COVID-19 Detection from Chest X-rays Using Trained Output Based Transfer Learning Approach.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
• Chest X-rays are crucial for early detection and management of COVID-19.
• Existing detection methods require improvement in efficiency and efficacy.

Purpose of the Study:

• To propose a novel Trained Output-based Transfer Learning (TOTL) approach for COVID-19 detection using chest X-rays.
• To enhance the accuracy and efficiency of COVID-19 diagnosis through an advanced deep learning model.
• To aggregate the capabilities of multiple pre-trained transfer learning models for improved performance.

Main Methods:

• Preprocessing chest X-rays using denoising, contrasting, and segmentation.
• Fine-tuning various pre-trained transfer learning models (InceptionV3, ResNet50, VGG19, etc.) on processed X-rays.
• Further training model outputs with a deep neural network architecture for enhanced performance.

Main Results:

• The TOTL model demonstrated high efficiency and efficacy in COVID-19 detection across four datasets.
• Performance was evaluated using standard metrics and compared favorably against contemporary methods.
• The proposed approach successfully aggregated the strengths of multiple deep transfer learning models.

Conclusions:

• The novel TOTL approach offers a promising solution for accurate and efficient COVID-19 detection from chest X-rays.
• This method can aid in early diagnosis, thereby helping to curtail the spread of the disease.
• The findings highlight the potential of combining transfer learning models for superior diagnostic performance.