Three-dimensional choroid neovascularization growth prediction from longitudinal retinal OCT images based on a hybrid model

Abstract Choroid neovascularization (CNV) is a pathological manifestation of retinal-choroidal diseases such as age-related macular degeneration and pathological myopia, which can cause permanent loss of central vision. Prediction of its growth is important in treatment planning. In this paper, based on longitudinal optical coherence tomography (OCT) volumes, a three-dimensional CNV growth prediction framework is proposed. A hybrid model which combines the reaction-diffusion model and the hyperelastic biomechanical model through mass effect is adopted to characterize the growth of CNV region and its reaction with surround tissues. A treatment factor is also included so that the model can adjust to different treatment plan each patient receives. Tested on a dataset with 6 subjects, each with 12 longitudinal 3D images, the proposed method achieved average true positive rate (TPR), false positive rate (FPR) and Dice coefficient (DC) of 80.0 ± 7.62%, 23.4 ± 8.36% and 78.9 ± 7.54%, respectively, in predicting the future CNV regions, and outperforms those achieved by the single reaction-diffusion model.

[1]  Gerhard A. Holzapfel,et al.  Nonlinear Solid Mechanics: A Continuum Approach for Engineering Science , 2000 .

[2]  Xinjian Chen,et al.  Automated 3-D Retinal Layer Segmentation of Macular Optical Coherence Tomography Images With Serous Pigment Epithelial Detachments , 2015, IEEE Transactions on Medical Imaging.

[3]  Ronald M. Summers,et al.  Tumor Growth Prediction with Hyperelastic Biomechanical Model, Physiological Data Fusion, and Nonlinear Optimization , 2014, MICCAI.

[4]  Xinjian Chen,et al.  3D choroid neovascularization growth prediction based on reaction-diffusion model , 2016, SPIE Medical Imaging.

[5]  Kyu Hyung Park,et al.  Association between choroidal morphology and anti-vascular endothelial growth factor treatment outcome in myopic choroidal neovascularization. , 2013, Investigative ophthalmology & visual science.

[6]  Bingbing Ni,et al.  Image Classification by Selective Regularized Subspace Learning , 2016, IEEE Transactions on Multimedia.

[7]  R. Tadayoni,et al.  INJECT AND EXTEND DOSING VERSUS DOSING AS NEEDED: A Comparative Retrospective Study of Ranibizumab in Exudative Age-Related Macular Degeneration , 2011, Retina.

[8]  P. Rosenfeld,et al.  Visual Acuity Outcomes Following a Variable–Dosing Regimen for Ranibizumab (LucentisTM) in Neovascular AMD: The PrONTO Study , 2006 .

[9]  Bianca S. Gerendas,et al.  Machine Learning to Analyze the Prognostic Value of Current Imaging Biomarkers in Neovascular Age-Related Macular Degeneration. , 2018, Ophthalmology. Retina.

[10]  K. Bathe.  Finite Element Procedures , 1995 .

[11]  Christos Davatzikos,et al.  Finite Element Modeling of Brain Tumor Mass-Effect from 3D Medical Images , 2005, MICCAI.

[12]  Hervé Delingette,et al.  Realistic simulation of the 3-D growth of brain tumors in MR images coupling diffusion with biomechanical deformation , 2005, IEEE Transactions on Medical Imaging.

[13]  Ronald M. Summers,et al.  Kidney Tumor Growth Prediction by Coupling Reaction–Diffusion and Biomechanical Model , 2013, IEEE Transactions on Biomedical Engineering.

[14]  Xinjian Chen,et al.  Three-Dimensional Segmentation of Fluid-Associated Abnormalities in Retinal OCT: Probability Constrained Graph-Search-Graph-Cut , 2012, IEEE Transactions on Medical Imaging.

[15]  Bianca S. Gerendas,et al.  Prediction of Anti-VEGF Treatment Requirements in Neovascular AMD Using a Machine Learning Approach. , 2017, Investigative ophthalmology & visual science.

[16]  Kalyanmoy Deb,et al.  A fast and elitist multiobjective genetic algorithm: NSGA-II , 2002, IEEE Trans. Evol. Comput..

[17]  Xinjian Chen,et al.  Choroid Neovascularization Growth Prediction With Treatment Based on Reaction-Diffusion Model in 3-D OCT Images , 2017, IEEE Journal of Biomedical and Health Informatics.

[18]  Milan Sonka,et al.  Prediction of treatment response from retinal OCT in patients with exudative age-related macular degeneration , 2014 .

[19]  P. Campochiaro,et al.  VEGF is major stimulator in model of choroidal neovascularization. , 2000, Investigative ophthalmology & visual science.

[20]  Ronald M. Summers,et al.  FEM-Based 3-D Tumor Growth Prediction for Kidney Tumor , 2010, IEEE Transactions on Biomedical Engineering.

[21]  Christos Davatzikos,et al.  An image-driven parameter estimation problem for a reaction–diffusion glioma growth model with mass effects , 2008, Journal of mathematical biology.

[22]  Yue Gao,et al.  View-Based Discriminative Probabilistic Modeling for 3D Object Retrieval and Recognition , 2013, IEEE Transactions on Image Processing.

[23]  Meng Wang,et al.  Visual Classification by ℓ1-Hypergraph Modeling , 2015, IEEE Trans. Knowl. Data Eng..

[24]  Jiashi Feng,et al.  Zero-Shot Learning via Attribute Regression and Class Prototype Rectification. , 2018, IEEE transactions on image processing : a publication of the IEEE Signal Processing Society.

[25]  Lawrence A. Yannuzzi,et al.  PRECLINICAL AND PHASE 1A CLINICAL EVALUATION OF AN ANTI-VEGF PEGYLATED APTAMER (EYE001) FOR THE TREATMENT OF EXUDATIVE AGE-RELATED MACULAR DEGENERATION , 2002, Retina.

[26]  J. Friedman.  Special Invited Paper-Additive logistic regression: A statistical view of boosting , 2000 .

[27]  I J Constable,et al.  Overexpression of vascular endothelial growth factor (VEGF) in the retinal pigment epithelium leads to the development of choroidal neovascularization. , 2000, The American journal of pathology.