A Veterinary DICOM-Based Deep Learning Denoising Algorithm Can Improve Subjective and Objective Brain MRI Image...

Area of Science:

• Veterinary Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Magnetic Resonance Imaging (MRI) is crucial for diagnosing brain conditions in veterinary patients.
• Image noise can degrade diagnostic quality in veterinary brain MRI.
• Deep learning (DL) algorithms show potential for improving medical image quality.

Purpose of the Study:

• To evaluate the efficacy of a DICOM-based deep-learning (DL) denoising algorithm for veterinary brain MRI.
• To quantitatively and qualitatively compare native MRI scans with those processed by the DL algorithm.
• To assess the impact of DL denoising on image quality metrics in canine and feline brain MRI.

Main Methods:

• Analytical cross-sectional method comparison study involving 30 dogs and cats.
• Quantitative analysis of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) on T2W, T2-FLAIR, and GRE sequences.
• Qualitative assessment by three blinded veterinary radiologists evaluating coarseness, contrast, and overall image quality.

Main Results:

• Statistically significant increases in SNR for cortical gray matter, subcortical white matter, deep gray matter, and internal capsule.
• Significantly higher CNR between cortical gray and white matter, and deep gray matter and internal capsule.
• Radiologists reported generally better coarseness, contrast, and overall quality scores for denoised images.

Conclusions:

• The DICOM-based DL denoising algorithm effectively reduces noise in 1.5T MRI of canine and feline brains.
• DL denoising demonstrably improves quantitative image quality metrics (SNR and CNR).
• The algorithm leads to improved perceived image quality by veterinary radiologists, aiding diagnosis.