A novel patches-selection method for the classification of point-of-care biosensing lateral flow assays with cardiac...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Medical Diagnostics

Background:

• Cardiovascular Disease (CVD) is a leading global cause of mortality, necessitating prompt diagnosis and management.
• Point-of-care (POC) biosensing devices, particularly lateral flow assays (LFAs), are valuable for CVD diagnosis but often struggle with low sensitivity for detecting minimal analyte concentrations.
• Advancements in artificial intelligence (AI) and image processing offer potential solutions to enhance detection sensitivity and improve disease diagnosis.

Purpose of the Study:

• To develop and evaluate a novel patches-selection approach for analyzing lateral flow assay (LFA) images.
• To leverage image features from LFA test and control lines for reliable prediction and classification of cardiovascular disease.
• To utilize Convolutional Neural Networks (CNNs) for accurate risk stratification of cardiovascular conditions.

Main Methods:

• A novel patches-selection method was employed to generate relevant image segments from LFA images, focusing on test and control lines.
• Image features were extracted from these generated patches for subsequent analysis.
• Classification algorithms, specifically Convolutional Neural Networks (CNNs), were deployed to predict and classify LFA images based on extracted features.

Main Results:

• The CNN model achieved approximately 98% accuracy in classifying LFA images.
• The proposed patches-selection approach demonstrated high performance in extracting crucial image information for disease classification.
• This methodology represents a novel investigation in the field, with no prior studies reported using this specific approach.

Conclusions:

• The developed AI-driven image analysis method shows significant promise for enhancing the sensitivity and accuracy of LFAs in detecting cardiovascular disease.
• The high classification accuracy suggests the potential applicability of this approach for identifying a broad spectrum of diseases and conditions.
• This technique offers a valuable tool for medical professionals in risk stratification and early diagnosis of cardiovascular problems.