Twelve automated thresholding methods for segmentation of PET images: a phantom study

Area of Science:

• Medical Imaging
• Image Processing
• Radiology

Background:

• Accurate tumor volume delineation in positron emission tomography (PET) is crucial for diagnosis and therapy planning.
• Current manual and semi-automated segmentation methods are time-consuming and operator-dependent.
• Fully automated segmentation often involves complex mathematical development or specialized calibration.

Purpose of the Study:

• To develop and evaluate a fully automated method for segmenting PET images.
• To implement 12 automated thresholding algorithms for PET image segmentation.
• To compare the performance of automated thresholding algorithms against standard PET segmentation techniques.

Main Methods:

• Implemented 12 automated thresholding algorithms, originally from optical character recognition and image analysis fields.
• Applied algorithms to segmented spherical (18)F-filled objects acquired on clinical PET/CT and small animal PET scanners.
• Compared automated segmentation results with a standard 42% maximum uptake threshold reference.

Main Results:

• Automated thresholding algorithms selected image-specific thresholds without prior spatial information or tomograph calibration.
• Ridler and Ramesh algorithms, utilizing clustering and histogram-shape analysis, demonstrated superior performance.
• These algorithms significantly outperformed the classical 42%-based threshold (p < 0.05).

Conclusions:

• Fully automated thresholding algorithms can effectively segment PET images.
• These methods provide a more robust and accurate alternative to classical PET segmentation tools.
• The study demonstrates the potential of automated thresholding for improved PET image analysis in clinical settings.