Segmentation of heterogeneous or small FDG PET positive tissue based on a 3D-locally adaptive random walk algorithm

• Published in: Computerized medical imaging and graphics Vol. 38; no. 8; pp. 753 - 763
• Main Authors: Onoma, D.P, Ruan, S, Thureau, S, Nkhali, L, Modzelewski, R, Monnehan, G.A, Vera, P, Gardin, I
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Ltd 01.12.2014; Elsevier
• Subjects: Adaptive algorithms; Algorithms; Applied sciences; Artificial intelligence; Biological and medical sciences; Computer Science; Computer science; control theory; systems; Data Interpretation, Statistical; Density; Exact sciences and technology; Fluorodeoxyglucose F18; Heterogeneous tumors; Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Imaging, Three-Dimensional - methods; Internal Medicine; Investigative techniques, diagnostic techniques (general aspects); Medical sciences; Neoplasms - diagnostic imaging; Other; Pattern recognition. Digital image processing. Computational geometry; PET imaging; Phantoms, Imaging; Positron-Emission Tomography - instrumentation; Positron-Emission Tomography - methods; Radionuclide investigations; Radiopharmaceuticals; Random walk; Reproducibility of Results; Segmentation; Sensitivity and Specificity; Signal and Image Processing; Tomography; Tumor segmentation; Tumors; Uptakes; Heterogeneous tumors; PET imaging; Tumor segmentation; Random walk; Euclidean geometry; Adaptive algorithm; Threshold detection; Randomized algorithm; Modeling; Sphere; Probability density; Biomedical data processing; Tissue; Image segmentation; Extreme value; Voxel; Medical imagery; Tumor; Tridimensional image; Lesion; Positron emission tomography; Threshold
• ISSN: 0895-6111; 1879-0771
• DOI: 10.1016/j.compmedimag.2014.09.007

Highlights • We propose to use the distance between adjacent nodes in place of a constant β parameter to take into account the different distances between adjacent nodes in 26 connectivity. • To reduce the partial volume effect in PET imaging, we propose to strengthen the grouping of voxels having similar intensity by adding the likelihood of probability to each class (tumor and non-tumor). • The accuracy in the small and heteregenous tumor segmentation is improved using our improvements.