Fig 1.
(a)-(b) Two example patients imaged with 3D T1-Weighted (T1w) VIBE & T2-Weighted (T2w) MRI (c) The eye lens (dark blue), vitreous humor (red) and sclera / cornea (light blue) are highlighted. Endophytic tumors delineated in yellow. All annotated regions were delineated manually by an expert radiologist.
Fig 2.
Proposed framework for automatic whole eye segmentation.
T1w and T2w 3D volumes are combined with EPSF features. These features are used to train a RF / Convolutional Neural Network (CNN) classifier, serving as the data term in Graph-Cut optimization.
Fig 3.
Learning a Pathological Eye Model (PM).
We follow the steps in [15] to (a) automatically detect the eye in the 3D MRI, followed by a set of b) image pre-processing techniques to learn information of pathological and healthy structures jointly using c) intensity profiles containing pathological information.
Table 1.
Eye anatomy DSC: Our Pathological Model (PM) shows more accurate results than the Healthy Model (HM) from [15] and the Combined Model (CM), especially for the region of the lens.
(*) p < 0.05. The table indicates both the Dice Similarity Coefficient (DSC) and the maximum surface segmentation error or Hausdorff Distance (HD), in mm.
Fig 4.
(a) ROC curve depicting the effect of varying amounts of (training/test) data for RF classification with and without EPSF. (b) ROC curve for both experiments (CNN / RF) with STD and with EPSF. (c) DSC tumor segmentation results for STD vs. EPSF. The latter shows better results for both cases (** = p < 0.01).
Table 2.
DSC performance for different scenarios before and after Graph-cut (GC) inference.
Hausdorff Distance (HD) and Mean Distance Error after GC inference. Complete results in S1 Table. Experiments were computed on ⋄: Macbook Pro Intel-Core™ i7 16GB—2, 5 GHZ & †: Intel-Core™ i7 6700 32GB with Nvidia GTX Titan X®.
Fig 5.
Average results for different combination of classifiers and feature sets. EPSF improves overall classification results over STD features consistently.
Fig 6.
the tumor ground truth is delineated in red. Probability Maps, P(fi|Yi), for worst (RF-STD) and best (CNN-EPSF) scenarios. Final column shows the pathological model (PM) eye segmentation results.