W/cm3, and tissue can be heated to >70 ◦C in 1–3 s [12]. The region of high temperature around the focal point of the transducer typically has a “cigar” shape, which is longer along the direction of wave propagation. Biological tissues such as fish muscle are non-transparent, and the effects of HIFU treatment can therefore be hard to visualize. For this reason, experiments on HIFU “phantoms” are included in this paper. A phantom is a homogenous gel or solid that mimics the properties of biological tissue. Most methods for creating HIFU phantoms are based on fixing some kind of protein in a transparent gel [13–15]. When heated, the protein denatures and causes the treated area to appear white and opaque. This enables a direct visualization of the shape and size of the treated area. The work described in this paper consists of three main parts: (1) A study of various parameters related to HIFU treatment, performed using HIFU phantoms; (2) a study of the heating effect on fish muscle samples, employing thermal imaging; and (3) a study of HIFU treatment on bones in fillets and its effect on bone attachment in the muscle.

2. Materials and Methods

2.1. HIFU Instrumentation A Tektronix AFG2021 signal generator and an E&I 1040L amplifier were used to drive the HIFU transducers. Two different HIFU transducers manufactured by Precision Acoustics were used. The transducer specifications are listed in Table1. The maximum input power to the transducers for continuous operation was established through practical tests and discussions with the manufacturer. The transducers were operated at the maximum continuous input power in all experiments. The reason for this was twofold: Firstly, at low powers, the HIFU heating effect is low compared to the rate of heat dissipation in the rest of the sample, and thus the temperature never rises above a relatively low level; Secondly, the experimental conditions should give some insight into the potential for high-speed treatment on an industrial scale. Note that although the 680 kHz transducer is operated at a higher power (120 W) than the 1 MHz transducer (100 W), the conversion of energy from sound waves to heat is generally more efficient at higher frequencies [12]. The transducers are therefore expected to yield approximately the same heating effect.

Table 1. HIFU transducer specifications.

Serial Operating Active Area Focal Length Width of Focal Zone Max. Continuous Number Frequency (Hz) Diameter (mm) (mm) (FWHM) (mm) Input Power (W) PA439 680 60 75 4.3 120 PA500 1000 60 75 2.8 100

The treatment setup is shown in Figure2. The transducer was placed in a water bath, pointing towards the water surface. The sample to be treated was placed on a plate, and the plate was placed over the transducer at the water surface. A hole in the plate allowed the ultrasound to travel from the water, directly into the sample. Figure3 shows the complete setup placed in a plastic tub (used for submersion in water). A laser pointer was placed above the sample, pointing towards the focal point of the transducer. In this way, it was possible to see which part of the sample was being treated. For continuous treatment along a line, the sample was placed on a second plate and moved across Foods 2017, 6, 82 4 of 15 the transducer beam, as indicated by the white arrow in Figure3. The second plate was pulled by a stepper motor with an adjustable speed, and the position of the object under treatment was adjusted by hand during pulling, to keep the row of bones in the focal zone. Foods 2017, 6, 82 4 of 15

Foods 2017, 6, 82 4 of 15

(a) (a) (b) (b)

FigureFigure 2. HIFU 2. HIFU setup setup with with a transducer a transducer and and plateplate to holdhold the the fillet. fillet. (a )( aSchematic) Schematic illustration illustration with awith a Figure 2. HIFU setup with a transducer and plate to hold the fillet. (a) Schematic illustration with a transducertransducer placed placed in ain water a water bath; bath; (b ()b )Image Image showingshowing aa transducer transducer an dan plated plate mounted mounted to a metalto a metal transducer placed in a water bath; (b) Image showing a transducer and plate mounted to a metal stand. stand.stand.

Figure 3. CompleteFigure 3. Complete setup setup placed placed in a in plastic a plastic tub tub forfor immersion immersion in water. in water. The fish The fillet fish in placed fillet inon placeda on a perforated plate with a slit in the middle to allow the ultrasound to pass through to the fillet. The perforated plate with a slit in the middle to allow the ultrasound to pass through to the fillet. The plate Figureplate 3. Complete is pulled by setup a stepper placed motor in ain plastic the direction tub for of immersionthe white arrow. in water. A laser The pointer fish mountedfillet in placed above on a is pulled by a stepper motor in the direction of the white arrow. A laser pointer mounted above the perforatedthe fillet plate indicates with a theslit positionin the middle of the transducerto allow the focal ultrasound point, enabling to pass the th operatorrough to to the aim fillet. the The fillet indicatesultrasound the position at the bone of row the duri transducerng continuous focal treatment. point, enabling the operator to aim the ultrasound at plate is pulled by a stepper motor in the direction of the white arrow. A laser pointer mounted above the bone row during continuous treatment. the2.2 . filletUltrasound indicates Phantoms the position of the transducer focal point, enabling the operator to aim the ultrasound at the bone row during continuous treatment. 2.2. UltrasoundOne Phantoms common method for making HIFU phantoms is to use bovine serum albumin (BSA) as the protein and mix it with a polyacrylamide gel [13]. However, the polyacrylamide gel is made using 2.2. Ultrasound Phantoms One commonacrylamide, method which is fora known making neurotoxin. HIFU phantomsWith some experimentation, is to use bovine we serumfound that albumin a standard (BSA) as the agar gel could be mixed with a BSA solution to form a physically stable and relatively transparent protein andOne mix common it with method a polyacrylamide for making HIFU gel phantoms [13]. However, is to use the bovine polyacrylamide serum albumin gel (BSA) is made as the using proteinphantom. and mix The it agarwith gel a ispolyacrylamide completely non- toxicgel [13]. and wasHowever, therefore the preferred polyacrylamide over the pol gelyacrylamide. is made using acrylamide, which is a known neurotoxin. With some experimentation, we found that a standard acrylamide,This approach which isis similar a known to that neurotoxin. described in With [16], forsome the fabricationexperimentation, of thermal we therapy found phantoms. that a standard The method for making phantoms was to dissolve the BSA and agarose in separate solutions, agaragar gel could gel could be mixedbe mixed with with a BSAa BSA solution solution toto formform a physically physically stable stable and and relatively relatively transparent transparent carefully mixing them, pouring the mix into a container, and then letting it set in a refrigerator to phantom.phantom. The The agar agar gel gel is completelyis completely non-toxic non-toxic and was was therefore therefore preferred preferred over over the pol theyacrylamide. polyacrylamide. generate a solid form. Because the agarose needs to be dissolved at high temperatures (close to This approach is similar to that described in [16], for the fabrication of thermal therapy phantoms. This approach100 °C), the is similar BSA was to dissolved that described in a separate in [16], solution for the atfabrication room temperature, of thermal to therapy avoid prematu phantoms.re The methodThedenaturation method for offor making the making protein. phantoms phantoms waswas toto dissolve the the BSA BSA and and agarose agarose in separate in separate solutions, solutions, carefullycarefully mixing mixingInitial them, experiments them, pouring pouring were the conductedthe mix mix intointo to establish aa container,contai howner, muchand and then BSA then lettingcould letting be it dissolvedset it setin a in refrigeratorin a water, refrigerator to to generategenerate aand solid 23.1a solid form.% was form. Becausefound Because to be the an agarose theapproximate agarose needs limit. needs to An be to agarose dissolved be dissolved concentration at high at high temperaturesof 4.3% temperatures was found (close to (close be to 100 to ◦C), suitable—as high as possible without gelling during the mixing of the two solutions. The the BSA100 was°C), dissolved the BSA was in a dissolved separate in solution a separate at room solution temperature, at room temperature, to avoid premature to avoid denaturation premature of denaturationtemperature of the of theprotein. solutions was brought to approximately 30 °C (BSA) and 70 °C (agarose), and the the protein. Initial experiments were conducted to establish how much BSA could be dissolved in water, Initial experiments were conducted to establish how much BSA could be dissolved in water, and 23.1% was found to be an approximate limit. An agarose concentration of 4.3% was found to be and 23.1%suitable was—as found high to as be possible an approximate without gelling limit. An during agarose the concentrationmixing of the of two 4.3% solutions. was found The to be temperature of the solutions was brought to approximately 30 °C (BSA) and 70 °C (agarose), and the

