A deep neural network to assess spontaneous pain from mouse facial expressions

Development of PainFace software to simplify, standardize, and scale up mouse grimace analyses

ABSTRACT

Facial grimacing is used to quantify spontaneous pain in mice and other mammals, but scoring relies on humans with different levels of proficiency. Here, we developed a cloud-based software platform called PainFace ( http://painface.net ) that uses machine learning to detect 4 facial action units of the mouse grimace scale (orbitals, nose, ears, whiskers) and score facial grimaces of black-coated C57BL/6 male and female mice on a 0 to 8 scale. Platform accuracy was validated in 2 different laboratories, with 3 conditions that evoke grimacing-laparotomy surgery, bilateral hindpaw injection of carrageenan, and intraplantar injection of formalin. PainFace can generate up to 1 grimace score per second from a standard 30 frames/s video, making it possible to quantify facial grimacing over time, and operates at a speed that scales with computing power. By analyzing the frequency distribution of grimace scores, we found that mice spent 7x more time in a "high grimace" state following laparotomy surgery relative to sham surgery controls. Our study shows that PainFace reproducibly quantifies facial grimaces indicative of nonevoked spontaneous pain and enables laboratories to standardize and scale-up facial grimace analyses.

From facial expressions to algorithms: a narrative review of animal pain recognition technologies

Facial expressions are essential for communication and emotional expression across species. Despite the improvements brought by tools like the Horse Grimace Scale (HGS) in pain recognition in horses, their reliance on human identification of characteristic traits presents drawbacks such as subjectivity, training requirements, costs, and potential bias. Despite these challenges, the development of facial expression pain scales for animals has been making strides. To address these limitations, Automated Pain Recognition (APR) powered by Artificial Intelligence (AI) offers a promising advancement. Notably, computer vision and machine learning have revolutionized our approach to identifying and addressing pain in non-verbal patients, including animals, with profound implications for both veterinary medicine and animal welfare. By leveraging the capabilities of AI algorithms, we can construct sophisticated models capable of analyzing diverse data inputs, encompassing not only facial expressions but also body language, vocalizations, and physiological signals, to provide precise and objective evaluations of an animal's pain levels. While the advancement of APR holds great promise for improving animal welfare by enabling better pain management, it also brings forth the need to overcome data limitations, ensure ethical practices, and develop robust ground truth measures. This narrative review aimed to provide a comprehensive overview, tracing the journey from the initial application of facial expression recognition for the development of pain scales in animals to the recent application, evolution, and limitations of APR, thereby contributing to understanding this rapidly evolving field.

Automated recognition of emotional states of horses from facial expressions

Animal affective computing is an emerging new field, which has so far mainly focused on pain, while other emotional states remain uncharted territories, especially in horses. This study is the first to develop AI models to automatically recognize horse emotional states from facial expressions using data collected in a controlled experiment. We explore two types of pipelines: a deep learning one which takes as input video footage, and a machine learning one which takes as input EquiFACS annotations. The former outperforms the latter, with 76% accuracy in separating between four emotional states: baseline, positive anticipation, disappointment and frustration. Anticipation and frustration were difficult to separate, with only 61% accuracy.

The grimace scale: a useful tool for assessing pain in laboratory animals

Accurately and promptly assessing pain in experimental animals is extremely important to avoid unnecessary suffering of the animals and to enhance the reproducibility of experiments. This is a key concern for veterinarians, animal caretakers, and researchers from the perspectives of veterinary care and animal welfare. Various methods including ethology, immunohistochemistry, electrophysiology, and molecular biology are used for pain assessment. However, the grimace scale, which was developed by taking cues from interpreting pain through facial expressions of non-verbal infants, has become recognized as a very simple and practical method for objectively evaluating pain levels by scoring changes in an animal's expressions. This method, which was first implemented with mice approximately 10 years ago, is now being applied to various experimental animals and is widely used in research settings. This review focuses on the usability of the grimace scale from the "cage-side" perspective, aiming to make it a more user-friendly tool for those involved in animal experiments. Differences in facial expressions in response to pain in various animals, examples of applying the grimace scale, current automated analytical methods, and future prospects are discussed.

Repeated stress triggers seeking of a starvation-like state in anxiety-prone female mice

Elevated anxiety often precedes anorexia nervosa and persists after weight restoration. Patients with anorexia nervosa often describe self-starvation as pleasant, potentially because food restriction can be anxiolytic. Here, we tested whether repeated stress can cause animals to prefer a starvation-like state. We developed a virtual reality place preference paradigm in which head-fixed mice can voluntarily seek a starvation-like state induced by optogenetic stimulation of hypothalamic agouti-related peptide (AgRP) neurons. Prior to stress exposure, males but not females showed a mild aversion to AgRP stimulation. Strikingly, following multiple days of stress, a subset of females developed a strong preference for AgRP stimulation that was predicted by high baseline anxiety. Such stress-induced changes in preference were reflected in changes in facial expressions during AgRP stimulation. Our study suggests that stress may cause females predisposed to anxiety to seek a starvation state and provides a powerful experimental framework for investigating the underlying neural mechanisms.

Deep Learning for Face Detection and Pain Assessment in Japanese macaques (Macaca fuscata)

Facial expressions have increasingly been used to assess emotional states in mammals. The recognition of pain in research animals is essential for their well-being and leads to more reliable research outcomes. Automating this process could contribute to early pain diagnosis and treatment. Artificial neural networks have become a popular option for image classification tasks in recent years due to the development of deep learning. In this study, we investigated the ability of a deep learning model to detect pain in Japanese macaques based on their facial expression. Thirty to 60 min of video footage from Japanese macaques undergoing laparotomy was used in the study. Macaques were recorded undisturbed in their cages before surgery (No Pain) and one day after the surgery before scheduled analgesia (Pain). Videos were processed for facial detection and image extraction with the algorithms RetinaFace (adding a bounding box around the face for image extraction) or Mask R-CNN (contouring the face for extraction). ResNet50 used 75% of the images to train systems; the other 25% were used for testing. Test accuracy varied from 48 to 54% after box extraction. The low accuracy of classification after box extraction was likely due to the incorporation of features that were not relevant for pain (for example, background, illumination, skin color, or objects in the enclosure). However, using contour extraction, preprocessing the images, and fine-tuning, the network resulted in 64% appropriate generalization. These results suggest that Mask R-CNN can be used for facial feature extractions and that the performance of the classifying model is relatively accurate for nonannotated single-frame images.

