Automated tumour boundary delineation on (18)F-FDG PET images using active contour coupled with shifted-optimal...

Area of Science:

• Medical Imaging
• Image Processing
• Radiology

Background:

• Accurate tumor delineation in Positron Emission Tomography (PET) is crucial for effective cancer treatment planning.
• Traditional segmentation methods, like Otsu's thresholding, can be biased by significant differences in object-background variance.
• Existing active contour models may lack precision in complex imaging scenarios.

Purpose of the Study:

• To develop an automated, accurate, and robust method for tumor boundary tracing in PET images.
• To address the limitations of conventional thresholding techniques in PET image segmentation.
• To improve the precision and accuracy of tumor volume delineation for radiation oncology.

Main Methods:

• An automatic tumor boundary tracing method integrating a novel double-stage threshold search with a region-based active contour model.
• The double-stage threshold search minimizes energy between Otsu's threshold and maximum intensity, overcoming bias in variance.
• The combined algorithm was validated using phantom inserts and clinically applied to oesophageal cancer patients.

Main Results:

• The combined algorithm demonstrated higher accuracy in segmenting tumor volumes compared to traditional active contour methods.
• Clinical implementation in oesophageal cancer patients showed reduced erroneous delineation, improving PET tumor contouring precision.
• The method proved robust, independent of source-to-background ratio (SBR) curves, and did not require prior lesion size estimation.

Conclusions:

• The proposed automatic method offers a significant improvement in PET tumor boundary tracing accuracy and precision.
• This approach provides a robust and reliable tool for radiation oncologists, enhancing treatment planning.
• The algorithm's independence from SBR and lesion size makes it broadly applicable in clinical settings.