RATIONALE AND OBJECTIVES: The authors developed a real-time, interactive three-dimensional (3D) segmentation pipeline that uses relatively low-level segmentation operations and provides two-dimensional and 3D visualization through a user-friendly graphical interface. MATERIALS AND METHODS: The low-level segmentation processes were implemented on a massively parallel computer; the graphical user interface was written with a public domain software toolkit. Since their implementation 2 years ago, these segmentation tools have been applied to approximately 300 computed tomographic and magnetic resonance imaging data sets. Two typical clinical cases are presented to demonstrate their applications. RESULTS: The entire processing pipeline can be executed in a few seconds. The tools are simple to learn because they involve the use of low-level procedures and a user-friendly graphical interface with a short interactive response time. Segmentation of the bones, aorta, kidneys, and kidney cysts in case 1 could be performed in about 16 minutes. The time needed to segment each organ in case 2 ranged from about 15 minutes for the skin and brain to about 1 minute for the tumor. CONCLUSION: Satisfactory results can be obtained in a relatively short time with the real-time interactive 3D segmentation system. Operation of the system can be easily learned by a wide variety of nonspecialized users with some medical background.
RATIONALE AND OBJECTIVES: The authors developed a real-time, interactive three-dimensional (3D) segmentation pipeline that uses relatively low-level segmentation operations and provides two-dimensional and 3D visualization through a user-friendly graphical interface. MATERIALS AND METHODS: The low-level segmentation processes were implemented on a massively parallel computer; the graphical user interface was written with a public domain software toolkit. Since their implementation 2 years ago, these segmentation tools have been applied to approximately 300 computed tomographic and magnetic resonance imaging data sets. Two typical clinical cases are presented to demonstrate their applications. RESULTS: The entire processing pipeline can be executed in a few seconds. The tools are simple to learn because they involve the use of low-level procedures and a user-friendly graphical interface with a short interactive response time. Segmentation of the bones, aorta, kidneys, and kidney cysts in case 1 could be performed in about 16 minutes. The time needed to segment each organ in case 2 ranged from about 15 minutes for the skin and brain to about 1 minute for the tumor. CONCLUSION: Satisfactory results can be obtained in a relatively short time with the real-time interactive 3D segmentation system. Operation of the system can be easily learned by a wide variety of nonspecialized users with some medical background.