Quantitative orbital tightening for pain assessment using machine learning with DeepLabCut

Pain assessment in animal models is essential for understanding mechanisms underlying pathological pain and developing effective pain medicine. The grimace scale (GS), facial expression features in pain such as orbital tightening (OT), is a valuable measure for assessing pain in animal models. However, the classical grimace scale for pain assessment is labor-intensive, subject to subjectivity and inconsistency, and is not a quantitative measure. In the present study, we utilized machine learning with DeepLabCut to annotate the superior and inferior eyelid margins and the medial and lateral canthus of the eyes in animals' video images. Based on the annotation, we quantified the eyelid distance and palpebral fissure width of the animals' eyes so that the degree of OT in animals with pain could be measured and described quantitatively. We established criteria for the inclusion and exclusion of the annotated images for quantifying OT, and validated our quantitative grimace scale (qGS) in the mice with pain caused by capsaicin injections in the orofacial or hindpaw regions, the Nav1.8-ChR2 mice following orofacial noxious stimulation with laser light, and the oxaliplatin-treated mice following tactile stimulation with a von Frey filament. We showed that both the eyelid distance and the palpebral fissure width were shortened significantly in the animals in pain compared to the control animals without nociceptive stimulation. Collectively, the present study has established a quantitative orbital tightening for pain assessment in mice using DeepLabCut, providing a new tool for pain assessment in preclinical studies with mice.

Functional diversity of dopamine axons in prefrontal cortex during classical conditioning

Midbrain dopamine neurons impact neural processing in the prefrontal cortex (PFC) through mesocortical projections. However, the signals conveyed by dopamine projections to the PFC remain unclear, particularly at the single-axon level. Here, we investigated dopaminergic axonal activity in the medial PFC (mPFC) during reward and aversive processing. By optimizing microprism-mediated two-photon calcium imaging of dopamine axon terminals, we found diverse activity in dopamine axons responsive to both reward and aversive stimuli. Some axons exhibited a preference for reward, while others favored aversive stimuli, and there was a strong bias for the latter at the population level. Long-term longitudinal imaging revealed that the preference was maintained in reward- and aversive-preferring axons throughout classical conditioning in which rewarding and aversive stimuli were paired with preceding auditory cues. However, as mice learned to discriminate reward or aversive cues, a cue activity preference gradually developed only in aversive-preferring axons. We inferred the trial-by-trial cue discrimination based on machine learning using anticipatory licking or facial expressions, and found that successful discrimination was accompanied by sharper selectivity for the aversive cue in aversive-preferring axons. Our findings indicate that a group of mesocortical dopamine axons encodes aversive-related signals, which are modulated by both classical conditioning across days and trial-by-trial discrimination within a day.

Functional Diversity of Dopamine Axons in Prefrontal Cortex During Classical Conditioning

Midbrain dopamine neurons impact neural processing in the prefrontal cortex (PFC) through mesocortical projections. However, the signals conveyed by dopamine projections to the PFC remain unclear, particularly at the single-axon level. Here, we investigated dopaminergic axonal activity in the medial PFC (mPFC) during reward and aversive processing. By optimizing microprism-mediated two-photon calcium imaging of dopamine axon terminals, we found diverse activity in dopamine axons responsive to both reward and aversive stimuli. Some axons exhibited a preference for reward, while others favored aversive stimuli, and there was a strong bias for the latter at the population level. Long-term longitudinal imaging revealed that the preference was maintained in reward- and aversive-preferring axons throughout classical conditioning in which rewarding and aversive stimuli were paired with preceding auditory cues. However, as mice learned to discriminate reward or aversive cues, a cue activity preference gradually developed only in aversive-preferring axons. We inferred the trial-by-trial cue discrimination based on machine learning using anticipatory licking or facial expressions, and found that successful discrimination was accompanied by sharper selectivity for the aversive cue in aversive-preferring axons. Our findings indicate that a group of mesocortical dopamine axons encodes aversive-related signals, which are modulated by both classical conditioning across days and trial-by-trial discrimination within a day.

Central neuropathic pain

Central neuropathic pain arises from a lesion or disease of the central somatosensory nervous system such as brain injury, spinal cord injury, stroke, multiple sclerosis or related neuroinflammatory conditions. The incidence of central neuropathic pain differs based on its underlying cause. Individuals with spinal cord injury are at the highest risk; however, central post-stroke pain is the most prevalent form of central neuropathic pain worldwide. The mechanisms that underlie central neuropathic pain are not fully understood, but the pathophysiology likely involves intricate interactions and maladaptive plasticity within spinal circuits and brain circuits associated with nociception and antinociception coupled with neuronal hyperexcitability. Modulation of neuronal activity, neuron-glia and neuro-immune interactions and targeting pain-related alterations in brain connectivity, represent potential therapeutic approaches. Current evidence-based pharmacological treatments include antidepressants and gabapentinoids as first-line options. Non-pharmacological pain management options include self-management strategies, exercise and neuromodulation. A comprehensive pain history and clinical examination form the foundation of central neuropathic pain classification, identification of potential risk factors and stratification of patients for clinical trials. Advanced neurophysiological and neuroimaging techniques hold promise to improve the understanding of mechanisms that underlie central neuropathic pain and as predictive biomarkers of treatment outcome.

Fully automated deep learning models with smartphone applicability for prediction of pain using the Feline Grimace Scale

This study used deep neural networks and machine learning models to predict facial landmark positions and pain scores using the Feline Grimace Scale© (FGS). A total of 3447 face images of cats were annotated with 37 landmarks. Convolutional neural networks (CNN) were trained and selected according to size, prediction time, predictive performance (normalized root mean squared error, NRMSE) and suitability for smartphone technology. Geometric descriptors (n = 35) were computed. XGBoost models were trained and selected according to predictive performance (accuracy; mean square error, MSE). For prediction of facial landmarks, the best CNN model had NRMSE of 16.76% (ShuffleNetV2). For prediction of FGS scores, the best XGBoost model had accuracy of 95.5% and MSE of 0.0096. Models showed excellent predictive performance and accuracy to discriminate painful and non-painful cats. This technology can now be used for the development of an automated, smartphone application for acute pain assessment in cats.

