Application and development of a CNN model to optimize an OligoFISSEQ image obtention pipeline

Martí Aranda, Aina

Application and development of a CNN model to optimize an OligoFISSEQ image obtention pipeline

License

Citation

Martí Aranda, A. Application and development of a CNN model to optimize an OligoFISSEQ image obtention pipeline. 2021. handle: http://hdl.handle.net/10230/49015

This citation was generated automatically.

Abstract

In this project we have explored the application of an artificial intelligence deep learning algorithm (DLA) in the OligoFISSEQ image acquisition protocol to classify the obtained images according to their expected quality. We developed and tested several approaches for image preprocessing, model architecture, and data augmentation. This study achieved the best regression models with mean squared errors of 0.0149 for model 4, 0.0131 for model 3, 0.0109 for model 2, and 0.0169 for model 1, which used as binary classifiers yielded accuracies of 0.64, 0.80, 0.88, and 0.88 in the external validation dataset, respectively.