Comparison of three image segmentation techniques for target volume delineation in positron emission tomography

Abstract

Incorporation of positron emission tomography (PET) data into radiotherapy planning is currently under investigation for numerous sites including lung, brain, head and neck, breast, and prostate. Accurate tumor‐volume quantification is essential to the proper utilization of the unique information provided by PET. Unfortunately, target delineation within PET currently remains a largely unaddressed problem. We therefore examined the ability of three segmentation methods—thresholding, Sobel edge detection, and the watershed approach—to yield accurate delineation of PET target cross‐sections. A phantom study employing well‐defined cylindrical and spherical volumes and activity distributions provided an opportunity to assess the relative efficacy with which the three approaches could yield accurate target delineation in PET. Results revealed that threshold segmentation can accurately delineate target cross‐sections, but that the Sobel and watershed techniques both consistently fail to correctly identify the size of experimental volumes. The usefulness of threshold‐based segmentation is limited, however, by the dependence of the correct threshold (that which returns the correct area at each image slice) on target size.PACS numbers: 87.58.Fg, 87.57.Nk

Comparison of Edge Detection Algorithms for Automated Radiographic Measurement of the Carrying Angle

“…Edge detection is considered an important and primary step in many analyses and for that reason, it is under continuous research. [3] Its operators and algorithms have been extensively employed in the digital analysis of images from various kinds of medical imaging techniques such as: radiography [4][5][6][7][8], mammography [9], ultrasonography [10,11], echocardiography [12], computed tomography [13], magnetic resonance [14][15][16][17], radioisotope scanning [18], positron emission tomography [19], optical coherence tomography [20,21], near-infrared [22], fundography [23][24][25][26][27], angiography [28], microscopy [29], confocal microscopy [30][31][32] and prosthetic vision [33].…”

“…Nevertheless, the literature lacks a large scale study that specifically tests edge detection techniques, both quantitatively and qualitatively, for use in analyzing medical images. Additionally, the vast majority of studies carried out on edge detection employed high-level mathematical packages, such as the MATLAB environment [4,5,7,10,15,19,[25][26][27]30], to perform their analyses; however, although these programs are powerful, they are not readily available for wide use and are not always needed for simpler functions.…”

Segmentation of Lung Tumours in Positron Emission Tomography Scans: A Machine Learning Approach

“…The first broad group aims to segment the tumour by searching for some inhomogeneity throughout the PET scan. Although there are some interesting examples from this group, such as gradient-based (watershed) methods [4,5] and a multimodal generalisation of level set method [6], they are not as well established nor as frequently cited in current reviews as the methods from the second group, which aim to define the optimal threshold value of the uptake in order to segment a tumour. This second group includes approaches that define the optimal threshold as some fixed uptake value, or a fixed percentage of the maximum uptake value; other more sophisticated approaches determine the optimal threshold as the weighted sum of mean target uptake and mean background uptake, among other tecniques [7][8][9][10][11][12].…”

Implementation of a Fused Approach for Segmentation of Brain MR Images for Tumor Extraction

“…Both MRI and CT scan use computers to create detailed images of the brain. The other methods include PET (Positron Emission Tomography) scan, DTI ( Diffusion Tensor Imaging ), MRS (Magnetic Resonance Spectroscopy) scan and biopsy (tissue sample analysis) [6][7].…”