Protease-Activated Receptor 2 (PAR2) Expressed in Sensory Neurons Contributes to Signs of Pain and Neuropathy in Paclitaxel Treated Mice

Chemotherapy-induced peripheral neuropathy (CIPN) is a common, dose-limiting side effect of cancer therapy. Protease-activated receptor 2 (PAR2) is implicated in a variety of pathologies, including CIPN. In this study, we demonstrate the role of PAR2 expressed in sensory neurons in a paclitaxel (PTX)-induced model of CIPN in mice. PAR2 knockout/wildtype (WT) mice and mice with PAR2 ablated in sensory neurons were treated with PTX administered via intraperitoneal injection. In vivo behavioral studies were done in mice using von Frey filaments and the Mouse Grimace Scale. We then examined immunohistochemical staining of dorsal root ganglion (DRG) and hind paw skin samples from CIPN mice to measure satellite cell gliosis and intra-epidermal nerve fiber (IENF) density. The pharmacological reversal of CIPN pain was tested with the PAR2 antagonist C781. Mechanical allodynia caused by PTX treatment was alleviated in PAR2 knockout mice of both sexes. In the PAR2 sensory neuronal conditional knockout (cKO) mice, both mechanical allodynia and facial grimacing were attenuated in mice of both sexes. In the DRG of the PTX-treated PAR2 cKO mice, satellite glial cell activation was reduced compared to control mice. IENF density analysis of the skin showed that the PTX-treated control mice had a reduction in nerve fiber density while the PAR2 cKO mice had a comparable skin innervation as the vehicle-treated animals. Similar results were seen with satellite cell gliosis in the DRG, where gliosis induced by PTX was absent in PAR cKO mice. Finally, C781 was able to transiently reverse established PTX-evoked mechanical allodynia. PERSPECTIVE: Our work demonstrates that PAR2 expressed in sensory neurons plays a key role in PTX-induced mechanical allodynia, spontaneous pain, and signs of neuropathy, suggesting PAR2 as a possible therapeutic target in multiple aspects of PTX CIPN.

An automated rat grimace scale for the assessment of pain

Pain is a complex neuro-psychosocial experience that is internal and private, making it difficult to assess in both humans and animals. In pain research, animal models are prominently used, with rats among the most commonly studied. The rat grimace scale (RGS) measures four facial action units to quantify the pain behaviors of rats. However, manual recording of RGS scores is a time-consuming process that requires training. While computer vision models have been developed and utilized for various grimace scales, there are currently no models for RGS. To address this gap, this study worked to develop an automated RGS system which can detect facial action units in rat images and predict RGS scores. The automated system achieved an action unit detection precision and recall of 97%. Furthermore, the action unit RGS classifiers achieved a weighted accuracy of 81-93%. The system's performance was evaluated using a blast traumatic brain injury study, where it was compared to trained human graders. The results showed an intraclass correlation coefficient of 0.82 for the total RGS score, indicating that the system was comparable to human graders. The automated tool could enhance pain research by providing a standardized and efficient method for the assessment of RGS.

Deep learning for video-based automated pain recognition in rabbits

Despite the wide range of uses of rabbits (Oryctolagus cuniculus) as experimental models for pain, as well as their increasing popularity as pets, pain assessment in rabbits is understudied. This study is the first to address automated detection of acute postoperative pain in rabbits. Using a dataset of video footage of n = 28 rabbits before (no pain) and after surgery (pain), we present an AI model for pain recognition using both the facial area and the body posture and reaching accuracy of above 87%. We apply a combination of 1 sec interval sampling with the Grayscale Short-Term stacking (GrayST) to incorporate temporal information for video classification at frame level and a frame selection technique to better exploit the availability of video data.

Classification of multiple emotional states from facial expressions in head-fixed mice using a deep learning-based image analysis

Facial expressions are widely recognized as universal indicators of underlying internal states in most species of animals, thereby presenting as a non-invasive measure for assessing physical and mental conditions. Despite the advancement of artificial intelligence-assisted tools for automated analysis of voluminous facial expression data in human subjects, the corresponding tools for mice still remain limited so far. Considering that mice are the most prevalent model animals for studying human health and diseases, a comprehensive characterization of emotion-dependent patterns of facial expressions in mice could extend our knowledge on the basis of emotions and the related disorders. Here, we present a framework for the development of a deep learning-powered tool for classifying facial expressions in head-fixed mouse. We demonstrate that our machine vision was capable of accurately classifying three different emotional states from lateral facial images in head-fixed mouse. Moreover, we objectively determined how our classifier characterized the differences among the facial images through the use of an interpretation technique called Gradient-weighted Class Activation Mapping. Importantly, our machine vision presumably discerned the data by leveraging multiple facial features. Our approach is likely to facilitate the non-invasive decoding of a variety of emotions from facial images in head-fixed mice.

Spared nerve injury causes motor phenotypes unrelated to pain in mice

Most animal models of neuropathic pain use targeted nerve injuries quantified with motor reflexive measures in response to an applied noxious stimulus. These motor reflexive measures can only accurately represent a pain response if motor function in also intact. The commonly used spared nerve injury (SNI) model, however, damages the tibial and common peroneal nerves that should result in motor phenotypes (i.e., an immobile or "flail" foot) not typically captured in sensory assays. To test the extent of these issues, we used DeepLabCut, a deep learning-based markerless pose estimation tool to quantify spontaneous limb position in C57BL/6J mice during tail suspension following either SNI or sham surgery. Using this granular detail, we identified the expected flail foot-like impairment, but we also found SNI mice hold their injured limb closer to the body midline compared to shams. These phenotypes were not present in the Complete Freunds Adjuvant model of inflammatory pain and were not reversed by multiple analgesics with different mechanisms of action, suggesting these SNI-specific phenotypes are not directly related to pain. Together these results suggest SNI causes previously undescribed phenotypes unrelated to altered sensation that are likely underappreciated while interpreting preclinical pain research outcomes.

Post-injury pain and behaviour: a control theory perspective

Injuries of various types occur commonly in the lives of humans and other animals and lead to a pattern of persistent pain and recuperative behaviour that allows safe and effective recovery. In this Perspective, we propose a control-theoretic framework to explain the adaptive processes in the brain that drive physiological post-injury behaviour. We set out an evolutionary and ethological view on how animals respond to injury, illustrating how the behavioural state associated with persistent pain and recuperation may be just as important as phasic pain in ensuring survival. Adopting a normative approach, we suggest that the brain implements a continuous optimal inference of the current state of injury from diverse sensory and physiological signals. This drives the various effector control mechanisms of behavioural homeostasis, which span the modulation of ongoing motivation and perception to drive rest and hyper-protective behaviours. However, an inherent problem with this is that these protective behaviours may partially obscure information about whether injury has resolved. Such information restriction may seed a tendency to aberrantly or persistently infer injury, and may thus promote the transition to pathological chronic pain states.

