A Study on Staging Cystic Echinococcosis Using Machine Learning Methods.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Medicine
• Parasitic Disease Diagnostics

Background:

• Cystic echinococcosis (CE) is a slow-progressing parasitic disease with non-specific symptoms, often delaying diagnosis and treatment.
• Accurate staging of CE is critical for effective management, as defined by the World Health Organization (WHO).

Purpose of the Study:

• To develop and evaluate an advanced artificial intelligence (AI) and machine learning (ML) system for classifying CE cyst stages.
• To compare the performance of ten ML algorithms across CT, ultrasound (US), and MRI datasets.

Main Methods:

• Utilized computed tomography (CT), ultrasound (US), and magnetic resonance imaging (MRI) datasets for CE cyst classification.
• Evaluated ten ML algorithms, including hybrid models like CNN+ResNet and Inception+ResNet.
• Developed a normalization and scoring technique to consolidate performance metrics (accuracy, precision, recall, specificity, F1 score) for model selection.

Main Results:

• Hybrid models, specifically CNN+ResNet and Inception+ResNet, demonstrated superior performance across all imaging modalities.
• The CNN+ResNet model achieved 97.55% accuracy on CT, 93.99% on US, and 100% on MRI.
• The proposed scoring technique effectively identified the best-performing model for CE cyst classification.

Conclusions:

• AI and ML, particularly hybrid and pre-trained models, show significant potential for advancing medical image classification.
• This research offers a promising approach to enhance the differential diagnosis of cystic echinococcosis.
• The developed system can aid in more precise and timely diagnosis of CE stages.