Foods 2017, 6, 82 5 of 15 suitable—as high as possible without gelling during the mixing of the two solutions. The temperature of the solutions was brought to approximately 30 ◦C (BSA) and 70 ◦C (agarose), and the solutions were carefully combined in a 60/40% mixture (BSA/agarose). This resulted in a final concentration of 13.8% BSA and 1.72% agarose in the phantoms. To study the interaction between ultrasound and fish bones, phantoms with implanted pin bones were made. The bones were first collected from a number of cod fillets and sorted according to size. The phantom container was then partly filled with BSA/agar mixture, which was allowed to set for a few minutes, and the bones were placed on the sticky surface of the partly solidified gel. Lastly, the container was completely filled with the BSA/agar mixture, leaving the bones in the middle of the finished phantom.

2.3. Fish Samples Cod for experiments was sourced from the Tromsø Research Station. The cod was caught in 2014, fed at the facility for approximately one and a half years, and slaughtered on 23 October 2015. The weight of the cod after gutting and heading was in the range 3.0–6.9 kg, with 75% of the fish in the 3.0–5.0 kg range. The cod was stored on ice for three days, filleted at Nofima, and stored for an additional day on ice before the experiments. Whole salmon were purchased from Dragøy, a local fishmonger in Tromsø, and the details of the handling are therefore not known. The gutted weights of the salmon were in the 5–6 kg range. The salmon was filleted at Nofima on 3 November 2015, at which time it was in a state of rigor mortis. However, this did not seem to affect the condition of the finished fillet. The fillets were subsequently stored on ice for two days and were past the rigor state by the time of the experiment.

2.4. Thermal Imaging of Heating Effect To study the HIFU heating effect in fish muscle tissue, samples of cod and salmon were first subjected to HIFU treatment and then imaged using a thermal camera. Loins from cod and salmon were cut to fit on the HIFU treatment plate and were treated at different transducer-sample distances (25–75 mm) using the 680 kHz and 1 MHz transducer. Cod and salmon samples were treated for 15 and 20 s, respectively. A slightly higher treatment time was used for salmon, as initial tests indicated that the heating effect was lower in salmon. Experiments were conducted with both skinned and unskinned fillets. After treatment, the loin was sliced along the treated area, perpendicular to the fillet surface. The slice was subsequently imaged using a Testo 870 thermal camera, yielding information on the temperature distribution as a function of the lateral position and depth inside the fillet. Each image was taken 15 s after the end of treatment.

2.5. HIFU Treatment Protocol for Bone Pulling Experiments In the experiment, to study the effect of HIFU treatment on the bone pulling force, cod fillets were treated with the 1 MHz transducer and salmon fillets with the 680 kHz transducer, as the initial experiments suggested that these had the best heating effect for the respective species. The transducer-fillet distance was set to 50 mm, based on initial experiments that showed that for cod fillets, the best bone heating effect was obtained with distances in the range of 45–55 mm (for salmon, initial experiments showed no obvious optimal distance). The transducer was tilted approximately 15 degrees to align the transducer beam with the direction of the bones in the fillets. Cod fillets were treated from the skin side, with the muscle side facing upwards, enabling easy tracking of the bone position. Salmon fillets were treated from the muscle (opposite) side, since initial experiments indicated that skin effectively blocks the ultrasound. Cod fillets were divided into groups with and without skin, with eight fish (16 fillets) in each group. Each group was treated at a speed of 1.5 mm/s. The 10 salmon which were purchased for the experiment were intended for a single group with skin, treated at a speed of 1.5 mm/s. However, after treating three fillets, it became apparent that the treatment was heating the surface to such a Foods 2017, 6, 82 6 of 15 Foods 2017, 6, 82 6 of 15 In each experiment series, one of the fillets (right or left) was used as the control, and the other was treated with HIFU. The left and right fillets were alternated for the treatment and control, to degree that it began to split up and fall apart. Consequently, the speed was adjusted to 3 mm/s for the avoid any possible systematic differences between left and right fillets. remaining seven fillets. 2.6. BoneIn each Pulling experiment series, one of the fillets (right or left) was used as the control, and the other was treated with HIFU. The left and right fillets were alternated for the treatment and control, to avoid any possibleThe bones systematic were pulled differences using a betweenStable Micro left andSystems right TA fillets.-HDi texture analyzer with a 25 kg load cell. The bones were gripped with artery forceps and pulled vertically upwards at 10 mm/s (maximum2.6. Bone Pulling speed), whilst the fillet was held down by hand. The force time series from the texture analyzer were analyzed to calculate the peak force during pulling and the integrated force (work) The bones were pulled using a Stable Micro Systems TA-HDi texture analyzer with a 25 kg load from the start to end of the pulling. During the experiments, it was observed that cod bones would cell. The bones were gripped with artery forceps and pulled vertically upwards at 10 mm/s (maximum generally loosen after the peak value was reached, while salmon bones would detach from the speed), whilst the fillet was held down by hand. The force time series from the texture analyzer were muscle in a short series of tugs. analyzed to calculate the peak force during pulling and the integrated force (work) from the start to end For cod, 10–12 bones were pulled, starting at the bone closest to the tail, and for salmon, 12 of the pulling. During the experiments, it was observed that cod bones would generally loosen after the bones were pulled, starting with bone number 4 from the tail (bones 1–3 were very small and peak value was reached, while salmon bones would detach from the muscle in a short series of tugs. therefore not included). For cod, 10–12 bones were pulled, starting at the bone closest to the tail, and for salmon, 12 bones were pulled, starting with bone number 4 from the tail (bones 1–3 were very small and therefore not included). 3. Results 3. Results 3.1. Phantom Experiment Results 3.1. Phantom Experiment Results 3.1.1. Distance between Transducer and Sample—Boneless Phantom 3.1.1. Distance between Transducer and Sample—Boneless Phantom An experiment was performed to investigate how the distance between the transducer and the sampleAn affects experiment the treatment. was performed A 30 mm to investigatethick phantom how was the treated distance with between a 1 MHz the transducer for and 15 the s, withsample the affects transducer the treatment.-sample distance A 30 mm adjusted thick phantom between was 40 and treated 80 mm with in a 5 1 mm MHz steps. transducer The phantom for 15 s, withwas at the room transducer-sample temperature, approximately distance adjusted 20 °C. between 40 and 80 mm in 5 mm steps. The phantom was atThe room results temperature, are shown approximately in Figure 4. When 20 ◦C. the transducer is close to the sample, most of the heat is dissipatedThe results in the are upper shown surface in Figure of 4the. Whenphantom, the transducerwith denaturation is close causing to the sample, a bright most spot ofand the a depressionheat is dissipated in the in surface. the upper With surface increasing of the distance, phantom, the with denatured denaturation area causing extends a further bright spot into and the phantom,a depression and in at the a 60 surface. mm distance, With increasing when the distance,focal point the is denaturedcentered in area the phantom, extends further the denatured into the phantom,area has a and“cigar at a” 60shape. mm distance, This shape when corresponds the focal point to that is observed centered in in thethe phantom, application the of denatured HIFU in medicine,area has a under “cigar” ideal shape. conditions This shape [12]. However, corresponds at distances to that observed of 65 mm in and the above, application there is of no HIFU visible in denaturationmedicine, under inside ideal the conditions phantom, [12 but]. However, rather a series at distances of small of 65 holes mm on and the above, bottom there surface is no visible of the phantom.denaturation When inside the thefocal phantom, point is close but rather to the a water series-phantom of small holesinterface, on the the bottom sound field surface seems of the to phantom.have a kind When of “hammering the focal point” effec is closet rather to thethan water-phantom a heating effect. interface, the sound field seems to have a kindThe of results “hammering” illustrate effect thatrather for a relatively than a heating thin sample, effect. such as a fish fillet, the distance between the transducerThe results and illustrate the sample that for is a critical relatively for thin the sample,shape of such the astreated a fish fillet,area and the distancethe intensity between of the treatment.transducer and the sample is critical for the shape of the treated area and the intensity of the treatment.