Preclinical orofacial pain assays and measures and chronic primary orofacial pain research: where we are and where we need to go

Chronic primary orofacial pain (OFP) conditions such as painful temporomandibular disorders (pTMDs; i.e., myofascial pain and arthralgia), idiopathic trigeminal neuralgia (TN), and burning mouth syndrome (BMS) are seemingly idiopathic, but evidence support complex and multifactorial etiology and pathophysiology. Important fragments of this complex array of factors have been identified over the years largely with the help of preclinical studies. However, findings have yet to translate into better pain care for chronic OFP patients. The need to develop preclinical assays that better simulate the etiology, pathophysiology, and clinical symptoms of OFP patients and to assess OFP measures consistent with their clinical symptoms is a challenge that needs to be overcome to support this translation process. In this review, we describe rodent assays and OFP pain measures that can be used in support of chronic primary OFP research, in specific pTMDs, TN, and BMS. We discuss their suitability and limitations considering the current knowledge of the etiology and pathophysiology of these conditions and suggest possible future directions. Our goal is to foster the development of innovative animal models with greater translatability and potential to lead to better care for patients living with chronic primary OFP.

Poor Association between Facial Expression and Mild Lameness in Thoroughbred Trot-Up Examinations

Musculoskeletal injuries in racehorses are difficult to detect prior to catastrophic breakdown. Lameness is commonly attributed to orthopaedic pain in horses, therefore, subtle lameness may be a pre-clinical sign of injury and, if identified early, could allow for preventative intervention. Our objective was to determine if facial expressions could be used to detect mild lameness as an indicator of orthopaedic pain in 'fit to race' horses. The Horse Grimace Scale (HGS) and the facial expressions in ridden horses (FEReq), were used to score images (n = 380) of mildly lame (n = 21) and non-lame (n = 17) Thoroughbred horses by two independent observers. Using an Equinosis Lameness Locator®, the lameness status of each horse was determined according to published thresholds [forelimb lameness (>|14.5 mm|) and hindlimb lameness (>|7.5 mm|)]. Inter and intraobserver reliability were assessed using two-way random-effects models. Univariable associations between lameness and facial expression parameters were identified using logistic and linear regression. Interobserver reliability was moderate (κ 0.45; 95% CI 0.36, 0.55). Horses with moderate mouth strain (HGS) and tense and extended upper lip (FEReq) were less likely to be lame (p = 0.042 and p = 0.027, respectively). Exposed sclera was associated with lameness (p = 0.045). Higher orbital tightening (HGS) scores were associated with a lower degree of maximum head amplitude (HDmax) lameness (p = 0.044). Tension and moderate tension above the eye, for the HGS and FEReq scores, were associated with increasing amplitude of HDmax (p = 0.048 and p = 0.034, respectively). Inconsistent associations between lameness status and HGS and FEReq scores may limit the potential use of the facial expression for the prediction of mild orthopaedic pain during pre-race lameness examinations. More objective parameters associated with mild orthopaedic pain should be explored.

Chronic stress triggers seeking of a starvation-like state in anxiety-prone female mice

Elevated anxiety often precedes anorexia nervosa and persists after weight restoration. Patients with anorexia nervosa often describe hunger as pleasant, potentially because food restriction can be anxiolytic. Here, we tested whether chronic stress can cause animals to prefer a starvation-like state. We developed a virtual reality place preference paradigm in which head-fixed mice can voluntarily seek a starvation-like state induced by optogenetic stimulation of hypothalamic agouti-related peptide (AgRP) neurons. Prior to stress induction, male but not female mice showed mild aversion to AgRP stimulation. Strikingly, following chronic stress, a subset of females developed a strong preference for AgRP stimulation that was predicted by high baseline anxiety. Such stress-induced changes in preference were reflected in changes in facial expressions during AgRP stimulation. Our study suggests that stress may cause females predisposed to anxiety to seek a starvation state, and provides a powerful experimental framework for investigating the underlying neural mechanisms.

C781, a β-Arrestin Biased Antagonist at Protease-Activated Receptor-2 (PAR2), Displays in vivo Efficacy Against Protease-Induced Pain in Mice

Given the limited options and often harmful side effects of current analgesics and the suffering caused by the opioid crisis, new classes of pain therapeutics are needed. Protease-activated receptors (PARs), particularly PAR2, are implicated in a variety of pathologies, including pain. Since the discovery of the role of PAR2 in pain, development of potent and specific antagonists has been slow. In this study, we describe the in vivo characterization of a novel small molecule/peptidomimetic hybrid compound, C781, as a β-arrestin-biased PAR2 antagonist. In vivo behavioral studies were done in mice using von Frey filaments and the Mouse Grimace Scale. Pharmacokinetic studies were done to assess pharmacokinetic/pharmacodynamic relationship in vivo. We used both prevention and reversal paradigms with protease treatment to determine whether C781 could attenuate protease-evoked pain. C781 effectively prevented and reversed mechanical and spontaneous nociceptive behaviors in response to small molecule PAR2 agonists, mast cell activators, and neutrophil elastase. The ED50 of C781 (intraperitoneal dosing) for inhibition of PAR2 agonist (20.9 ng 2-AT)-evoked nociception was 6.3 mg/kg. C781 was not efficacious in the carrageenan inflammation model. Pharmacokinetic studies indicated limited long-term systemic bioavailability for C781 suggesting that optimizing pharmacokinetic properties could improve in vivo efficacy. Our work demonstrates in vivo efficacy of a biased PAR2 antagonist that selectively inhibits β-arrestin/MAPK signaling downstream of PAR2. Given the importance of this signaling pathway in PAR2-evoked nociception, C781 exemplifies a key pharmacophore for PAR2 that can be optimized for clinical development. PERSPECTIVE: Our work provides evidence that PAR2 antagonists that only block certain aspects of signaling by the receptor can be effective for blocking protease-evoked pain in mice. This is important because it creates a rationale for developing safer PAR2-targeting approaches for pain treatment.

Intervertebral disc degeneration and how it leads to low back pain

The purpose of this review was to evaluate data generated by animal models of intervertebral disc (IVD) degeneration published in the last decade and show how this has made invaluable contributions to the identification of molecular events occurring in and contributing to pain generation. IVD degeneration and associated spinal pain is a complex multifactorial process, its complexity poses difficulties in the selection of the most appropriate therapeutic target to focus on of many potential candidates in the formulation of strategies to alleviate pain perception and to effect disc repair and regeneration and the prevention of associated neuropathic and nociceptive pain. Nerve ingrowth and increased numbers of nociceptors and mechanoreceptors in the degenerate IVD are mechanically stimulated in the biomechanically incompetent abnormally loaded degenerate IVD leading to increased generation of low back pain. Maintenance of a healthy IVD is, thus, an important preventative measure that warrants further investigation to preclude the generation of low back pain. Recent studies with growth and differentiation factor 6 in IVD puncture and multi-level IVD degeneration models and a rat xenograft radiculopathy pain model have shown it has considerable potential in the prevention of further deterioration in degenerate IVDs, has regenerative properties that promote recovery of normal IVD architectural functional organization and inhibits the generation of inflammatory mediators that lead to disc degeneration and the generation of low back pain. Human clinical trials are warranted and eagerly anticipated with this compound to assess its efficacy in the treatment of IVD degeneration and the prevention of the generation of low back pain.

