Fatima Haimour

Brief summary of my PhD work:

Medical image translation is the task of converting images between different modalities, such as Magnetic Resonance (MR) to Computed Tomography (CT) or Positron Emission Tomography (PET) to CT. One approach to solve this problem is to use an attention Generative Adversarial Network (GAN) and transfer learning. Attention GANs are GANs that use an attention mechanism to focus on specific regions of the input image, allowing for more accurate image generation. Transfer learning is a technique that involves using a pre-trained model as a starting point to train a new model, rather than training the new model from scratch.

By using an attention GAN and transfer learning, it is possible to train a model to convert images between different modalities. The attention mechanism can be trained to focus on specific features of the input image, such as organs or structures, to generate more accurate images in the target modality. Additionally, transfer learning allows the model to take advantage of the knowledge learned by the pre-trained model, reducing the amount of data and computational resources needed to train the new model. This approach has been shown to produce more realistic and accurate images compared to other methods, making it a promising solution for medical image translation tasks.

Image-to-image translation

Figure 1: The training process of the proposed paired–unpaired AGGAN with transfer learning.