Early Development of Locomotion in the Term Piglet Model: Does Size Matter?

Intrauterine undernutrition in humans typically results in low birth weight ([small for gestational age] SGA) and delayed postnatal neuromotor maturation. Since SGA and intrauterine growth retardation are also common in domestic pigs, piglets are premised as models to study delayed motor development. Applied to the locomotor paradigm, however, questions emerge: (i) how to map the developmental time scale of the precocial model onto the altricial target species and (ii) how to distinguish size from maturation effects? Gait data were collected at self-selected voluntary walking speed during early development (0-96 hours postpartum; pp) for SGA- and normal ([appropriate for gestational age] AGA) piglets. Dimensionless spatiotemporal gait characteristics (according to dynamic similarity) become invariant already after 4 hours pp, suggesting rapid postnatal neuromotor maturation. Moreover, dimensionless gait data are largely identical for SGA- and AGA-siblings, indicating that primarily size effects explain absolute locomotor differences. This is further supported by (i) normalized force-generating capacity of limb muscles, (ii) joint kinematics (<10 hours pp), and (iii) normalized ground reaction forces (<5 days pp) being indifferent between SGA- and AGA- piglets. Furthermore, predictive modeling based on limb joint kinematics is unable to discern the majority of SGA- from AGA-piglets (<10 hours pp). All this leads to the conclusion that, although smaller than the AGA piglets in absolute terms, SGA-piglets mature (neuromechanically speaking) just like, and equally fast as their AGA littermates. Yet, it remains a fact that early SGA piglets are reported to be less mobile, less vital, and less competitive than their AGA siblings (even often die before day 3 pp). This conspicuous difference likely results from the energy level (blood glucose and glycogen) and its mobilization being considerably different between the piglet categories during early development.

A workflow for automatic, high precision livestock diagnostic screening of locomotor kinematics.

Locomotor kinematics have been challenging inputs for automated diagnostic screening of livestock. Locomotion is a highly variable behavior, and influenced by subject characteristics (e.g., body mass, size, age, disease). We assemble a set of methods from different scientific disciplines, composing an automatic, high through-put workflow which can disentangle behavioral complexity and generate precise individual indicators of non-normal behavior for application in diagnostics and research. For this study, piglets (Sus domesticus) were filmed from lateral perspective during their first 10 h of life, an age at which maturation is quick and body mass and size have major consequences for survival. We then apply deep learning methods for point digitization, calculate joint angle profiles, and apply information-preserving transformations to retrieve a multivariate kinematic data set. We train probabilistic models to infer subject characteristics from kinematics. Model accuracy was validated for strides from piglets of normal birth weight (i.e., the category it was trained on), but the models infer the body mass and size of low birth weight (LBW) piglets (which were left out of training, out-of-sample inference) to be "normal." The age of some (but not all) low birth weight individuals was underestimated, indicating developmental delay. Such individuals could be identified automatically, inspected, and treated accordingly. This workflow has potential for automatic, precise screening in livestock management.

Automatic landmark detection and mapping for 2D/3D registration with BoneNet.

The 3D musculoskeletal motion of animals is of interest for various biological studies and can be derived from X-ray fluoroscopy acquisitions by means of image matching or manual landmark annotation and mapping. While the image matching method requires a robust similarity measure (intensity-based) or an expensive computation (tomographic reconstruction-based), the manual annotation method depends on the experience of operators. In this paper, we tackle these challenges by a strategic approach that consists of two building blocks: an automated 3D landmark extraction technique and a deep neural network for 2D landmarks detection. For 3D landmark extraction, we propose a technique based on the shortest voxel coordinate variance to extract the 3D landmarks from the 3D tomographic reconstruction of an object. For 2D landmark detection, we propose a customized ResNet18-based neural network, BoneNet, to automatically detect geometrical landmarks on X-ray fluoroscopy images. With a deeper network architecture in comparison to the original ResNet18 model, BoneNet can extract and propagate feature vectors for accurate 2D landmark inference. The 3D poses of the animal are then reconstructed by aligning the extracted 2D landmarks from X-ray radiographs and the corresponding 3D landmarks in a 3D object reference model. Our proposed method is validated on X-ray images, simulated from a real piglet hindlimb 3D computed tomography scan and does not require manual annotation of landmark positions. The simulation results show that BoneNet is able to accurately detect the 2D landmarks in simulated, noisy 2D X-ray images, resulting in promising rigid and articulated parameter estimations.