Minocycline Prevents the Development of Key Features of Inflammation and Pain in DSS-induced Colitis in Mice

Abdominal pain is a common feature in inflammatory bowel disease (IBD) patients, and greatly compromises their quality of life. Therefore, the identification of new therapeutic tools to reduce visceral pain is one of the main goals for IBD therapy. Minocycline, a broad-spectrum tetracycline antibiotic, has gained attention in the scientific community because of its immunomodulatory and anti-inflammatory properties. The aim of this study was to evaluate the potential of this antibiotic as a therapy for the management of visceral pain in dextran sodium sulfate (DSS)-induced colitis in mice. Preemptive treatment with minocycline markedly reduced histological features of intestinal inflammation and the expression of inflammatory markers (Tlr4, Tnfα, Il1ß, Ptgs2, Inos, Cxcl2, and Icam1), and attenuated the decrease of markers of epithelial integrity (Tjp1, Ocln, Muc2, and Muc3). In fact, minocycline restored normal epithelial permeability in colitic mice. Treatment with the antibiotic also reversed the changes in the gut microbiota profile induced by colitis. All these ameliorative effects of minocycline on both inflammation and dysbiosis correlated with a decrease in ongoing pain and referred hyperalgesia, and with the improvement of physical activity induced by the antibiotic in colitic mice. Minocycline might constitute a new therapeutic approach for the treatment of IBD-induced pain. PERSPECTIVE: This study found that the intestinal anti-inflammatory effects of minocycline ameliorate DSS-associated pain in mice. Therefore, minocycline might constitute a novel therapeutic strategy for the treatment of IBD-induced pain.

Rodent Animal Models of Endometriosis-Associated Pain: Unmet Needs and Resources Available for Improving Translational Research in Endometriosis

Chronic pain induced by endometriosis is a maladaptive pain experienced by half of women with this disease. The lack of pharmacological treatments suitable for the long-term relief of endometriosis-associated pain, without an impact on fertility, remains an urgent unmet need. Progress has been slowed by the absence of a reproducible rodent endometriosis model that fully replicates human physiopathological characteristics, including pain symptoms. Although pain assessment in rodents is a complicated task requiring qualified researchers, the choice of the behavioral test is no less important, since selecting inappropriate tests can cause erroneous data. Pain is usually measured with reflex tests in which hypersensitivity is evaluated by applying a noxious stimulus, yet this ignores the associated emotional component that could be evaluated via non-reflex tests. We conducted a systematic review of endometriosis models used in rodents and the number of them that studied pain. The type of behavioral test used was also analyzed and classified according to reflex and non-reflex tests. Finally, we determined the most used reflex tests for the study of endometriosis-induced pain and the main non-reflex behavioral tests utilized in visceral pain that can be extrapolated to the study of endometriosis and complement traditional reflex tests.

Pain Recognition in Rodents

Available methods for recognizing and assessing pain in rodents have increased over the last 10 years, including the development of validated pain assessment scales. Much of this work has been driven by the needs of biomedical research, and there are specific challenges to applying these scales in the clinical environment. This article provides an introduction to pain assessment scale validation, reviews current methods of pain assessment, highlighting their strengths and weaknesses, and makes recommendations for assessing pain in a clinical environment.

Explainable automated recognition of emotional states from canine facial expressions: the case of positive anticipation and frustration

In animal research, automation of affective states recognition has so far mainly addressed pain in a few species. Emotional states remain uncharted territories, especially in dogs, due to the complexity of their facial morphology and expressions. This study contributes to fill this gap in two aspects. First, it is the first to address dog emotional states using a dataset obtained in a controlled experimental setting, including videos from (n = 29) Labrador Retrievers assumed to be in two experimentally induced emotional states: negative (frustration) and positive (anticipation). The dogs' facial expressions were measured using the Dogs Facial Action Coding System (DogFACS). Two different approaches are compared in relation to our aim: (1) a DogFACS-based approach with a two-step pipeline consisting of (i) a DogFACS variable detector and (ii) a positive/negative state Decision Tree classifier; (2) An approach using deep learning techniques with no intermediate representation. The approaches reach accuracy of above 71% and 89%, respectively, with the deep learning approach performing better. Secondly, this study is also the first to study explainability of AI models in the context of emotion in animals. The DogFACS-based approach provides decision trees, that is a mathematical representation which reflects previous findings by human experts in relation to certain facial expressions (DogFACS variables) being correlates of specific emotional states. The deep learning approach offers a different, visual form of explainability in the form of heatmaps reflecting regions of focus of the network's attention, which in some cases show focus clearly related to the nature of particular DogFACS variables. These heatmaps may hold the key to novel insights on the sensitivity of the network to nuanced pixel patterns reflecting information invisible to the human eye.

Clinical Assessments of Fracture Healing and Basic Science Correlates: Is There Room for Convergence?

PURPOSE OF REVIEW

The purpose of this review is to summarize the clinical and basic science methods used to assess fracture healing and propose a framework to improve the translational possibilities.

RECENT FINDINGS

Mainstays of fracture healing assessment include clinical examination, various imaging modalities, and assessment of function. Pre-clinical studies have yielded insight into biomechanical progression as well as the genetic, molecular, and cellular processes of fracture healing. Efforts are emerging to identify early markers to predict impaired healing and possibly early intervention to alter these processes. Despite of the differences in clinical and preclinical research, opportunities exist to unify and improve the translational efforts between these arenas to develop and optimize our ability to assess and predict fracture healing, thereby improving the clinical care of these patients.

Going Deeper than Tracking: A Survey of Computer-Vision Based Recognition of Animal Pain and Emotions

Advances in animal motion tracking and pose recognition have been a game changer in the study of animal behavior. Recently, an increasing number of works go ‘deeper’ than tracking, and address automated recognition of animals’ internal states such as emotions and pain with the aim of improving animal welfare, making this a timely moment for a systematization of the field. This paper provides a comprehensive survey of computer vision-based research on recognition of pain and emotional states in animals, addressing both facial and bodily behavior analysis. We summarize the efforts that have been presented so far within this topic—classifying them across different dimensions, highlight challenges and research gaps, and provide best practice recommendations for advancing the field, and some future directions for research.

