Um Método Baseado em Inteligência Artificial para Estimar o Peso e a Altura a partir de Imagens Clínicas de Múltiplas Visões
Resumo
Este estudo propõe um método baseado em aprendizado profundo e visão computacional para automatizar a medição de altura e peso, reduzindo erros inerentes aos humanos em avaliações clínicas. Medidas imprecisas podem causar erros na dosagem e diagnósticos, afetando a saúde do paciente. Foram utilizadas 19.172 imagens clínicas de múltiplas visões (posterior, anterior e laterais) de 4.793 avaliações idividuais. O modelo DeepLabV3 extraiu silhuetas, posteriormente usadas para treinar as arquiteturas WideResNet e Scale Equivariant WideResNet. A abordagem combina silhuetas de múltiplas visões, arquiteturas especializadas para invariância de escala e regularizadores que consideram a relação entre peso e altura. O método obteve um Erro Absoluto Médio de 3,7 kg e de 3,6 cm para peso e altura respectivamente, utilizando 3.834 imagens ou 958 avaliações para os dados de teste. Os melhores resultados foram obtidos onde o Erro Quadrático Médio (EQM) foi o menor. Os resultados demonstram a eficácia da abordagem na obtenção de medidas biométricas precisas, ressaltando seu potencial para aprimorar a precisão clínica e auxiliar no monitoramento da saúde de pacientes.Referências
Altinigne, C. Y., Thanou, D., and Achanta, R. (2020). Height and weight estimation from unconstrained images. In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 2298–2302. IEEE.
Bigalke, A., Hansen, L., Diesel, J., and Heinrich, M. P. (2021). Seeing under the cover with a 3d u-net: point cloud-based weight estimation of covered patients. International Journal of Computer Assisted Radiology and Surgery, 16(12):2079–2087.
Cruz-Cunha, M. M. (2016). Encyclopedia of e-health and telemedicine. IGI global.
Deleu, E., Elez, S., Gadodia, A., Macvaugh, K., and Zhao, G. (2021). Using deep learning for urban pedestrian counting. In 2021 IEEE MIT Undergraduate Research Technology Conference (URTC), pages 1–5. IEEE.
Guerra, R. S., Sousa-Santos, A. R., Sousa, A. S., Valdiviesso, R., Afonso, C., Moreira, P., Padrão, P., Borges, N., Santos, A., Ferro, G., et al. (2021). Prediction equations for estimating body weight in older adults. Journal of Human Nutrition and Dietetics, 34(5):841–848.
Han, D., Zhang, J., and Shan, S. (2020). Leveraging auxiliary tasks for height and weight estimation by multi task learning. In 2020 IEEE International Joint Conference on Biometrics (IJCB), pages 1–7. IEEE.
Jin, Z., Huang, J., Wang, W., Xiong, A., and Tan, X. (2022). Estimating human weight from a single image. IEEE Transactions on Multimedia.
Kumar, A., Deeksha, K., Pooja, G. S., Reddy, T. T., and Reddy, T. A. (2022). Estimate height weight and body mass index from face image using machine learning. In 2022 5th International Conference on Multimedia, Signal Processing and Communication Technologies (IMPACT), pages 1–5. IEEE.
Lima, L. D. B., Teixeira, S., Bordalo, V., Lacoste, S., Guimond, S., Sousa, D. L., Pinheiro, D. N., Moreira, R., and Teles, A. S. (2024). A scale-equivariant cnn-based method for estimating human weight and height from multi-view clinic silhouette images. Expert Systems with Applications, 256:124879.
Maskin, L., Attie, S., Setten, M., Rodriguez, P., Bonelli, I., Stryjewski, M., and Valentini, R. (2010). Accuracy of weight and height estimation in an intensive care unit. Anaesthesia and intensive care, 38(5):930–934.
McDonald, C. M., Olofin, I., Flaxman, S., Fawzi, W. W., Spiegelman, D., Caulfield, L. E., Black, R. E., Ezzati, M., Danaei, G., and Study, N. I. M. (2013). The effect of multiple anthropometric deficits on child mortality: meta-analysis of individual data in 10 prospective studies from developing countries. The American journal of clinical nutrition, 97(4):896–901.
Medhi, I., Jain, M., Tewari, A., Bhavsar, M., Matheke-Fischer, M., and Cutrell, E. (2012). Combating rural child malnutrition through inexpensive mobile phones. In Proceedings of the 7th Nordic conference on human-computer interaction: making sense through design, pages 635–644.
Nyholm, B., Obling, L., Hassager, C., Grand, J., Møller, J., Othman, M., Kondziella, D., and Kjaergaard, J. (2022). Superior reproducibility and repeatability in automated quantitative pupillometry compared to standard manual assessment, and quantitative pupillary response parameters present high reliability in critically ill cardiac patients. PLoS One, 17(7):e0272303.
Schulte, J., Kocherovsky, M., Paul, N., Pleune, M., and Chung, C.-J. (2022). Autonomous human-vehicle leader-follower control using deep-learning-driven gesture recognition. Vehicles, 4(1):243–258.
Sinaga, H. T., Siregar, M., and Sitanggang, B. (2024). Anthropometric training impact on chws’ skills in child height measurement in indonesia. Journal of Hunan University Natural Sciences, 51(3).
Wells, M., Goldstein, L. N., and Cattermole, G. (2022). Development and validation of a length-and habitus-based method of total body weight estimation in adults. The American Journal of Emergency Medicine, 53:44–53.
