Advancements in Deep Learning for Disease Detection: a Comprehensive Survey on T-Fusion Net and Spatial Attention in Infectious Disease Imaging

EasyChair Preprint 11713

Abstract

In this analytical survey, we showcase one of our research in the area of virological science and its critical impact on infectious disease detection and diagnosis. The T-Fusion Net, an innovative deep neural network meticulously tailored for the precise detection of COVID-19, leveraging SARS-CoV-2 CT scans. Distinguished by its focus on virology, our model integrates Multiple Localizations-Based Spatial Attention Mechanisms (MLSAM) to enrich feature extraction and representation in medical image analysis, specifically emphasizing the nuanced patterns associated with viral infections. Through the strategic assembly of an ensemble of T-Fusion Nets orchestrated via fuzzy max fusion, we achieve unparalleled classification accuracy, boasting rates of 97.59% for T-Fusion Net and an impressive 98.4% for its ensemble counterpart. Our findings not only underscore the robustness of MLSAM in selective feature extraction but also spotlight its profound impact on enhancing diagnostic capabilities in infectious diseases. This study, with a virological lens, unveils a technological advancement poised to contribute in medical imaging in the realm of virology, offering a valuable tool for frontline healthcare practitioners in their battle against COVID-19.

Keyphrases: COVID-19 Diagnosis, Deep Neural Network, Spatial attention mechanisms, Virology, ensemble learning, infectious disease detection

@booklet{EasyChair:11713,
  author    = {Susmita Ghosh and Abhiroop Chatterjee},
  title     = {Advancements in Deep Learning for Disease Detection: a Comprehensive Survey on T-Fusion Net and Spatial Attention in Infectious Disease Imaging},
  howpublished = {EasyChair Preprint 11713},
  year      = {EasyChair, 2024}}