Grimace scale, burrowing, and nest building for the assessment of post-surgical pain in mice and rats-A systematic review

Several studies suggested an informative value of behavioral and grimace scale parameters for the detection of pain. However, the robustness and reliability of the parameters as well as the current extent of implementation are still largely unknown. In this study, we aimed to systematically analyze the current evidence-base of grimace scale, burrowing, and nest building for the assessment of post-surgical pain in mice and rats. The following platforms were searched for relevant articles: PubMed, Embase via Ovid, and Web of Science. Only full peer-reviewed studies that describe the grimace scale, burrowing, and/or nest building as pain parameters in the post-surgical phase in mice and/or rats were included. Information about the study design, animal characteristics, intervention characteristics, and outcome measures was extracted from identified publications. In total, 74 papers were included in this review. The majority of studies have been conducted in young adult C57BL/6J mice and Sprague Dawley and Wistar rats. While there is an apparent lack of information about young animals, some studies that analyzed the grimace scale in aged rats were identified. The majority of studies focused on laparotomy-associated pain. Only limited information is available about other types of surgical interventions. While an impact of surgery and an influence of analgesia were rather consistently reported in studies focusing on grimace scales, the number of studies that assessed respective effects was rather low for nest building and burrowing. Moreover, controversial findings were evident for the impact of analgesics on post-surgical nest building activity. Regarding analgesia, a monotherapeutic approach was identified in the vast majority of studies with non-steroidal anti-inflammatory (NSAID) drugs and opioids being most commonly used. In conclusion, most evidence exists for grimace scales, which were more frequently used to assess post-surgical pain in rodents than the other behavioral parameters. However, our findings also point to relevant knowledge gaps concerning the post-surgical application in different strains, age levels, and following different surgical procedures. Future efforts are also necessary to directly compare the sensitivity and robustness of different readout parameters applied for the assessment of nest building and burrowing activities.

Next generation behavioral sequencing for advancing pain quantification

With symptoms such as spontaneous pain and pathologically heightened sensitivity to stimuli, chronic pain accounts for about 20% of physician visits and up to 2/3 of patients are dissatisfied with current treatments. Much of our knowledge on pain processing and analgesics has emerged from behavioral studies performed on animals presenting the same symptoms under pathological conditions. While humans can verbally describe their pain, studies on rodents have relied on behavioral assays providing non-exhaustive characterization or altering animals' original sensitivity through repetitive stimulations. The emergence of what we term "next-generation behavioral sequencing" is now permitting us to quantitatively describe behavioral features on millisecond to minutes long timescales that lie beyond easy detection with the unaided eye. Here, we summarize emerging videography and computational based behavioral approaches that have the potential to significantly improve pain research.

The Impact of Activity-Based Interventions on Neuropathic Pain in Experimental Spinal Cord Injury

Physical activity-based rehabilitative interventions represent the main treatment concept for people suffering from spinal cord injury (SCI). The role such interventions play in the relief of neuropathic pain (NP) states is emerging, along with underlying mechanisms resulting in SCI-induced NP (SCI-NP). Animal models have been used to investigate the benefits of activity-based interventions (ABI), such as treadmill training, wheel running, walking, swimming, and bipedal standing. These activity-based paradigms have been shown to modulate inflammatory-related alterations as well as induce functional and structural changes in the spinal cord gray matter circuitry correlated with pain behaviors. Thus far, the research available provides an incomplete picture of the cellular and molecular pathways involved in this beneficial effect. Continued research is essential for understanding how such interventions benefit SCI patients suffering from NP and allow the development of individualized rehabilitative therapies. This article reviews preclinical studies on this specific topic, goes over mechanisms involved in SCI-NP in relation to ABI, and then discusses the effectiveness of different activity-based paradigms as they relate to different forms, intensity, initiation times, and duration of ABI. This article also summarizes the mechanisms of respective interventions to ameliorate NP after SCI and provides suggestions for future research directions.

Machine learning and deep learning frameworks for the automated analysis of pain and opioid withdrawal behaviors

The automation of behavioral tracking and analysis in preclinical research can serve to advance the rate of research outcomes, increase experimental scalability, and challenge the scientific reproducibility crisis. Recent advances in the efficiency, accuracy, and accessibility of deep learning (DL) and machine learning (ML) frameworks are enabling this automation. As the ongoing opioid epidemic continues to worsen alongside increasing rates of chronic pain, there are ever-growing needs to understand opioid use disorders (OUDs) and identify non-opioid therapeutic options for pain. In this review, we examine how these related needs can be advanced by the development and validation of DL and ML resources for automated pain and withdrawal behavioral tracking. We aim to emphasize the utility of these tools for automated behavioral analysis, and we argue that currently developed models should be deployed to address novel questions in the fields of pain and OUD research.

Automated detection of squint as a sensitive assay of sex-dependent calcitonin gene-related peptide and amylin-induced pain in mice

ABSTRACT

We developed an automated squint assay using both black C57BL/6J and white CD1 mice to measure the interpalpebral fissure area between the upper and lower eyelids as an objective quantification of pain. The automated software detected a squint response to the commonly used nociceptive stimulus formalin in C57BL/6J mice. After this validation, we used the automated assay to detect a dose-dependent squint response to a migraine trigger, the neuropeptide calcitonin gene-related peptide, including a response in female mice at a dose below detection by the manual grimace scale. Finally, we found that the calcitonin gene-related peptide amylin induced squinting behavior in female mice, but not males. These data demonstrate that an automated squint assay can be used as an objective, real-time, continuous-scale measure of pain that provides higher precision and real-time analysis compared with manual grimace assessments.

A model-specific simplification of the Mouse Grimace Scale based on the pain response of intraperitoneal CCl4 injections

Despite its long establishment and applicability in mice pain detection, the Mouse Grimace Scale still seems to be underused in acute pain detection during chronic experiments. However, broadening its applicability can identify possible refinement approaches such as cumulative severity and habituation to painful stimuli. Therefore, this study focuses on two main aspects: First, five composite MGS criteria were evaluated with two independent methods (the MoBPs algorithm and a penalized least squares regression) and ranked for their relative importance. The most important variable was used in a second analysis to specifically evaluate the context of pain after an i.p. injection (intervention) in two treatment groups (CCl₄ and oil (control)) at fixed times throughout four weeks in 24 male C57BL/6 N mice. One hour before and after each intervention, video recordings were taken, and the MGS assessment was performed. In this study, the results indicate orbital tightening as the most important criterion. In this experimental setup, a highly significant difference after treatment between week 0 and 1 was found in the CCl₄ group, resulting in a medium-sized effect (W = 62.5, p value < 0.0001, r_CCl4 = 0.64). The oil group showed no significant difference (week 0 vs 1, W = 291.5, p value = 0.7875, r_control = 0.04). Therefore, the study showed that the pain caused by i.p. injections was only dependent on the applied substance, and no significant cumulation or habituation occurred due to the intervention. Further, the results indicated that the MGS system can be simplified.

