An efficient computational framework for gastrointestinal disorder prediction using attention-based transfer...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Gastroenterology

Background:

• Diagnosing gastrointestinal (GI) disorders is challenging due to varied presentations and the time-intensive nature of manual review.
• Computer-aided diagnostic (CAD) systems offer automation to improve efficiency and accuracy in GI diagnosis.
• Existing CAD systems require enhancement for safety and reliability on large medical datasets.

Purpose of the Study:

• To develop an effective CAD system for classifying eight types of GI images.
• To improve the safety and reliability of AI-driven diagnostic tools for gastroenterology.

Main Methods:

• Utilized transfer learning with an attention mechanism for image classification.
• Employed the ConvNeXt pre-trained network for feature extraction.
• Evaluated the proposed ConvNeXt+Attention CAD system on a dataset of GI images.

Main Results:

• The ConvNeXt+Attention system demonstrated superior performance compared to other state-of-the-art methods.
• Achieved an area under the receiver operating characteristic curve (AUC) of 0.9997.
• Achieved an area under the precision-recall curve (AUPRC) of 0.9973, indicating excellent diagnostic capability.

Conclusions:

• The proposed ConvNeXt+Attention CAD system is a robust and effective tool for classifying GI images.
• The study highlights the potential of combining transfer learning and attention mechanisms for advanced medical diagnostics.
• The developed system shows promise for enhancing the accuracy and efficiency of diagnosing GI disorders.