Intelligent Algorithm-Based Multislice Spiral Computed Tomography to Diagnose Coronary Heart Disease.

Area of Science:

• Medical Imaging
• Radiology
• Computational Imaging

Background:

• Coronary artery disease (CAD) diagnosis relies on imaging techniques.
• 64-slice spiral CT is a common diagnostic tool.
• Image noise can affect diagnostic accuracy.

Purpose of the Study:

• To evaluate the clinical value of dictionary learning and expectation maximization reconstruction (DLEM) for denoising 64-slice spiral CT images.
• To assess the diagnostic performance of DLEM-enhanced CT in identifying coronary artery stenosis.

Main Methods:

• A retrospective study involving 120 patients with confirmed coronary heart disease.
• Patients were divided into a control group (conventional CT) and an observation group (DLEM-enhanced CT).
• Coronary angiography (CAG) served as the gold standard for comparison.

Main Results:

• DLEM algorithm significantly improved image quality, with 70% of images achieving grade 4 or above post-processing.
• The observation group showed superior diagnostic performance: consistency (88.33%), sensitivity (86.67%), specificity (80%), and accuracy (85%).
• DLEM achieved comparable denoising to MLEM with fewer iterations.

Conclusions:

• The DLEM algorithm offers effective denoising for 64-slice spiral CT images.
• DLEM significantly enhances the accuracy of diagnosing coronary artery stenosis.
• The DLEM algorithm demonstrates valuable clinical utility for CAD diagnosis and warrants promotion.