Researchers from the ibs.GRANADA design an AI to predict cell migration in breast cancer
The study has been published in the journal 'Computers in Biology and Medicine', number two in the world in its field.
A team of researchers from the Granada Biosanitary Research Institute (ibs.GRANADA), the University of Granada and the University of Seville, led by Juan Antonio Marchal Corrales and Miguel Ángel Gutiérrez Naranjo respectively, has published an innovative study in which an Artificial Intelligence is designed to improve the prediction of the evolution of cell migration in breast cancer. The study, entitled Using Deep Learning for Predicting the Dynamic Evolution of Breast Cancer Migration (Using Deep Learning to Predict the Dynamic Evolution of Breast Cancer Migration, in Spanish), represents an important advance in the combination of techniques deep learning (deep learning) and computational biology.
The multidisciplinary work, with the participation of Francisco M. García Moreno and the doctoral student Jesús Ruiz Espigares, both from the University of Granada, focuses on the development of a predictive framework called Prediction Wound Progression Framework (PWPF or Wound Progression Prediction Framework in Spanish). This framework harnesses the power of deep learning to analyze and predict cell migration in two-dimensional models—technically known as Wound Healing or wound healing—providing new insights into the understanding of the metastatic process of breast cancer.
“Metastasis is the main cause of mortality in patients with breast cancer and understanding how cell migration occurs is crucial to developing better therapeutic strategies,” explains Jesús Ruiz, co-principal investigator of the Department of Human Anatomy and Embryology at the University of Granada and member of the Biomedical Research Centre (CIBM).
The team has developed a neural network architecture based on Conv-LSTM, which takes advantage of both the spatial and temporal characteristics of cell migration data. This architecture allows to accurately predict the evolution of the technique Wound Healing over time, improving the ability to analyze dynamics in the context of breast cancer models. This automated approach can be applied to more complex 3D models that better mimic tumor characteristics and promises to open new avenues for cancer research and treatment.
The research is the result of a multidisciplinary collaboration between different departments and centres: the Department of Computer Languages and Systems (LSI), the Department of Human Anatomy and Embryology and the CITIC of the University of Granada, the Singular Laboratory BioFabi3D_Biofabrication and 3D (bio)printing of the CIBM, the Unit of Excellence "Modeling Nature" and the Biosanitary Research Institute ibs.GRANADA, in addition to the Department of Computer Science and Artificial Intelligence of the University of Seville.
The team's progress is notable not only for its scientific contribution, but also for its accessibility and promotion of open access, as the code and data generated are publicly available in its GitHub and Zenodo repositories, promoting open access and international collaboration in cancer research.
The project has been carried out thanks to funding from the Ministry of Science, Innovation and Universities (MICIN), the Department of Health and Families of the Andalusian Government and the Doctores Galera y Requena Chair of Research in Cancer Stem Cells of the UGR
Bibliographic reference
Garcia-Moreno FM, Ruiz-Espigares J, Gutiérrez-Naranjo MA, Marchal JA. Using deep learning for predicting the dynamic evolution of breast cancer migration. Comput Biol Med. 2024 Sep;180:108890.
https://doi.org/10.1016/j.compbiomed.2024.108890
