ReviewSegmentation of positron emission tomography images: Some recommendations for target delineation in radiation oncology
Section snippets
From delineation to segmentation
The delineation of target volumes on PET images can obviously be done by hand with an appropriate computer interface. The major drawback of this approach is the variability of the resulting contours [15]. Because of their modest resolution [1], PET images look blurred and the human eye cannot easily distinguish the boundaries of the target. An additional difficulty is related to display settings (windowing level and width). Changing the colour scale or saturation can dramatically change the
Review of automatic delineation methodologies
Image segmentation can be achieved in various ways. The simplest method consists in considering each pixel independently and to determine its class label by looking solely at its value. This turns out to be equivalent to building the image histogram and to split it into several parts thanks to one or several thresholds. In a binary problem (e.g., with two classes: target and non-target), one threshold is sufficient.
Thresholding can be refined in several ways. The simplest and most common
Thresholding in question
As a matter of fact, thresholding is an old [41] but still very popular automatic method of segmenting PET images (see e.g., [15], [17], [18], [42], [43], [44], [45], [46], [47] and references therein). As the goal is to separate a region with high uptake from a background with lower uptake, the idea of an intensity threshold naturally emerges. It is both intuitive to understand and easy to implement: all pixels having an intensity that is lower than the threshold are labeled as non-target
Validation issues
The purpose of validation is basically to check that the considered delineation method is applicable to a broad range of cases with a reasonable accuracy. Quality and pertinence of validation thus depends on (i) the set of images it involves and (ii) the quality criteria it uses to assess the discrepancy between the obtained result and the desired one.
Several types of images can be used in validation. Computer-generated images are typically useful in the primary steps of validation, as a proof
Summary and conclusions
Accuracy of automatic target delineation is directly conditioned by image quality [61], [72]. Therefore, image acquisition and reconstruction are as important as the delineation technique itself. Increasing the acquisition duration or the tracer dose can contribute to improving the signal-to-noise ratio of the image (the larger amount of collected data reduces the statistical uncertainty in the reconstructed image). These requirements must naturally be put in the balance with patient comfort
Financial support
J.A.L. is a Research Associate with the Belgian fund of scientific research (Fonds National de la recherché scientifique, FRS-FNRS).
The authors have no financial relationship with the organizations that sponsored the research.
The authors have had full control of all primary data and agree to allow the journal to review their data if requested.
Acknowledgements
The author wishes to thank the reviewer for his/her useful comments and Anne Bol for her careful proofreading.
References (76)
- et al.
FDG-PET/CT imaging for staging and target volume delineation in preoperative conformal radiotherapy of rectal cancer
Int J Radiat Oncol Biol Phys
(2008) - et al.
Use of PET and PET/CT for radiation therapy planning: IAEA expert report 2006–2007
Radiother Oncol
(2009) - et al.
The integration of PET–CT scans from different hospitals into radiotherapy treatment planning
Radiother Oncol
(2008) - et al.
Adaptive biological image-guided IMRT with anatomic and functional imaging in pharyngolaryngeal tumors: impact on target volume delineation and dose distribution using helical tomotherapy
Radiother Oncol
(2007) - et al.
Using fluorodeoxyglucose positron emission tomography to assess tumor volume during radiotherapy for non-small-cell lung cancer and its potential impact on adaptive dose escalation and normal tissue sparing
Int J Radiat Oncol Biol Phys
(2009) - et al.
Biological image-guided radiotherapy in rectal cancer: challenges and pitfalls
Int J Radiat Oncol Biol Phys
(2009) - et al.
Comparison of five segmentation tools for 18F-fluorodeoxy-glucose-positron emission tomography-based target volume definition in head and neck cancer
Int J Radiat Oncol Biol Phys
(2007) - et al.
Variability of gross tumor volume delineation in head-and-neck cancer using CT and PET/CT fusion
Int J Radiat Oncol Biol Phys
(2006) - et al.
Combined 18F-FDG-PET/CT imaging in radiotherapy target delineation for head-and-neck cancer
Int J Radiat Oncol Biol Phys
(2009) - et al.
Comparison of CT- and FDG-PET-defined gross tumor volume in intensity modulated radiotherapy for head-and-neck cancer
Int J Radiat Oncol Biol Phys
(2005)