Figure 4.4. PhantomPhantom treated treated with with a a 1 MHz transducer for 15 s at distances between 40 and 80 mm (distance marked below each treatment position).Theposition).The transducertransducer waswas placedplaced belowbelow thethe phantom.phantom.

Foods 2017, 6, 82 7 of 15

3.1.2.Foods Treatment 2017, 6, 82Time—Boneless Phantom 7 of 15 Foods 2017, 6, 82 7 of 15 A second experiment was performed to study the effect of treatment time on the heating effect. 3.1.2. 3.1.2Treatment. Treatment Time Time—Boneless—Boneless Phantom Phantom A 30 mm thick phantom was treated with the 1 MHz transducer at a 55 mm distance, at points placed A second experiment was performed to study the effect of treatment time on the heating effect. 10 mm apart.A Thesecond treatment experiment time was performed was adjusted to study between the effect of 3 treatment and 30 time s. The on the phantom heating effect. was at room A 30 A mm 30 thickmm thick phantom phantom was was treated treated with with the the 1 MHz transducer transducer at ata 55a 55 mm mm distance, distance, at points at points temperatureplacedplaced 10 prior mm 10 mmapart. to treatment. apart. The The treatment treatment time time was was adjusted adjusted betweenbetween 3 3and and 30 30s. The s. The phantom phantom was atwas room at room Thetemperature resultstemperature arepriorshown prior to treatment. to intreatment. Figure 5. For treatment times of 3–5 s, the treatment causes small specks to appearThe in the resultsThe phantom. results are are shown shown These in in Figure may Figure be5. 5. For dueFor treatment treatment to denaturation times times of of 3 – close35– s,5 thes, to the treatment pre-existing treatment causes causes gas small bubbles small or specks to appear in the phantom. These may be due to denaturation close to pre-existing gas bubbles ultrasound-inducedspecks to appear in cavitation the phantom. bubbles These [ 13may]. Forbe due treatment to denaturation times ofclose 10 to s pre and-existing above, gas a cigar-shaped bubbles or ultrasound-induced cavitation bubbles [13]. For treatment times of 10 s and above, a cigar-shaped denaturedor ultrasounddenatured area begins-induced area tobegins form, cavitation to growingform, bubblesgrowing thicker [13thicker]. withFor with treatment longer longer treatmenttreatment times of 10times. times. s and Note above, Note also also thata cigar for that -25shaped for 25 and 30 s,denatured the denaturedand 30 area s, the begins area denatured is to thicker form, area growing at is thickerthe bottom. thicker at the bottom.with This longer is This consistent istreatment consistent with times. with observations observations Note also that from from for previous 25 studiesand [12 30prev] showing s, ious the studies denatured that [12] ultrasonic showing area is cavitationthat thicker ultrasonic at results the cavitation bottom. in more results This heating in is more consistent in heating the area within the closest observationsarea closest to the to transducer. from previousthe transducer.studies [12] The showing results illustrate that ultrasonic that not onlycavitation the intensity results but in also more the heating shape of in the the treated area areaclosest to The results illustrate that not only the intensity but also the shape of the treated area is dependent on time. the transducer.is dependent The on resultstime. illustrate that not only the intensity but also the shape of the treated area is dependent on time.

Figure 5. HIFU phantom treated with a 1 MHz transducer at 55 mm distance, for 3–30 s (treatment Figure 5. HIFU phantom treated with a 1 MHz transducer at 55 mm distance, for 3–30 s (treatment time marked below each treatment position). The phantom was at room temperature prior to time markedtreatment. below The each horizontal treatment stripes position). in the image The are phantom due to imperfect was at cutting room of temperature the phantom slice. prior to treatment. The horizontalFigure 5. HIFU stripes phantom in the tre imageated arewith due a 1 toMHz imperfect transducer cutting at 55 ofmm the distance, phantom for slice.3–30 s (treatment time3.1.3 . marked Sample Temperature below each — treatmentBoneless position)Phantom. The phantom was at room temperature prior to 3.1.3. Sampletreatment. Temperature—BonelessIn HIFU The treatment,horizontal thestripes treated in Phantomthe area image must are be due heated to imperfect above a certaincutting thresholdof the phantom temperature slice. to cause denaturation. The input power and treatment time required to reach this threshold are In3.1.3 HIFU. dependentSample treatment, Temperature on the the initial treated— temperatureBoneless area Phantom must of the besample. heated In fish above processing a certain plants threshold the temperature temperature of the to cause product is kept close to 0 °C in order to minimize unwanted microbiological and enzymatic activity, denaturation.In HIFU The treatment, input power the treated and treatment area must timebe heated required above to a reachcertain this threshold threshold temperature are dependent to and an experiment was conducted to study the effect of treatment at these low, more realistic on the initial temperature of the sample. In fish processing plants the temperature of the product cause temperatures. denaturation. HIFU The phantoms input power were cooled and treatment to approximately time required 4 °C and to treated reach with this the threshold 1 MHz are ◦ is keptdependent closetransducer to on 0 theC atin ainitial 55 order mm temperature distance to minimize for treatment of the unwanted sample. times between In microbiological fish 3processing and 50 s. plants and enzymatic the temperature activity, of the and an experimentproduct wasisImages kept conducted close of the to sliced 0 to°C studyinphantoms order the to are effectminimize shown of in treatment unwanted Figure 6. For atmicrobiological thesetreatment low, times more and of 20enzymatic realistic s or less, temperatures. activity,no HIFUand phantoms andenaturation experiment were is cooled visible. was conductedHowever, to approximately for to 30 study–50 s, 4 a the◦thinC and effectcigar treated-shaped of treatment denaturation with the at 1 these MHzarea is low, transducerclearly m orevisible. realistic at a 55 mm Comparing these results to the results for the phantom at room temperature (Figure 5), it is clearly distancetemperatures. for treatment HIFU times phantoms between were 3 and cooled 50 s. to approximately 4 °C and treated with the 1 MHz transducerseen that at athe 55 heating mm distance effect is forsignifi treatmentcantly reduced times bybetween the lower 3 and sample 50 s.temperature. Images of the sliced phantoms are shown in Figure6. For treatment times of 20 s or less, no Images of the sliced phantoms are shown in Figure 6. For treatment times of 20 s or less, no denaturation is visible. However, for 30–50 s, a thin cigar-shaped denaturation area is clearly visible. denaturation is visible. However, for 30–50 s, a thin cigar-shaped denaturation area is clearly visible. ComparingComparing these these results results to to the the results results for for thethe phantomphantom at at room room temperature temperature (Figure (Figure 5), it5 ),is itclearly is clearly seen thatseen thethat heatingthe heating effect effect is significantlyis significantly reduced reduced byby the lower lower sample sample temperature. temperature.

