Document Type : Research Note

Author

• Mohammad Heydari

Isfahan

Abstract

Deep learning–based super-resolution has become an important tool for enhancing brain magnetic resonance imaging (MRI), particularly when acquisition constraints limit spatial resolution. Lightweight autoencoder architectures have recently been proposed to achieve computational efficiency while maintaining reconstruction quality. However, certain architectural choices adopted in these models—most notably the incorporation of encoder–decoder skip connections—raise methodological concerns regarding the preservation of the information bottleneck principle that defines autoencoder-based learning. This note critically examines whether such designs genuinely rely on latent representation learning or instead introduce shortcut pathways that weaken the inferential nature of super-resolution reconstruction.

Keywords

• Autoencoder-based super-resolution
• Encoder–decoder architectures
• Latent representation
• Shortcut connections
• Deep learning

Main Subjects

• H.3. Artificial Intelligence