Automated recognition of pain in cats

Facial expressions in non-human animals are closely linked to their internal affective states, with the majority of empirical work focusing on facial shape changes associated with pain. However, existing tools for facial expression analysis are prone to human subjectivity and bias, and in many cases also require special expertise and training. This paper presents the first comparative study of two different paths towards automatizing pain recognition in facial images of domestic short haired cats (n = 29), captured during ovariohysterectomy at different time points corresponding to varying intensities of pain. One approach is based on convolutional neural networks (ResNet50), while the other-on machine learning models based on geometric landmarks analysis inspired by species specific Facial Action Coding Systems (i.e. catFACS). Both types of approaches reach comparable accuracy of above 72%, indicating their potential usefulness as a basis for automating cat pain detection from images.

Measuring Mouse Somatosensory Reflexive Behaviors with High-speed Videography, Statistical Modeling, and Machine Learning

Objectively measuring and interpreting an animal's sensory experience remains a challenging task. This is particularly true when using preclinical rodent models to study pain mechanisms and screen for potential new pain treatment reagents. How to determine their pain states in a precise and unbiased manner is a hurdle that the field will need to overcome. Here, we describe our efforts to measure mouse somatosensory reflexive behaviors with greatly improved precision by high-speed video imaging. We describe how coupling sub-second ethograms of reflexive behaviors with a statistical reduction method and supervised machine learning can be used to create a more objective quantitative mouse "pain scale." Our goal is to provide the readers with a protocol of how to integrate some of the new tools described here with currently used mechanical somatosensory assays, while discussing the advantages and limitations of this new approach.

Deep Learning-Based Grimace Scoring Is Comparable to Human Scoring in a Mouse Migraine Model

Pain assessment is essential for preclinical and clinical studies on pain. The mouse grimace scale (MGS), consisting of five grimace action units, is a reliable measurement of spontaneous pain in mice. However, MGS scoring is labor-intensive and time-consuming. Deep learning can be applied for the automatic assessment of spontaneous pain. We developed a deep learning model, the DeepMGS, that automatically crops mouse face images, predicts action unit scores and total scores on the MGS, and finally infers whether pain exists. We then compared the performance of DeepMGS with that of experienced and apprentice human scorers. The DeepMGS achieved an accuracy of 70–90% in identifying the five action units of the MGS, and its performance (correlation coefficient = 0.83) highly correlated with that of an experienced human scorer in total MGS scores. In classifying pain and no pain conditions, the DeepMGS is comparable to the experienced human scorer and superior to the apprentice human scorers. Heatmaps generated by gradient-weighted class activation mapping indicate that the DeepMGS accurately focuses on MGS-relevant areas in mouse face images. These findings support that the DeepMGS can be applied for quantifying spontaneous pain in mice, implying its potential application for predicting other painful conditions from facial images.

Sharing pain: Using pain domain transfer for video recognition of low grade orthopedic pain in horses

Orthopedic disorders are common among horses, often leading to euthanasia, which often could have been avoided with earlier detection. These conditions often create varying degrees of subtle long-term pain. It is challenging to train a visual pain recognition method with video data depicting such pain, since the resulting pain behavior also is subtle, sparsely appearing, and varying, making it challenging for even an expert human labeller to provide accurate ground-truth for the data. We show that a model trained solely on a dataset of horses with acute experimental pain (where labeling is less ambiguous) can aid recognition of the more subtle displays of orthopedic pain. Moreover, we present a human expert baseline for the problem, as well as an extensive empirical study of various domain transfer methods and of what is detected by the pain recognition method trained on clean experimental pain in the orthopedic dataset. Finally, this is accompanied with a discussion around the challenges posed by real-world animal behavior datasets and how best practices can be established for similar fine-grained action recognition tasks. Our code is available at https://github.com/sofiabroome/painface-recognition.

A functional subdivision within the somatosensory system and its implications for pain research

Somatosensory afferents are traditionally classified by soma size, myelination, and their response specificity to external and internal stimuli. Here, we propose the functional subdivision of the nociceptive somatosensory system into two branches. The exteroceptive branch detects external threats and drives reflexive-defensive reactions to prevent or limit injury. The interoceptive branch senses the disruption of body integrity, produces tonic pain with strong aversive emotional components, and drives self-caring responses toward to the injured region to reduce suffering. The central thesis behind this functional subdivision comes from a reflection on the dilemma faced by the pain research field, namely, the use of reflexive-defensive behaviors as surrogate assays for interoceptive tonic pain. The interpretation of these assays is now being challenged by the discovery of distinct but interwoven circuits that drive exteroceptive versus interoceptive types of behaviors, with the conflation of these two components contributing partially to the poor translation of therapies from preclinical studies.

Innovations and advances in modelling and measuring pain in animals

Best practices in preclinical algesiometry (pain behaviour testing) have shifted over the past decade as a result of technological advancements, the continued dearth of translational progress and the emphasis that funding institutions and journals have placed on rigour and reproducibility. Here we describe the changing trends in research methods by analysing the methods reported in preclinical pain publications from the past 40 years, with a focus on the last 5 years. We also discuss how the status quo may be hampering translational success. This discussion is centred on four fundamental decisions that apply to every pain behaviour experiment: choice of subject (model organism), choice of assay (pain-inducing injury), laboratory environment and choice of outcome measures. Finally, we discuss how human tissues, which are increasingly accessible, can be used to validate the translatability of targets and mechanisms identified in animal pain models.

Engaging endogenous opioid circuits in pain affective processes

The pervasive use of opioid compounds for pain relief is rooted in their utility as one of the most effective therapeutic strategies for providing analgesia. While the detrimental side effects of these compounds have significantly contributed to the current opioid epidemic, opioids still provide millions of patients with reprieve from the relentless and agonizing experience of pain. The human experience of pain has long recognized the perceived unpleasantness entangled with a unique sensation that is immediate and identifiable from the first-person subjective vantage point as "painful." From this phenomenological perspective, how is it that opioids interfere with pain perception? Evidence from human lesion, neuroimaging, and preclinical functional neuroanatomy approaches is sculpting the view that opioids predominately alleviate the affective or inferential appraisal of nociceptive neural information. Thus, opioids weaken pain-associated unpleasantness rather than modulate perceived sensory qualities. Here, we discuss the historical theories of pain to demonstrate how modern neuroscience is revisiting these ideas to deconstruct the brain mechanisms driving the emergence of aversive pain perceptions. We further detail how targeting opioidergic signaling within affective or emotional brain circuits remains a strong avenue for developing targeted pharmacological and gene-therapy analgesic treatments that might reduce the dependence on current clinical opioid options.

