Related Experiment Video
Updated: Sep 9, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Automated Kidney Tumor Segmentation in CT Images Using Deep Learning: A Multi-Stage Approach
Hung-Cheng Kan1, Geng-Ming Fan2, Ming-Hao Wei3
1In-Service Master Program in Artificial Intelligence in Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan (H.-C.K., S.-J.P.); Division of Urology, Department of Surgery, Linkou Chang Gung Memorial Hospital, Taoyuan, Taiwan (H.-C.K., P-H.L., I.-H.S., K.-J.Y., S.-T.P., T.W.); College of Medicine, Chang Gung University, Taoyuan, Taiwan (H.-C.K., P-H.L., I.-H.S., K.-J.Y., S.-T.P., T.W.).
This study introduces an automated DeepMedic 3D convolutional neural network for segmenting kidneys and renal tumors on CT scans. The AI model provides accurate and reproducible results, improving diagnostic efficiency in renal oncology.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Radiology
- Oncology
Background:
- Computed tomography (CT) is crucial for renal tumor assessment, but manual segmentation is time-consuming and variable.
- Accurate tumor delineation is vital for diagnosis, treatment planning, and prognosis.
- Current methods face challenges due to tumor heterogeneity and indistinct margins.
Purpose of the Study:
- To develop and validate a fully automated segmentation pipeline for kidneys and renal tumors using CT images.
- To address the clinical need for reliable, accurate, and reproducible automated segmentation tools.
- To enhance diagnostic workflows in renal oncology.
Main Methods:
- Development of a DeepMedic 3D convolutional neural network for multi-scale feature extraction.
- Training and evaluation on 382 contrast-enhanced CT scans with physician annotations.
- Implementation of image preprocessing (Hounsfield unit conversion, windowing, 3D reconstruction, resampling) and post-processing.
Main Results:
- High performance in kidney segmentation: Dice coefficient 93.82%, precision 94.86%, recall 93.66%.
- Accurate renal tumor segmentation: Dice coefficient 88.19%, precision 90.36%, recall 88.23%.
- Visual validation confirmed clinical relevance and accuracy of automated segmentation.
Conclusions:
- The DeepMedic-based framework offers robust and accurate segmentation of kidneys and renal tumors from CT scans.
- The automated tool has the potential for real-time application, improving diagnostic efficiency.
- This technology can significantly aid in treatment planning and patient management in renal oncology.
