Kidney Segmentation from Dynamic Contrast-Enhanced Magnetic Resonance Imaging Integrating Deep Convolutional Neural...

Area of Science:

• Medical Imaging
• Nephrology
• Artificial Intelligence

Background:

• Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is vital for diagnosing and treating chronic kidney disease.
• Accurate kidney segmentation from DCE-MRI is essential for assessing kidney function.
• Deep convolutional neural networks (CNNs) show promise for medical image segmentation tasks.

Purpose of the Study:

• To propose a fully automated, two-phase approach for kidney segmentation in DCE-MRI scans.
• To integrate CNNs and level set methods for precise kidney delimitation.
• To evaluate the proposed method's effectiveness compared to existing techniques.

Main Methods:

• Development of two U-Net (UNT) based CNNs to generate kidney probability maps from DCE-MRI scans.
• Integration of predicted probability maps with a level set method incorporating shape prior information.
• Utilizing a dataset of 45 subjects for training, validation, and testing the automated segmentation approach.

Main Results:

• The proposed two-phase method achieved high performance in kidney segmentation.
• Key metrics include a Dice similarity coefficient of 0.95 ± 0.02 and intersection over union of 0.91 ± 0.03.
• A 95% Hausdorff distance of 1.54 ± 1.6 was recorded, demonstrating accuracy.

Conclusions:

• The developed two-phase approach effectively segments kidneys in DCE-MRI scans.
• The method demonstrates superior performance compared to standalone UNT models and other level set-based techniques.
• This automated approach holds significant potential for improving chronic kidney disease evaluation.