Figure 6. Slices of HIFU phantoms, treated with a 1 MHz transducer at a 55 mm distance for 10, 20, 30, 40, and 50 s (left to right). The phantom temperature was approx. 4 °C before treatment.

FigureFigure 6. Slices 6. Slices of HIFU of HIFU phantoms, phantoms, treated treated with with aa 11 MHz transducer transducer at ata 55 a 55mm mm distance distance for 10, for 20, 10, 30, 20, 30, 40, and 50 s (left to right). The phantom temperature was approx. 4 °C before treatment. 40, and 50 s (left to right). The phantom temperature was approx. 4 ◦C before treatment.

Foods 2017, 6, 82 8 of 15 Foods 2017, 6, 82 8 of 15

3.1.4.3.1.4. Bone in Phantom AA phantomphantom experimentexperiment was performed to studystudy thethe interactioninteraction betweenbetween bones and the HIFUHIFU soundsound field.field. PhantomsPhantoms of of a a 30 30 mm mm thickness thickness with with implanted implanted bones bones were were treated treated with with the 680the kHz680 kHz and 1and MHz 1 MHz transducers transducers at distances at distances of 25, of 35, 25, and 35, 45and mm 45 formm 30, for 45, 30, and 45, 60and s at60 each s at each distance. distance. The 25The mm 25 distancemm distance was includedwas included in order in order to study to study the effect the effect of defocusing of defocusing the sound the sound field. field. The phantoms The phantoms were cooledwere cooled to 4 ◦C to prior 4 °C toprior treatment. to treatment. After treatment,After treatment, the phantoms the phantoms were sliced were andsliced imaged. and imaged. The results The areresults shown are inshown Figure in7 Figure. 7. Generally,Generally, thethe heatingheating effecteffect atat aa 2525 mmmm distancedistance isis lowerlower thanthan atat 3535 andand 4545 mm,mm, duedue toto thethe defocuseddefocused naturenature of of the the sound sound field. field. However, However, with with the the 25 mm25 mm distance, distance, the denaturationthe denaturation is only is closeonly toclose the to bone, the bone, while while for 35 for and 35 45 and mm, 45 mm, there there are clouds are clouds of denaturation of denaturation outside outside of the of bone.the bone. For theFor 680the 680 kHz kHz transducer, transducer, the cloudsthe clouds are centeredare centered on the onbone, the bone, while while for the for1 the MHz 1 MHz transducer, transducer, there there is also is analso area an area of denaturation of denaturation in the in upperthe upper surface surface of the of phantom.the phantom. The The result result for for the the 1 MHz 1 MHz treatment treatment at 25at mm25 mm for for 60 s60 (lower s (lower left cornerleft corner of Figure of Figure7b) is especially7b) is especially interesting. interesting. In this In case, this there case, is there a clearly is a visibleclearly areavisible of denaturationarea of denaturation along almost along allalmost of the all bone, of the with bone, no with additional no additional “clouds” “clouds outside” outside of this. Thisof this. result This indicates result indicates that the that use ofthe a use defocused of a defocus sounded field sound may field be useful may be to useful reduce to the reduce amount the ofamount denaturation of denaturation in areas in apart areas from apart outside from outside of the of bone, the bone, while while still yieldingstill yielding a heating a heating effect effect close close to theto the bone. bone.

(a) (b)

FigureFigure 7.7. SlicesSlices ofof phantomsphantoms with embedded bones,bones, treated at a 25–4525–45 mm distance and for 30 30–60–60 s. TheThe phantomphantom temperaturetemperature waswas approx.approx. 4 ◦°CC before treatment. ( a) Treated with a 680 kHz transducer;transducer; ((bb)) TreatedTreated withwith aa 11 MHzMHz transducer.transducer.

3.2.3.2. Salmon Experiment Results

3.2.1.3.2.1. Effect of TransducerTransducer DistanceDistance andand PresencePresence ofof SkinSkin TheThe results of of thermal thermal imaging imaging of of HIFU HIFU treated treated salmon salmon samples samples are are shown shown in Figure in Figure 8. The8. Theexperiment experiment was was performed performed with with both both the the 680 680 kHz kHz and and the the 1 1 MHz MHz transducer, transducer, but but to to limit the the numbernumber of figures,figures, only the results for for the the 680 680 kHz kHz transducer transducer are are shown shown here. here. Th Thee effects effects using using the the 1 1MHz MHz transducer transducer were were slightly slightly weaker weaker,, but but otherwise otherwise quite quite similar. similar. TheThe thermalthermal imagesimages clearly show that there is aa significantlysignificantly strongerstronger heatingheating effect inin samplessamples withoutwithout skin skin compared compared to skinned to skinned fillets. The fillets. heat distributionThe heat distribution is also affected is by also the transducer-sampleaffected by the distance,transducer with-sample short distance, distances with resulting short indistances more heat resulting dissipation in more at the heat upper dissipation surface, at and the longer upper distancessurface, and resulting longer indistances more heat resulting dissipating in more at heat the lowerdissipating surface. at the However, lower surface. the patterns However, of heat the distributionpatterns of heat are lessdistribution pronounced are less than pro whatnounced was observed than what with was the observed phantoms. with There the phantoms. is a tendency There for mostis a tendency of the heat for to most be dissipated of the heat at theto be lower dissipated surface, at i.e., the the lower side closestsurface, to i.e. the, the transducer. side closest Using to the transducer. Using the 680 kHz transducer on skinned fillets, a distance of 55 mm produced significant heating throughout the samples.

Foods 2017, 6, 82 9 of 15

680 kHz transducer on skinned fillets, a distance of 55 mm produced significant heating throughout Foods 2017, 6, 82 9 of 15 the samples.

Figure 8. Thermal images of salmon samples treated with the 680 kHz transducer for 20 s. The Figure 8. Thermal images of salmon samples treated with the 680 kHz transducer for 20 s. The transducer-sample distance is listed below each set of images. transducer-sample distance is listed below each set of images.

