The role of artificial intelligence in the pose estimation method for early diagnosis of cerebral palsy: advances in diagnostic medical imaging

• Autores: Maguiña, Alyssa N., Vasquez-Roque, Carlos E.
• Formato: artículo
• Fecha de Publicación: 2023
• Institución: Instituto Nacional de Salud del Niño San Borja
• Repositorio: INSNS - Revistas
• Lenguaje: español
• OAI Identifier: oai:ojs.pkp.sfu.ca:article/17
• Enlace del recurso: https://investigacionpediatrica.insnsb.gob.pe/index.php/iicqp/article/view/17
• Nivel de acceso: acceso abierto
• Materia: Parálisis Cerebral; Niño; Sistemas de Visión Computacional; Inteligencia Artificial; Diagnóstico por Imagen; Cerebral Palsy; Child; Artificial Intelligence; Vision Systems, Computer; Imaging Diagnosis

One of the traditional methods for diagnosing infantile cerebral palsy is general motor assessment, which analyzes the quality and complexity of the infant's movements by visual inspection of spontaneous movements. However, this assessment is subjective and requires highly trained clinicians, which can becostly and time-consuming. To overcome this limitation, computer vision-based solutions are currently being developed to analyze infant movements. These analyses are based on pose estimation, obtained from artificial intelligence models, and then artificial intelligence-based classification algorithms are used to determine whether the movements are normal or abnormal. In this article, we present the use of pose estimation as a computer vision method to analyze fidgety movements in children with cerebral palsy and compare these estimated movements with how artificial intelligence algorithms classify them. Finally, some challenges and future perspectives on this technology are identified.