Automated segmentation of kidney and renal mass and automated detection of renal mass in CT urography using 3D U-Net-based deep convolutional neural network

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Kidney and renal mass segmentation and detection are crucial for diagnosis and treatment planning.
• Manual segmentation is time-consuming and prone to inter-observer variability.
• Automated methods can improve efficiency and accuracy in medical image analysis.

Purpose of the Study:

• To develop and evaluate a 3D U-Net deep learning model for automated segmentation of kidneys and renal masses.
• To assess the model's capability in detecting renal masses in the corticomedullary phase of computed tomography urography (CTU).

Main Methods:

• A 3D U-Net architecture was employed for segmentation tasks.
• The model was trained and evaluated on CTU data from 441 patients (882 kidneys).
• Segmentation accuracy was quantified using the Dice Similarity Coefficient (DSC).
• Detection performance was assessed using the area under the receiver operating characteristic curve.

Main Results:

• The model achieved high segmentation accuracy for kidneys (DSC: 0.973) and renal tumors (DSC: 0.844).
• Moderate accuracy was observed for renal cyst segmentation (DSC: 0.536).
• The model demonstrated good performance in detecting both renal tumors and cysts.

Conclusions:

• The 3D U-Net model shows promise for automated kidney and renal tumor segmentation and detection in CTU.
• This automated tool can potentially reduce radiologist workload and enhance diagnostic accuracy.
• The model can provide valuable quantitative data for clinical decision-making, treatment, and surgical planning.