To testtest howhow thethe fatfat layerlayer underneathunderneath the skin affects the HIFUHIFU treatment, an additional test was performedperformed on a salmonsalmon sample. Here,Here, a salmon loin sample was prepared with three test areas; one withwith intact intact skin, skin, one one where where skin skin was wasremoved removed but with but the with fat and the dark fat and muscle dark underneath muscle underneath remaining, andremaining, one with and the one muscle with directly the muscl exposed.e directly Each exposed. area was Each treated area with was the treated 680 kHz with transducer the 680 kHz at a 55transducer mm distance at a 55 for mm 20 s.distance for 20 s. The result is shown shown in in Figure Figure 9.9. With With the the muscle muscle directly directly exposed, exposed, the the HIFU HIFU treatment treatment yields yields an anarea area of of denaturation denaturation throughout throughout the the whole whole sample. sample. However, However, w withith some some fat fat and and dark muscle muscle remaining, thethe effecteffect onlyonly reachedreached halfwayhalfway intointo thethe sample.sample. InIn thethe partpart ofof thethe loinloin withwith thethe skinskin intact,intact, no denaturation was observed. This indicates that for optimum efficiency efficiency of the treatment, both skin and fat/darkfat/dark muscle muscle should should be be removed. removed.

Figure 9.9. Salmon fillet fillet treated with a 680 kHz transducer at a 55 mm distance for 20 s, in three three different points (from left left to to right): right): Skinned, Skinned, with with skin skin removed removed but but fat fatand and dark dark muscle muscle remaining, remaining, and andwithwith skin. skin.

3.2.2.3.2.2. Bone Pulling ExperimentsExperiments The effect of HIFUHIFU treatment on bone pulling is shown in Figure 1010,, which shows the mean reduction inin peak pulling force and work. The reduction in peak pu pullinglling force is approximately 30 30%,%, and the reduction in pullingpulling work is approximately 12%12% for both treatment speeds. Note, however, thatthat onlyonly threethreefillets fillets were were treated treated at at 1.5 1.5 mm/s, mm/s, and and the the results results are are therefore therefore not not as reliableas reliable as thoseas those for thefor the 3.0 mm/s3.0 mm/s treatment. treatment. There is aa largelarge differencedifference betweenbetween thethe peakpeak pullingpulling forceforce andand thethe pullingpulling workwork values.values. DuringDuring pulling,pulling, thethe salmon bones wouldwould detachdetach from thethe filletfillet inin a short series of tugs, rather than one single tugtug with with a a peak peak force force value. value. The The pulling pulling work work metric metric therefore therefore seems seems more representativemore representative for salmon. for salmon.There was only one bone broken and no bones slipping in all of the salmon experiments. There was thusThere no was problem only one of “overtreatment” bone broken and in no this bones series slipping of experiments. in all of the salmon experiments. There was thus no problem of “overtreatment” in this series of experiments. The effect of HIFU treatment on the fillet surface was not systematically recorded, but the effect was generally more severe at treatment speeds of 1.5 mm/s. Example images of fillets after treatment are shown in Figure 11.

Foods 2017, 6, 82 10 of 15

The effect of HIFU treatment on the fillet surface was not systematically recorded, but the effect was generally more severe at treatment speeds of 1.5 mm/s. Example images of fillets after treatment FoodsareFoods shown 2017 2017, 6, ,6 in82, 82 Figure 11. 1010 of of 15 15

Figure 10. Mean values for the reduction of bone pulling force and work in salmon fillets, with error FigureFigure 10. 10. Mean Mean values values for for the the reduction reduction of of bone bone pulling pulling force force and and work work in in salmon salmon fillets, fillets, with with error error bars marking corresponding mean standard deviation values. Reduction percentages are calculated barsbars marking marking corresponding corresponding mean mean standard standard deviation deviation valu values.es. Reduction Reduction percentages percentages are are calculated calculated relative to the control fillet from the same fish. relativerelative to to the the control control fillet fillet from from the the same same fish. fish.

(a(a) ) (b(b) ) Figure 11. Examples of salmon fillets after HIFU treatment. (a) Treated at 1.5 mm/s. Denaturation Figure 11.11. ExamplesExamples ofof salmon salmon fillets fillets after after HIFU HIFU treatment. treatment. (a) ( Treateda) Treated at 1.5 at mm/s.1.5 mm/s. Denaturation Denaturation and andsplittingand splitting splitting clearly cl clearly visible;early visible; visible; (b) Treated ( b(b) )Treated Treated at 3 mm/s.at at 3 3mm/s. mm/s. Some Some Some denaturation denaturation denaturation visible, visible, visible, no splitting. no no splitting. splitting.

3.3. Cod Experiment Results 33.3..3. Cod Experiment Results

3.3.13.3.1.3.3.1. .Effect Effect of of Transducer Transducer Distance Distance and and Presence Presence ofof of SkinSkin Skin TheThe results results results of of of thermal thermal imaging imaging imaging of of of HIFU HIFU HIFU treated treated treated cod cod cod samples samples samples are are are shown shown shown in in Figure Figure in Figure 12. 12. The The12. Theexperimentexperiment experiment was was was performed performed performed with with with both both both thethe the 680 680 680 kHz kHz kHz and and and the the the 1 1 1 MHz MHz MHz transducer, transducer, but but to to to limit limit limit the the numbernumber of of figures, figures, figures, only only the the results results for for the the 1 1 MHz MHz transducer transducer are are shown shown here. here. The The effects effects using using the the 680680 MHz MHz were were slightly slightly weaker,weaker weaker, but,but but otherwiseotherwise otherwise quitequite quite similarsimilar similar (a(a (a tendencytenden tendencycy oppositeopposite opposite toto to thatthat that ofof of salmon).salmon). salmon). AsAs with withwith the the the salmon, salmon, salmon, the the theeffect effect effect of transducer-sample of of transducer transducer-sample-sample distance distance distance to the to cod to the theis similar cod cod is is to similar similar that observed to to that that observedinobserved phantoms, in in phantoms, phantoms, but the effect but but isthe the less effect effect pronounced. is is less less pronounced. pronounced. There is also There There a clear is is also also tendency a a clear clear fortendency tendency the heat for for to the bethe heat mostly heat to to bebe mostly mostly dissipated dissipated at at the the upper upper and and lower lower surfaces surfaces of of the the samp samples.les. This This is is probably probably due due to to the the transitionstransitions between between media media of of different different acoustic acoustic impedances impedances (water, (water, tissue, tissue, and and air). air). There There are are no no clearlyclearly visible visible differences differences between between samples samples with with or or wit withouthout skin skin in in the the thermal thermal image. image.

Foods 2017, 6, 82 11 of 15 dissipated at the upper and lower surfaces of the samples. This is probably due to the transitions between media of different acoustic impedances (water, tissue, and air). There are no clearly visible Foods 2017, 6, 82 11 of 15 differencesFoods 2017, 6, between82 samples with or without skin in the thermal image. 11 of 15

FigureFigure 12.12.12. ThermalThermal images images of of cod cod samplessamsamplesples treated treated with with the the the 111 MHz MHz MHz transducer transducer transducer for for for 15 15 15 s. s. s. The The The transducer-sampletransducertransducer--samplesample distancedistancedistance isisis listedlistedlisted belowbelowbelow eacheacheach setsetset ofofof images.images.images.

