Um modelo de visão computacional para análise biomecânica do valgismo dinâmico

Abstract

Introduction: Injuries in the lower limbs are frequent occurrences in physical activities that demand an impact on the extremities. Most of these injuries occur due to the lack of correct body alignment during the activity performed. An important predictor of global body assessment to prevent these types of injuries is the identification of dynamic valgism, which is the medial displacement of the knee during the impact of the lower limb in activities such as running and jumping. Several methods are used for the detection of dynamic valgism during physical examination, among them the analysis of movement by video. However, this type of analysis has certain limitations related mainly to the use of markers for fixed points or specific suits, which can lead to incorrect data and erroneous diagnostic conclusions. An alternative may be the use of "virtual markers" for analysis developed by software that model the biomechanics of movement. Objectives: This study aims to establish a standardized, effective and accessible method for analyzing and monitoring the evolution of dynamic valgus. Methods: Data will be collected in the form of video, captured by the Kinect v2® tool, and processed using specific software developed by the researchers. This software will be used to perform the test to analyze the dynamic valgus in an automated way. In the study, the variation of results, quantitative, presented between the evaluators and the developed software will be measured. For this, the average of the results of each evaluator of the trunk, hip, knee and ankle movements will be measured; during the two moments of analysis of the dynamic valgus. After that, the inter-rater and intra-rater angles will be compared using the paired t-test. Results: 34 samples were collected. The evaluation carried out between the analyzes of the computational model designed and the specialist in our research did not show a significant agreement index. However, the model has shown effective results for clinical practice. The software presented seven moments with significant agreement, while the evaluation by the specialist, there were 02 moments of significant agreement. Conclusion: In the current study it was identified that the capture and processing of images evaluating the dynamic valgus from the developed model was more effective than the analysis by the specialists. This shows that the use of our model can be used in basic clinical practice.