Alexander Saunders1,2,3, Kevin S King4, Stefan Blüml1,2, John C Wood5, Matthew Borzage2,6. 1. Children's Hospital Los Angeles, Department of Radiology, Los Angeles, California, United States. 2. Rudi Schulte Research Institute, Santa Barbara, California, United States. 3. University of Southern California, Viterbi School of Engineering, Los Angeles, California, United States. 4. Huntington Medical Research Institutes, Advanced Imaging and Spectroscopy Center, Pasadena, California, United States. 5. Children's Hospital Los Angeles, Division of Cardiology, Los Angeles, California, United States. 6. University of Southern California, Children's Hospital Los Angeles, Fetal and Neonatal Institute, Division of Neonatology, Department of Pediatrics, Los Angeles, California, United States.
Abstract
Purpose: To evaluate six cerebral arterial segmentation algorithms in a set of patients with a wide range of hemodynamic characteristics to determine real-world performance. Approach: Time-of-flight magnetic resonance angiograms were acquired from 33 subjects: normal controls ( N = 11 ), sickle cell disease ( N = 11 ), and non-sickle anemia ( N = 11 ) using a 3 Tesla Philips Achieva scanner. Six segmentation algorithms were tested: (1) Otsu's method, (2) K-means, (3) region growing, (4) active contours, (5) minimum cost path, and (6) U-net machine learning. Segmentation algorithms were tested with two region-selection methods: global, which selects the entire volume; and local, which iteratively tracks the arteries. Five slices were manually segmented from each patient by two readers. Agreement between manual and automatic segmentation was measured using Matthew's correlation coefficient (MCC). Results: Median algorithm segmentation times ranged from 0.1 to 172.9 s for a single angiogram versus 10 h for manual segmentation. Algorithms had inferior performance to inter-observer vessel-based ( p < 0.0001 , MCC = 0.65 ) and voxel-based ( p < 0.0001 , MCC = 0.73 ) measurements. There were significant differences between algorithms ( p < 0.0001 ) and between patients ( p < 0.0042 ). Post-hoc analyses indicated (1) local minimum cost path performed best with vessel-based ( p = 0.0261 , MCC = 0.50 ) and voxel-based ( p = 0.0131 , MCC = 0.66 ) analyses; and (2) higher vessel-based performance in non-sickle anemia ( p = 0.0002 ) and lower voxel-based performance in sickle cell ( p = 0.0422 ) compared with normal controls. All reported MCCs are medians. Conclusions: The best-performing algorithm (local minimum cost path, voxel-based) had 9.59% worse performance than inter-observer agreement but was 3 orders of magnitude faster. Automatic segmentation was non-inferior in patients with sickle cell disease and superior in non-sickle anemia.
Purpose: To evaluate six cerebral arterial segmentation algorithms in a set of patients with a wide range of hemodynamic characteristics to determine real-world performance. Approach: Time-of-flight magnetic resonance angiograms were acquired from 33 subjects: normal controls ( N = 11 ), sickle cell disease ( N = 11 ), and non-sickle anemia ( N = 11 ) using a 3 Tesla Philips Achieva scanner. Six segmentation algorithms were tested: (1) Otsu's method, (2) K-means, (3) region growing, (4) active contours, (5) minimum cost path, and (6) U-net machine learning. Segmentation algorithms were tested with two region-selection methods: global, which selects the entire volume; and local, which iteratively tracks the arteries. Five slices were manually segmented from each patient by two readers. Agreement between manual and automatic segmentation was measured using Matthew's correlation coefficient (MCC). Results: Median algorithm segmentation times ranged from 0.1 to 172.9 s for a single angiogram versus 10 h for manual segmentation. Algorithms had inferior performance to inter-observer vessel-based ( p < 0.0001 , MCC = 0.65 ) and voxel-based ( p < 0.0001 , MCC = 0.73 ) measurements. There were significant differences between algorithms ( p < 0.0001 ) and between patients ( p < 0.0042 ). Post-hoc analyses indicated (1) local minimum cost path performed best with vessel-based ( p = 0.0261 , MCC = 0.50 ) and voxel-based ( p = 0.0131 , MCC = 0.66 ) analyses; and (2) higher vessel-based performance in non-sickle anemia ( p = 0.0002 ) and lower voxel-based performance in sickle cell ( p = 0.0422 ) compared with normal controls. All reported MCCs are medians. Conclusions: The best-performing algorithm (local minimum cost path, voxel-based) had 9.59% worse performance than inter-observer agreement but was 3 orders of magnitude faster. Automatic segmentation was non-inferior in patients with sickle cell disease and superior in non-sickle anemia.
Authors: H A Kirişli; M Schaap; C T Metz; A S Dharampal; W B Meijboom; S L Papadopoulou; A Dedic; K Nieman; M A de Graaf; M F L Meijs; M J Cramer; A Broersen; S Cetin; A Eslami; L Flórez-Valencia; K L Lor; B Matuszewski; I Melki; B Mohr; I Oksüz; R Shahzad; C Wang; P H Kitslaar; G Unal; A Katouzian; M Örkisz; C M Chen; F Precioso; L Najman; S Masood; D Ünay; L van Vliet; R Moreno; R Goldenberg; E Vuçini; G P Krestin; W J Niessen; T van Walsum Journal: Med Image Anal Date: 2013-06-04 Impact factor: 8.545
Authors: Hrvoje Bogunović; José María Pozo; María Cruz Villa-Uriol; Charles B L M Majoie; Rene van den Berg; Hugo A F Gratama van Andel; Juan M Macho; Jordi Blasco; Luis San Román; Alejandro F Frangi Journal: Med Phys Date: 2011-01 Impact factor: 4.071
Authors: Matthew T Borzage; Adam M Bush; Soyoung Choi; Aart J Nederveen; Lena Václavů; Thomas D Coates; John C Wood Journal: J Appl Physiol (1985) Date: 2016-01-21