Institut für Lebensmittelwissenschaft und Biotechnologie

Permanent URI for this collectionhttps://hohpublica.uni-hohenheim.de/handle/123456789/6

Browse

Browsing Institut für Lebensmittelwissenschaft und Biotechnologie by Sustainable Development Goals "13"

Type the first few letters and click on the Browse button
Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Oral processing of anisotropic food structures: A modelling approach to dynamic mastication data
    (2024) Oppen, Dominic; Weiss, Jochen
    Materials that have been generated through a directionally oriented growth process often exhibit anisotropic properties. Plant materials such as tubers and roots or animal matter used to produce products such as steaks or pasta filata are characterized by an alignment of molecules, aggregates or cells in certain dimensions leading to differing properties depending on direction. Such an anisotropic property behavior is important for a wide range of quality attributes such as texture, appearance, stability and even aroma and taste. Especially the former is of critical importance to consumer liking and acceptance of foods. Structure-texture relationships have already been established for certain foods. For anisotropic foods though, a determination of such relationships is difficult, since the comminution of foods during chewing causes complex changes to the underlying anisotropic structure elements that are not easily measurable using conventional mechanical texture analysis tests such as cutting, shearing or compression. On the other hand, sensory tests using panels are very time consuming and often do not reveal structural causes for texture like or dislike by consumers. The lack of availability of suitable analytical techniques that allow for a description of texture properties relevant to mastication hampers especially the development of meat substitutes that are currently trending. The aim of this work was therefore to characterize changes to anisotropic structures induced by chewing (henceforth referred to as "oral processing") using a novel measurement approach that records kinematic and electromyographic properties of the chewing process. The kinematics of jaw movement were recorded using a 3D motion tracking system. Muscle activity was recorded using an electromyograph. From the measured data, characteristics for individual chews were calculated, which were represented in a linear mixed model as a function of the food structure. Section I provides the scientific basis for this work through a preface and a literature review. Grown and manufactured anisotropic foods are identified and described. A general overview of the production, phase phenomena and characterization methods for anisotropic food materials is given. Section II contains the oral processing experiments. In Chapter III, the focus was put on the impact of fiber length of grown structures on mastication behavior. Meat model systems with different microstructures but the same composition were produced. The model systems with anisotropic and isotropic microstructures were comminuted to different sizes, and the fiber length was inferred from the length of the particles, taking into account the particle size effect of chewing. The results indicate that longer fibers cause greater jaw movement and muscle activity. For instance, estimate peak muscle activity of anisotropic samples is 58.2857 µV higher (p=0.0156) compared to isotropic samples. Chapter IV describes minced meat products in which certain phase volumes were replaced by a finely comminuted meat mass. The aim of the study was to find detection limits beyond which an increase or decrease in muscle fiber cells does not lead to a further adjustment of the mastication properties. In the study, a transition point was identified at around 50 % of batter-like substances. Food models with more than 50 % of batter-like substance showed a smaller change in mastication parameters. The effect was more pronounced with higher proportions of fibrous material. Chapter V dealt with the topic of meat substitutes. A simple model of meat substitutes was used to test whether the effects found in anisotropic animal-based products can also be found in plant-based products. Hydrocolloid gels with different phase volumes of wet textured plant protein were produced. Similar effects for the animal-based products were observed, although the correlation was not as strong. It was hypothesized that a large part of the effect was due to the weak binding ability of hydrocolloid gels. Thus, the anisotropic particles could not be held together with a low proportion of the outer hydrocolloid gel and required less muscle activity despite a higher content of structured phase. Section III assessed alternative data evaluation strategies to the linear mixed model. The aim of the study in Chapter VI was to anticipate the model products from Chapter III using a classification approach. Algorithms of three categories were trained with the data set of the chewing processes. Two approaches were used to evaluate whether the algorithms could either resolve each individual food model with variations in microstructure (anisotropy) and macrostructure (particle size) or in microstructure only. For both approaches, the algorithms performed significantly better compared to a random guessing. The best classification results were achieved by the boosted ensemble learner "XGBoost", which assigned 96.617 % of all bites to the corresponding food microstructure. Furthermore, it was demonstrated that standardized and normalized oral processing data are almost not subject-dependent. In addition, feature importance analysis confirmed that lateral jaw movement is a good indicator of the presence of anisotropic food material and, with a weight of 0.39205, is the most important feature for classifying samples according to their structure. In summary, this work was able to show that the dynamic characteristics of mastication change depending on anisotropic properties. In general, modeling of mastication characteristics has never been conducted before and represents a promising advance over mean-based evaluation. The machine learning approach is also new in the field of oral processing and proved to be promising. For future research, it is proposed to correlate the dynamic features with sensory texture data to obtain direct correlations between chewing characteristics and texture attributes.