O26. A review on target volume delineation using FDG-PET/CT images for radiation oncology. (September 2016)

• Record Type: Journal Article
• Authors: du Toit, M.; Potgieter, A.; Heunis, M.; Boellaard, R.; Pruim, J.
• Is Part Of: Physica medica. Volume 32(2016)Supplement 2
• Page Start: 149
• Publication Date: 2016-09
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2016.07.034
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : Introduction: In recent years, Positron Emission Tomography (PET) became available as a complementary imaging modality to CT and MRI for radiotherapy planning. PET provides information on radiotracer uptake heterogeneity within a tumour which in turn can indicate the distribution of the molecular process. Thus the primary tumour can be characterized more accurately to deliver an effective treatment plan and reduce the side effects associated with radiation treatment. It has been shown that PET/CT reduces the intra- and inter-observer variation of the tumour delineation and it increases the turnover and accuracy of the delineation process from using only CT images. However to detect the tumour boundary visually from PET images alone is challenging, therefore numerous automatic delineation algorithms have been developed. The purpose was to assess the efficacy and accuracy of existing delineation methods for FDG-PET/CT images for the oncologist to improve tumour volume delineation for radiotherapy planning. Materials and methods: A systematic literature review was conducted to find tumour delineation techniques applied to FDG-PET scans which are likely to be considered by the oncologist as a possible solution to the delineation problem. Visual variation of PET images is shown in comparison to CT images with NEMA phantom images. Results: Apart from the thresholding techniques, current automated methods prove to be tumour site specific and need additional calibrations and/or measurements on the PET/CT units. A good departmental protocol such as Resear ch forL ife (EARL) should be followed to ensure a uniform platform for reporting and contouring of images especially if visual contouring is applied. Current research is investigating combinations of algorithms and prediction methods based on statistical information and atlas data. Machine learning methodologies are also under discussion since it can handle big data sets and assist with decision making. Conclusion: We believe that until these methods have been clinically tested and verified, the oncologist would be hesitant to apply the suggested changes to the GTV as indicated by the PET.