AdditionalAdditional experimentsexperiments werewere performedperformed tototo studystudy thethe heatingheating effecteffect onon bonesbones duringduring continuouscontinuous HIFUHIFU treatmenttreatment alongalong thethe bonebone row.row. CodCodCod filletsfilletsfillets withwithwith andandand withoutwithout skinskin werewere treatedtreated withwithwith thethe 680680 kHzkHz andand 11 MHzMHz transducer,transducer, atatat distancesdistances ofof 4545 andand 5555 mm,mm, whichwhich seemedseemed promisingpromising basedbased onon thethe thermalthermal imagingimagingimaging results.results. TheThe filletsfilletsfillets werewere treatedtreatedtreated atatat aaa speedspeedspeed ofofof 111 mm/s,mm/s,mm/s, and and were were cut cut along along the the rowrow ofof bonesbonesbones afterafteafterr treatment.trtreatment.eatment. TheThe strongest strongest strongest heating heating heating effect effect effect was was was seen seen seen for the for for cod the the fillet cod cod with fillet fillet skin, with with treated skin, skin, with treated treated the 1 with with MHz the the transducer 1 1 MHz MHz attransducertransducer a 45 mm distance.atat aa 4545 mmmm This distance.distance. fillet is shownThisThis filletfillet in Figure isis shownshown 13. Manyinin FigureFigure of the 13.13. bones ManyMany have ofof thethe clearly bonesbones detached havehave clearlyclearly from thedetacheddetached muscle, fromfrom and thethe there muscle,muscle, is also andand one therethere bone is whichis alsoalso hasoneone become bonebone whichwhich burned. hashas However, becomebecome burned.burned. the rest However,However, of the fish thethe muscle restrest doesofof thethe not fishfish have musclemuscle a very doesdoes “cooked” notnot havehave appearance, aa veryvery ““cookedcooked indicating”” appearance,appearance, again that indicatingindicating the bones againagain absorb thatthat heat thethe more bonesbones efficiently absorbabsorb thatheatheat the momo muscle.rere efficientlyefficiently thatthat thethe muscle.muscle. NoteNote that that that similar similar similar experiments experiments experiments were were were conducted conducted conducted on salmononon salmon salmon samples, samples, samples, but without but but without without any visible any any visiblevisible effect oneffecteffect the onon bones thethe bonesbones or the oror tissue thethe tissuetissue surrounding surroundingsurrounding the bone. thethe bone.bone. Images ImagesImages from fromfrom these thesethese experiments experimentsexperiments are therefore areare thereforetherefore not includednotnot includedincluded here. here.here.

FigureFigure 13. 13.13.Example ExampleExample of of ofa cod aa codcod fillet fillet fillet treated treated treated with awith with 1 MHz aa 11 transducer MHz MHz transducer transducer at a 45 mm at at distance,aa 4545 mm mm with distance, distance, continuous with with treatmentcontinuouscontinuous along treatmenttreatment the row alongalong of bones thethe rorow atw of aof speed bonesbones of atat 1aa mm/s.speedspeed ofof The 11 mm/s.mm/s. fillet was TheThe cutfilletfillet along waswas cutcut the alongalong row of thethe bones rowrow toofof visualizebonesbones toto visualizevisualize the treatment thethe treatmenttreatment effect. Note effect.effect. that NoteNote some thatthat bones somesome have bonesbones become havehave detached becomebecome detacheddetached from the from muscle,from thethe and muscle,muscle, that someandand thatthat have somesome been havehave “burned”, beenbeen ““ resultingburnedburned””,,in resultingresulting a brown inin color. aa brownbrown The image color.color. representsTheThe imageimagea representsrepresents “best case” aa with ““bestbest regard casecase”” towithwith the regardregard treatment toto thethe effect, treatmenttreatment and is effect,effect, not representative andand isis notnot representativerepresentative of all fillets treated. ofof allall filletsfillets treated.treated.

3.3.2.3.3.23.3.2.. BoneBone PullingPulling ExperimentsExperiments TheThe meanmean valuesvaluesvalues forforfor thethethe reductionreductionreduction of ofof pulling pullingpulling force/work forforce/workce/work areare shownshown inin FigureFigureFigure 1414a14aa... TheThe meanmeanmean reductionreductreductionion inin thethethe peakpeak pullingpulling forceforce isisis 50%5050%% forforfor filletsfillfilletsets withwithwith skin,skin,skin, andandand 72%7272%% forforfor filletsfilletsfillets withoutwithoutwithout skin.skin.skin. NoteNote alsoalso thatthat thethe standardstandard deviationdeviationdeviation isisis lowerlowerlower forforfor fillets filletsfillets without withoutwithout skin. skin.skin. AlthoughAlthough thethe resultsresults seemseem promisingpromising withwith regardregard toto reducingreducing thethe bonebone attachment,attachment, therethere werewere alsoalso aa significantsignificant numbernumber ooff failedfailed bonebone pullingpulling attempts.attempts. TheseThese areare divideddivided intointo attemptsattempts wherewhere thethe bonesbones brokebroke duringduring pullingpulling (most(most oftenoften duedue toto overheating)overheating) andand attemptsattempts wherewhere thethe bonebone slippedslipped fromfrom thethe arteryartery forceps.forceps. TheseThese numbernumberss areare summarizedsummarized inin FigureFigure 14b14b.. ForFor thethe HIFUHIFU treattreateded filletsfillets wwithoutithout skin, skin, 35 35%% ofof the the bones bones broke broke during during pulling, pulling, indicating indicating that that the the fillets fillets werewere overtreated. overtreated.

Foods 2017, 6, 82 12 of 15

Although the results seem promising with regard to reducing the bone attachment, there were also a significant number of failed bone pulling attempts. These are divided into attempts where the bones broke during pulling (most often due to overheating) and attempts where the bone slipped from theFoods artery 2017, 6 forceps., 82 These numbers are summarized in Figure 14b. For the HIFU treated fillets without12 of 15 Foodsskin, 2017 35%, 6,of 82 the bones broke during pulling, indicating that the fillets were overtreated. Studying12 of 15 theStudying pulling the force pulling values force in detail, values we in found detail, that we the found number that ofthe broken number bones of broken was much bones higher was inmuch the Studyingsmallesthigher in fillets. the the pullingsmallest For the force fillets. treated values For fillets the in withdetail,treated skin, we fil lets thefound numberwith that skin, the of the broken number number bones of ofbroken wasbroken only bones bones 9%. was Thus, was much only the higherpresence9%. Thus, in ofthe the skin smallest presence reduces fillets. of the skin effectFor reduces the of HIFUtreated the treatmenteffect fillets of with HIFU to someskin, treatment degree,the nu mberto but some alsoof degree,broken reduces bonesbut the also number was reduces only of 9brokenthe%. Thus,number bones. the of presence The broken number bones.of skin of bonesThe reduces number slipping the ofeffect is bones included of HIFUslipping in treatment the is figureincluded to to some show in the degree, that figure this butto number show also reducesthat is quite this thelownu mbernumber for HIFU is quite of treatedbroken low for filletsbones. HIFU compared The treated number tofillets the of bones controlcompared slipping fillets. to the is includedcontrol fillets. in the figure to show that this numberTheThe is amount amountquite low of of visible for visible HIFU denaturation denaturationtreated fillets on the oncompared fillets the fillets was to notthe was systematicallycontrol not systematically fillets. recorded, recorded, but the amount but the ofamount denaturationThe of amount denaturation was of visible generally was denaturation generally higher forhigher onthe the filletsfor filletsthe without fillets was without notskin. systematically Example skin. Example images recorded, images of fillets butof fillets after the amounttreatmentafter treatment of aredenaturation shown are shown in Figure was in generallyFigure 15. 15. higher for the fillets without skin. Example images of fillets after treatment are shown in Figure 15.