Progressive tracking: a novel procedure to facilitate manual digitization of videos.

Digitization of video recordings often requires the laborious procedure of manually clicking points of interest on individual video frames. Here, we present progressive tracking, a procedure that facilitates manual digitization of markerless videos. In contrast to existing software, it allows the user to follow points of interest with a cursor in the progressing video, without the need to click. To compare the performance of progressive tracking with the conventional frame-wise tracking, we quantified speed and accuracy of both methods, testing two different input devices (mouse and stylus pen). We show that progressive tracking can be twice as fast as frame-wise tracking while maintaining accuracy, given that playback speed is controlled. Using a stylus pen can increase frame-wise tracking speed. The complementary application of the progressive and frame-wise mode is exemplified on a realistic video recording. This study reveals that progressive tracking can vastly facilitate video analysis in experimental research.

Motion analysis of non-model organisms using a hierarchical model: Influence of setup enclosure dimensions on gait parameters of Swinhoe's striped squirrels as a test case.

In in-vivo motion analyses, data from a limited number of subjects and trials is used as proxy for locomotion properties of entire populations, yet the inherent hierarchy of the individual and population level is usually not accounted for. Despite the increasing availability of hierarchical model frameworks for statistical analyses, they have not been applied extensively to comparative motion analysis. As a case study for the use of hierarchical models, we analyzed locomotor parameters of four Swinhoe's striped squirrels. The small-bodied arboreal mammals exhibit brief bouts of rapid asymmetric gaits. Spatio-temporal parameters on runways with experimentally varied dimensions of the setup enclosure were compared to test for their potentially confounding effects. We applied principal component analysis to evaluate changes to the overall locomotor pattern. A common, non-hierarchical, pooled statistical analysis of the data revealed significant differences in some of the parameters depending on enclosure dimensions. In contrast, we used a hierarchical Bayesian generalized linear model (GLM) that considers subject specific differences and population effects to compare the effect of enclosure dimensions on the measured parameters and the principal components. None of the population effects were confirmed by the hierarchical GLM. The confounding effect of a single subject that deviates in its locomotor behavior is potentially bigger than the influence of the experimental variation in enclosure dimensions. Our findings justify the common practice of researchers to intuitively select an enclosure with dimensions assumed as "non-constraining". Hierarchical models can easily be designed to cope with limited sample size and bias introduced by deviating behavior of individuals. When limited data is available-a typical restriction of in-vivo motion analyses of non-model organisms-density distributions of the Bayesian GLM used here remain reliable and the hierarchical structure of the model optimally exploits all available information. We provide code to be adjusted to other research questions.

Vocalization-whisking coordination and multisensory integration of social signals in rat auditory cortex.

Social interactions involve multi-modal signaling. Here, we study interacting rats to investigate audio-haptic coordination and multisensory integration in the auditory cortex. We find that facial touch is associated with an increased rate of ultrasonic vocalizations, which are emitted at the whisking rate (∼8 Hz) and preferentially initiated in the retraction phase of whisking. In a small subset of auditory cortex regular-spiking neurons, we observed excitatory and heterogeneous responses to ultrasonic vocalizations. Most fast-spiking neurons showed a stronger response to calls. Interestingly, facial touch-induced inhibition in the primary auditory cortex and off-responses after termination of touch were twofold stronger than responses to vocalizations. Further, touch modulated the responsiveness of auditory cortex neurons to ultrasonic vocalizations. In summary, facial touch during social interactions involves precisely orchestrated calling-whisking patterns. While ultrasonic vocalizations elicited a rather weak population response from the regular spikers, the modulation of neuronal responses by facial touch was remarkably strong.