Diagnóstico de lesão no ligamento cruzado anterior por vibroartrografia e machine learning

Abstract

Knee injuries have a high incidence across all age groups, and their accurate diagnosis can represent a clinical challenge, especially in cases of partial injuries. Conventional methods such as physical examinations and imaging tests, although widely used, may present limitations related to sensitivity, cost, and the invasive nature of some procedures. This work proposes an alternative approach based on vibroarthrography (VAG) combined with machine learning (ML) techniques to support the diagnosis of knee injuries. A data acquisition system was developed using a MEMS sensor module — composed of an accelerometer and a gyroscope — to capture vibroacoustic signals during the execution of a standardized joint examination procedure. The results demonstrate the potential of the methodology as a non-invasive, low-cost tool to support clinical diagnosis. The study achieved 100% accuracy with certain bidimensional sensor combinations, based on a dataset of 33 samples.