Xia, L., Yang, J., Han, T., Xu, H., Yang, Q., Zhao, Y., and Wang, Y. (2019). A mobilized automatic human body measure system using neural network. Multimedia Tools and Applications, 78:11291–11311.
Zagoruyko, S. and Komodakis, N. (2016). Wide residual networks. arXiv preprint arXiv:1605.07146.
Bigalke, A., Hansen, L., Diesel, J., and Heinrich, M. P. (2021). Seeing under the cover with a 3d u-net: point cloud-based weight estimation of covered patients. International Journal of Computer Assisted Radiology and Surgery, 16(12):2079–2087.
Cruz-Cunha, M. M. (2016). Encyclopedia of e-health and telemedicine. IGI global.
Deleu, E., Elez, S., Gadodia, A., Macvaugh, K., and Zhao, G. (2021). Using deep learning for urban pedestrian counting. In 2021 IEEE MIT Undergraduate Research Technology Conference (URTC), pages 1–5. IEEE.
Guerra, R. S., Sousa-Santos, A. R., Sousa, A. S., Valdiviesso, R., Afonso, C., Moreira, P., Padrão, P., Borges, N., Santos, A., Ferro, G., et al. (2021). Prediction equations for estimating body weight in older adults. Journal of Human Nutrition and Dietetics, 34(5):841–848.
Han, D., Zhang, J., and Shan, S. (2020). Leveraging auxiliary tasks for height and weight estimation by multi task learning. In 2020 IEEE International Joint Conference on Biometrics (IJCB), pages 1–7. IEEE.
Jin, Z., Huang, J., Wang, W., Xiong, A., and Tan, X. (2022). Estimating human weight from a single image. IEEE Transactions on Multimedia.
Kumar, A., Deeksha, K., Pooja, G. S., Reddy, T. T., and Reddy, T. A. (2022). Estimate height weight and body mass index from face image using machine learning. In 2022 5th International Conference on Multimedia, Signal Processing and Communication Technologies (IMPACT), pages 1–5. IEEE.
Lima, L. D. B., Teixeira, S., Bordalo, V., Lacoste, S., Guimond, S., Sousa, D. L., Pinheiro, D. N., Moreira, R., and Teles, A. S. (2024). A scale-equivariant cnn-based method for estimating human weight and height from multi-view clinic silhouette images. Expert Systems with Applications, 256:124879.
Maskin, L., Attie, S., Setten, M., Rodriguez, P., Bonelli, I., Stryjewski, M., and Valentini, R. (2010). Accuracy of weight and height estimation in an intensive care unit. Anaesthesia and intensive care, 38(5):930–934.
McDonald, C. M., Olofin, I., Flaxman, S., Fawzi, W. W., Spiegelman, D., Caulfield, L. E., Black, R. E., Ezzati, M., Danaei, G., and Study, N. I. M. (2013). The effect of multiple anthropometric deficits on child mortality: meta-analysis of individual data in 10 prospective studies from developing countries. The American journal of clinical nutrition, 97(4):896–901.
Medhi, I., Jain, M., Tewari, A., Bhavsar, M., Matheke-Fischer, M., and Cutrell, E. (2012). Combating rural child malnutrition through inexpensive mobile phones. In Proceedings of the 7th Nordic conference on human-computer interaction: making sense through design, pages 635–644.
Nyholm, B., Obling, L., Hassager, C., Grand, J., Møller, J., Othman, M., Kondziella, D., and Kjaergaard, J. (2022). Superior reproducibility and repeatability in automated quantitative pupillometry compared to standard manual assessment, and quantitative pupillary response parameters present high reliability in critically ill cardiac patients. PLoS One, 17(7):e0272303.
Schulte, J., Kocherovsky, M., Paul, N., Pleune, M., and Chung, C.-J. (2022). Autonomous human-vehicle leader-follower control using deep-learning-driven gesture recognition. Vehicles, 4(1):243–258.
Sinaga, H. T., Siregar, M., and Sitanggang, B. (2024). Anthropometric training impact on chws’ skills in child height measurement in indonesia. Journal of Hunan University Natural Sciences, 51(3).
Wells, M., Goldstein, L. N., and Cattermole, G. (2022). Development and validation of a length-and habitus-based method of total body weight estimation in adults. The American Journal of Emergency Medicine, 53:44–53.
Xia, L., Yang, J., Han, T., Xu, H., Yang, Q., Zhao, Y., and Wang, Y. (2019). A mobilized automatic human body measure system using neural network. Multimedia Tools and Applications, 78:11291–11311.
Zagoruyko, S. and Komodakis, N. (2016). Wide residual networks. arXiv preprint arXiv:1605.07146.
Publicado
09/06/2025
Como Citar
L., Lucas D. Batista; TELES, Ariel Soares.
Um Método Baseado em Inteligência Artificial para Estimar o Peso e a Altura a partir de Imagens Clínicas de Múltiplas Visões. In: PRÊMIO ARTUR ZIVIANI - CONCURSO DE TESES E DISSERTAÇÕES (MESTRADO) - SIMPÓSIO BRASILEIRO DE COMPUTAÇÃO APLICADA À SAÚDE (SBCAS), 25. , 2025, Porto Alegre/RS.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 133-138.
ISSN 2763-8987.
DOI: https://doi.org/10.5753/sbcas_estendido.2025.7311.