Preclinical models of deep craniofacial nociception and temporomandibular disorder pain

Chronic pain in temporomandibular disorder (TMD) is a common health problem. Cumulating evidence indicates that the etiology of TMD pain is complex with multifactorial experience that could hamper the developments of treatments. Preclinical research is a resource to understand the mechanism for TMD pain, whereas limitations are present as a disease-specific model. It is difficult to incorporate multiple risk factors associated with the etiology that could increase pain responses into a single animal. This article introduces several rodent models which are often employed in the preclinical studies and discusses their validities for TMD pain after the elucidations of the neural mechanisms based on the clinical reports. First, rodent models were classified into two groups with or without inflammation in the deep craniofacial tissues. Next, the characteristics of each model and the procedures to identify deep craniofacial pain were discussed. Emphasis was directed on the findings of the effects of chronic psychological stress, a major risk factor for chronic pain, on the deep craniofacial nociception. Preclinical models have provided clinically relevant information, which could contribute to better understand the basis for TMD pain, while efforts are still required to bridge the gap between animal and human studies.

Brain circuits for pain and its treatment

[Figure: see text].

Pain assessment in horses using automatic facial expression recognition through deep learning-based modeling

The aim of this study was to develop and evaluate a machine vision algorithm to assess the pain level in horses, using an automatic computational classifier based on the Horse Grimace Scale (HGS) and trained by machine learning method. The use of the Horse Grimace Scale is dependent on a human observer, who most of the time does not have availability to evaluate the animal for long periods and must also be well trained in order to apply the evaluation system correctly. In addition, even with adequate training, the presence of an unknown person near an animal in pain can result in behavioral changes, making the evaluation more complex. As a possible solution, the automatic video-imaging system will be able to monitor pain responses in horses more accurately and in real-time, and thus allow an earlier diagnosis and more efficient treatment for the affected animals. This study is based on assessment of facial expressions of 7 horses that underwent castration, collected through a video system positioned on the top of the feeder station, capturing images at 4 distinct timepoints daily for two days before and four days after surgical castration. A labeling process was applied to build a pain facial image database and machine learning methods were used to train the computational pain classifier. The machine vision algorithm was developed through the training of a Convolutional Neural Network (CNN) that resulted in an overall accuracy of 75.8% while classifying pain on three levels: not present, moderately present, and obviously present. While classifying between two categories (pain not present and pain present) the overall accuracy reached 88.3%. Although there are some improvements to be made in order to use the system in a daily routine, the model appears promising and capable of measuring pain on images of horses automatically through facial expressions, collected from video images.

Measurement properties of grimace scales for pain assessment in non-human mammals: a systematic review

ABSTRACT

Facial expressions of pain have been identified in several animal species. The aim of this systematic review was to provide evidence on the measurement properties of grimace scales for pain assessment. The protocol was registered (SyRF#21-Nov-2019) and the study is reported according to the PRISMA guidelines. Studies reporting the development, validation, and the assessment of measurement properties of grimace scales were included. Data extraction and assessment were performed by two investigators, following the Consensus-based Standards for the Selection of Health Measurement Instruments (COSMIN) guidelines. Six categories of measurement properties were assessed: internal consistency, reliability, measurement error, criterion and construct validity, and responsiveness. Overall strength of evidence (high, moderate, low) of each instrument was based on methodological quality, number of studies and studies' findings. Twelve scales for nine species were included (mice, rats, rabbits, horses, piglets, sheep/lamb, ferrets, cats and donkeys). Considerable variability regarding their development and measurement properties was observed. The Mouse, Rat, Horse and Feline Grimace Scales exhibited high level of evidence. The Rabbit, Lamb, Piglet and Ferret Grimace Scales and Sheep Pain Facial Expression Scale exhibited moderate level of evidence. The Sheep Grimace Scale, EQUUS-FAP and EQUUS-Donkey-FAP exhibited low level of evidence for measurement properties. Construct validity was the most reported measurement property. Reliability and other forms of validity have been understudied. This systematic review identified gaps in knowledge on the measurement properties of grimace scales. Further studies should focus on improving psychometric testing, instrument refinement and the use of grimace scales for pain assessment in non-human mammals.

Neuropathic pain modeling: Focus on synaptic and ion channel mechanisms

Animal models of pain consist of modeling a pain-like state and measuring the consequent behavior. The first animal models of neuropathic pain (NP) were developed in rodents with a total lesion of the sciatic nerve. Later, other models targeting central or peripheral branches of nerves were developed to identify novel mechanisms that contribute to persistent pain conditions in NP. Objective assessment of pain in these different animal models represents a significant challenge for pre-clinical research. Multiple behavioral approaches are used to investigate and to validate pain phenotypes including withdrawal reflex to evoked stimuli, vocalizations, spontaneous pain, but also emotional and affective behaviors. Furthermore, animal models were very useful in investigating the mechanisms of NP. This review will focus on a detailed description of rodent models of NP and provide an overview of the assessment of the sensory and emotional components of pain. A detailed inventory will be made to examine spinal mechanisms involved in NP-induced hyperexcitability and underlying the current pharmacological approaches used in clinics with the possibility to present new avenues for future treatment. The success of pre-clinical studies in this area of research depends on the choice of the relevant model and the appropriate test based on the objectives of the study.

Towards Machine Recognition of Facial Expressions of Pain in Horses

Automated recognition of human facial expressions of pain and emotions is to a certain degree a solved problem, using approaches based on computer vision and machine learning. However, the application of such methods to horses has proven difficult. Major barriers are the lack of sufficiently large, annotated databases for horses and difficulties in obtaining correct classifications of pain because horses are non-verbal. This review describes our work to overcome these barriers, using two different approaches. One involves the use of a manual, but relatively objective, classification system for facial activity (Facial Action Coding System), where data are analyzed for pain expressions after coding using machine learning principles. We have devised tools that can aid manual labeling by identifying the faces and facial keypoints of horses. This approach provides promising results in the automated recognition of facial action units from images. The second approach, recurrent neural network end-to-end learning, requires less extraction of features and representations from the video but instead depends on large volumes of video data with ground truth. Our preliminary results suggest clearly that dynamics are important for pain recognition and show that combinations of recurrent neural networks can classify experimental pain in a small number of horses better than human raters.