FigureFigure 14. 14.( (aa)) Mean Mean valuesvalues for the reduction reduction of of bone bone pulling pulling force force and and work work in cod in cod fillets, fillets, with with error error bars Figurebarsmarking marking 14 corresponding. (a) Mean corresponding values mean for meanstandardthe reduction standard deviation of deviationbone values. pulling values.Reduction force Reductionand percentages work in percentages cod are fillets, calculated with are calculated error relative bars to markingrelativethe control tocorresponding thefillet control from the filletmean same from standard fish; the (b)same deviationNumber fish; of values. ( failedb) Number Reductionbone pulling of failed percentages attempts bone pullinginare cod calculated fillet attempts experiments. relative in cod to thefillet con experiments.trol fillet from the same fish; (b) Number of failed bone pulling attempts in cod fillet experiments.

(a) (b) (a) (b) Figure 15. Example of a cod fillet after HIFU treatment. (a) Fillet with skin: A small amount of white denaturated muscle visible along the bone row; (b) Fillet without skin. Denaturation is clearly visible Figure 15. Example of a cod fillet fillet after HIFU treatment. ( a) Fillet with skin: A small amount of white along the bone row, and some bones are burned. denaturated mus musclecle visible along the bone row; ( b) Fillet without skin. Denaturation Denaturation is clearly visible along the bone row, and somesome bonesbones areare burned.burned. 4. Discussion 4. Discussion For HIFU phantoms that are comparable to fish fillets in thickness (30 mm), correct adjustment of theFor transduHIFU phantomscer-sample that distance are comparable was critical to fish for fillets efficient in thickness heating (30 andmm), a correct well-shaped adjustment heat ofdistribution the transdu insidecer- samplethe samples. distance However, was criticalthe experiments for efficient on fish heating samples and did a not well show-shaped the same heat distributionsensitivity to inside distance. the samples.The distance However, affected the the experiments distribution on of fish heat, samples but most did of not the show heat wasthe sameeither sensitivitydistributed to at distance. the upper The or distancelower surfaces affected of the the distributionfillet. The treated of heat, areas but also most did of notthe haveheat wasthe “eithercigar” distributedshape seen atin the the upper phantoms. or lower This surfaces illustrates of thethat fillet. the behaviour The treated of areasHIFU also in fish did fillets not have is very the different“cigar” shape seen in the phantoms. This illustrates that the behaviour of HIFU in fish fillets is very different

Foods 2017, 6, 82 13 of 15

4. Discussion For HIFU phantoms that are comparable to fish fillets in thickness (30 mm), correct adjustment of the transducer-sample distance was critical for efficient heating and a well-shaped heat distribution inside the samples. However, the experiments on fish samples did not show the same sensitivity to distance. The distance affected the distribution of heat, but most of the heat was either distributed at the upper or lower surfaces of the fillet. The treated areas also did not have the “cigar” shape seen in the phantoms. This illustrates that the behaviour of HIFU in fish fillets is very different than in its original application of cancer treatment in humans, where the “sample” is several centimetres thick. In thinner samples such as the fish fillets, the heat distribution is more difficult to control. The experiments on phantoms also indicate that bones absorb ultrasound energy and convert it to heat more efficiently than muscle tissue. This effect is, to some degree, confirmed by the experiments on cod samples. The bones appeared to detach from the surrounding muscle, with only small denaturation effects in the muscle. This “self-focusing” effect for bones may be useful in an industrial application. Rather than targeting each bone in the fillet with high precision, it may be possible to apply a strong, non-focused sound field and rely on the bones to absorb the sound waves where needed. This could reduce the accuracy requirement for positioning of the transducer. The experiments on salmon samples with and without skin showed that its skin (and the dark muscle and fat underneath) effectively blocks the ultrasound and thereby reduces the heating effect. From a salmon processing perspective, this limits the usefulness of HIFU treatment, as a large portion of salmon fillets are sold with skin. In the experiments, a possible “workaround” of treating the fillets from the muscle side was tested. Unfortunately, the results show that the reduction in the bone pulling force after HIFU treatment is relatively small. One possible explanation for this is that the heating effect along the bone row appeared not to penetrate beyond a few millimetres into the fillet. However, some of the results obtained for single salmon samples indicate that the HIFU effect can penetrate throughout the whole fillet. This discrepancy is not yet understood, and further studies on salmon are needed before definite conclusions can be drawn. The initial experiments on single cod samples suggested that the presence of skin had no effect on the heat intensity and distribution in the samples. However, the bone pulling experiments showed that the heating effect was stronger for skinned fillets, resulting in a larger number of burned and broken bones. These unwanted effects were mainly seen for small bones, while larger bones were still strong enough to be pulled. This indicates that the problem could be alleviated by a better adjustment of the power of the HIFU treatment according to bone size (estimated from fish weight and bone position). For practical reasons, all of the experiments were conducted “post-rigor”. In many cases, fish fillet producers would like to perform filleting and bone removal before rigor mortis to reduce the processing time and increase the product shelf life [17]. Future research should include experiments on such pre-rigor fillets to determine its usefulness in these types of samples. The HIFU transducers were driven with 100–120 W of electrical input power, enabling the treatment of pin bone rows at 2–3 mm/s. Even assuming that 5 mm/s is sufficient, there is still a significant gap between this and industrial fish processing speeds, which are usually around 400 mm/s, i.e., 80 times faster. The industrial application of HIFU will therefore require custom made HIFU transducers with significantly higher power levels than any “off-the-shelf” transducers available today. A future study should also consider the practical aspects of acoustic coupling between the transducer and the fish fillets. In the experiments described here, coupling was achieved by submerging the transducer in water, and placing the fillet in the surface of the water. This may be impractical in an industrial implementation, and alternative “dry” coupling methods [18] should be researched and tested. The use of HIFU technology for localized heating may also be interesting in other areas of food processing. HIFU could potentially be used as a mild treatment to ease the deboning of products such as chicken or beef. Also, as a gastronomic technique, HIFU could, for example, be used to create dishes Foods 2017, 6, 82 14 of 15 that are cooked inside while remaining raw on the outside. Future research may pave the way for such new applications.

