PURPOSE: To introduce, implement, and assess an iterative modification to the active deformational image segmentation method as applied to cervical cancer tumors. MATERIALS AND METHODS: A comparison by Jaccard similarity (JS) between this active deformational method and manual segmentation was performed on tumors of various sizes across preradiation, 3 weeks postradiation, and 6 weeks postradiation using a General Linear Mixed Model across 121 studies from 52 patients with Stage IIB-IV cervical cancers. RESULTS: The deformable segmentation method produced promising levels of agreement including JS factors of 0.71+/-0.11 in the preradiation studies. The analysis illustrated a rate of improvement in JS with increasing tumor volume that differed between the preradiation and 6 weeks postradiation stage (P=0.0474). In the large preradiated tumors each additional cm3 of volume was associated with an increase or improvement in JS of 0.0008 (95% confidence interval [CI]: 0.0003, 0.0014). In the smaller postradiation tumors, each additional cm3 of volume was associated with a more robust improvement in JS of 0.0046 (95% CI: 0.0009, 0.0082). CONCLUSION: Agreement was strongly affected by tumor volume, and its performance was most impacted across volume in the later stages of radiation therapy. The deformation-based segmentation method appears to demonstrate utility for delineating cervical cancer tumors, particularly in the earliest stages of radiation treatment, where agreement is greatest. Copyright (c) 2008 Wiley-Liss, Inc.
PURPOSE: To introduce, implement, and assess an iterative modification to the active deformational image segmentation method as applied to cervical cancer tumors. MATERIALS AND METHODS: A comparison by Jaccard similarity (JS) between this active deformational method and manual segmentation was performed on tumors of various sizes across preradiation, 3 weeks postradiation, and 6 weeks postradiation using a General Linear Mixed Model across 121 studies from 52 patients with Stage IIB-IV cervical cancers. RESULTS: The deformable segmentation method produced promising levels of agreement including JS factors of 0.71+/-0.11 in the preradiation studies. The analysis illustrated a rate of improvement in JS with increasing tumor volume that differed between the preradiation and 6 weeks postradiation stage (P=0.0474). In the large preradiated tumors each additional cm3 of volume was associated with an increase or improvement in JS of 0.0008 (95% confidence interval [CI]: 0.0003, 0.0014). In the smaller postradiation tumors, each additional cm3 of volume was associated with a more robust improvement in JS of 0.0046 (95% CI: 0.0009, 0.0082). CONCLUSION: Agreement was strongly affected by tumor volume, and its performance was most impacted across volume in the later stages of radiation therapy. The deformation-based segmentation method appears to demonstrate utility for delineating cervical cancer tumors, particularly in the earliest stages of radiation treatment, where agreement is greatest. Copyright (c) 2008 Wiley-Liss, Inc.
Authors: Nilanjan Ray; Scott T Acton; Talissa Altes; Eduard E de Lange; James R Brookeman Journal: IEEE Trans Med Imaging Date: 2003-02 Impact factor: 10.048
Authors: Mark J Beresford; Anwar R Padhani; N Jane Taylor; Mei-Lin Ah-See; J James Stirling; Andreas Makris; James A d'Arcy; David J Collins Journal: J Magn Reson Imaging Date: 2006-12 Impact factor: 4.813
Authors: Dee H Wu; Nina A Mayr; Yasemin Karatas; Rifat Karatas; Mustafa Adli; Susan M Edwards; James D Wolff; Allen Movahed; Joseph F Montebello; William T C Yuh Journal: J Appl Clin Med Phys Date: 2005-11-21 Impact factor: 2.102
Authors: Rheal A Towner; Amy B Wisniewski; Dee H Wu; Samuel B Van Gordon; Nataliya Smith; Justin C North; Rayburt McElhaney; Christopher E Aston; S Abbas Shobeiri; Bradley P Kropp; Beverley Greenwood-Van Meerveld; Robert E Hurst Journal: J Urol Date: 2015-08-22 Impact factor: 7.450