5. Conclusions We have studied high intensity focused ultrasound (HIFU) as a potential technology for weakening the attachment of pin bones in fish fillets. Experiments were conducted on cod and salmon samples, in addition to HIFU phantoms. The results show that the thin geometry of fish fillets causes the heat to be mainly dissipated at the surfaces of the fillets, making the effects of HIFU treatment less controllable than in thicker geometries (e.g., human patients). The pin bones are seen to absorb the ultrasound energy more efficiently than the surrounding tissue, creating a “self-focusing” effect which can possibly be exploited for bone heating, avoiding the need to aim the HIFU transducer with a high degree of accuracy. For salmon fillets, the skin partly blocks the ultrasound from entering the fish muscle, significantly reducing the heating effect. This effect is less pronounced for cod fillets with skin, allowing for the significant heating of fillets both with and without skin. HIFU treatment was more effective for decreasing bone attachment in cod than in salmon. For cod, a reduction in pulling force of 50–70% was observed, but care must be taken to avoid the overtreatment of bones, which can result in bones breaking during pulling. For salmon, HIFU treatment resulted in a reduction of 30% for peak force and 10% for total pulling work. Thus, for the experimental setup tested here, HIFU treatment seems more promising for the processing of cod than for salmon. However, further experimentation and development is required to determine the full potential of HIFU treatment to aid in fish processing.

Acknowledgments: The work described in this paper was funded by the Norwegian Seafood Research Fund (FHF) through project 901125 (“Høy-intensitets fokusert ultralyd (HIFU) anvendt på fisk: En studie av effekten på ulike vevstyper”). The authors also wish to thank Kathryn E. Washburn at Nofima for her help on reviewing and proofreading the paper. Author Contributions: K.A.Þ., K.H., and M.H.S. conceived the experiments; M.H.S. and S.K.S. designed and performed the experiments; M.H.S. and S.K.S. analyzed the data; and M.H.S. and K.A.Þ. wrote the paper. Conflicts of Interest: The authors declare no conflict of interest. The funding sponsors (the Norwegian Seafood Research Fund) was involved in the design of the study, but otherwise had no role in the collection, analyses, and interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

1. Shadwick, R.E.; Gemballa, S. Structure, Kinematics, and Muscle Dynamics in Undulatory Swimming. Fish Physiol. 2005, 23, 241–280. [CrossRef] 2. Akse, L.; Tobiassen, T.; Halsebakke, H. Tykkfiskbein i Torskefilet. Plassering (antall), Trekkraft og Bruddstyrke. Available online: https://nofima.no/en/pub/1189715/ (accessed on 17 August 2017). 3. Thielemann, J.T.; Kirkhus, T.; Kavli, T.; Schumann-Olsen, H.; Haugland, O.; Westavik, H. System for Estimation of Pin Bone Positions in Pre-rigor Salmon. In Advanced Concepts for Intelligent Vision Systems; Springer: Berlin/Heidelberg, Germany, 2007; pp. 888–896. 4. Esaiassen, M.; Sørensen, N.K. Fjerning av Tykkfiskbein i Laks. Available online: https://nofima.no/ filearchive/Rapport28-1996_1.pdf (accessed on 17 August 2017). 5. Westavik, H. Fjerning av pinnebein i filet av laks slaktet ved oppdrettsmerd. In SINTEF Report Series; SINTEF Fiskeri og Havbruk AS: Trondheim, Norway, SFH80 A095019; ISBN 9788214046502. 6. Zhol, S.; Ackman, R.G.; Morrison, C. Storage of lipids in the myosepta of Atlantic salmon (Salmo salar). Fish Physiol. Biochem. 1995, 14, 171–178. [CrossRef][PubMed] 7. Rønning, S.B.; Østbye, T.K.; Krasnov, A.; Vuong, T.T.; Veiseth-Kent, E.; Kolset, S.O.; Pedersen, M.E. The role of extracellular matrix components in pin bone attachments during storage—A comparison between farmed Atlantic salmon (Salmo salar) and cod (Gadus morhua L.). Fish Physiol. Biochem. 2017, 43, 549–562. [CrossRef] [PubMed] 8. Sikorski, Z.E.; Scott, D.N.; Buisson, D.H.; Love, R.M. The role of collagen in the quality and processing of fish. CRC Crit. Rev. Food Sci. Nutr. 1984, 20, 301–343. [CrossRef][PubMed] Foods 2017, 6, 82 15 of 15

9. Hastings, R.J.; Rodger, G.W.; Park, R.; Matthews, A.D.; Anderson, E.M. Differential Scanning Calorimetry of Fish Muscle: The Effect of Processing and Species Variation. J. Food Sci. 2006, 50, 503–506. [CrossRef] 10. Moreno, H.M.; Montero, M.P.; Gómez-Guillén, M.C.; Fernández-Martín, F.; Mørkøre, T.; Borderías, J. Collagen characteristics of farmed Atlantic salmon with firm and soft fillet texture. Food Chem. 2012, 134, 678–685. [CrossRef][PubMed] 11. Jenne, J.W.; Preusser, T.; Günther, M. High-intensity focused ultrasound: Principles, therapy guidance, simulations and applications. Z. Med. Phys. 2012, 22, 311–322. [CrossRef][PubMed] 12. Bailey, M.R.; Khokhlova, V.A.; Sapozhnikov, O.A.; Kargl, S.G.; Crum, L.A. Physical mechanisms of the therapeutic effect of ultrasound (a review). Acoust. Phys. 2003, 49, 369–388. [CrossRef] 13. Lafon, C.; Zderic, V.; Noble, M.L.; Yuen, J.C.; Kaczkowski, P.J.; Sapozhnikov, O.A.; Chavrier, F.; Crum, L.A.; Vaezy, S. Gel phantom for use in high-intensity focused ultrasound dosimetry. Ultrasound Med. Biol. 2005, 31, 1383–1389. [CrossRef][PubMed] 14. Maxwell, A.D.; Wang, T.Y.; Yuan, L.; Duryea, A.P.; Xu, Z.; Cain, C.A. A tissue phantom for visualization and measurement of ultrasound-induced cavitation damage. Ultrasound Med. Biol. 2010, 36, 2132–2143. [CrossRef][PubMed] 15. Takegami, K.; Kaneko, Y.; Watanabe, T.; Maruyama, T.; Matsumoto, Y.; Nagawa, H. Polyacrylamide gel containing egg white as new model for irradiation experiments using focused ultrasound. Ultrasound Med. Biol. 2004, 30, 1419–1422. [CrossRef][PubMed] 16. Siddiqi, A.K.; Cho, S.H. Agar-based heat-sensitive gel with linear thermal response over 65–80 ◦C. J. Therm. Anal. Calorim. 2013, 111, 1805–1809. [CrossRef] 17. Skjervold, P.O.; Bencze Rørå, A.M.; Fjæra, S.O.; Vegusdal, A.; Vorre, A.; Einen, O. Effects of pre-, in-, or post-rigor filleting of live chilled Atlantic salmon. Aquaculture 2001, 194, 315–326. [CrossRef] 18. Yochev, B.; Kutzarov, S.; Ganchev, D.; Staykov, K. Investigation of Ultrasonic Properties of Hydrophilic Polymers for Dry-coupled Inspection. In Proceedings of the European Conference on Non-Destructive Testing, Berlin, Germany, 25–29 September 2006; pp. 1–10.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).