Computerized macular pathology diagnosis in spectral domain optical coherence tomography scans based on multiscale texture and shape features

Abstract

Purpose: To develop an automated method to identify the normal macula and three macular pathologies (macular hole [MH], macular edema [ME], and age-related macular degeneration [AMD]) from the fovea-centered cross sections in three-dimensional (3D) spectral-domain optical coherence tomography (SD-OCT) images.

Methods: A sample of SD-OCT macular scans (macular cube 200 × 200 or 512 × 128 scan protocol; Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, CA) was obtained from healthy subjects and subjects with MH, ME, and/or AMD (dataset for development: 326 scans from 136 subjects [193 eyes], and dataset for testing: 131 scans from 37 subjects [58 eyes]). A fovea-centered cross-sectional slice for each of the SD-OCT images was encoded using spatially distributed multiscale texture and shape features. Three ophthalmologists labeled each fovea-centered slice independently, and the majority opinion for each pathology was used as the ground truth. Machine learning algorithms were used to identify the discriminative features automatically. Two-class support vector machine classifiers were trained to identify the presence of normal macula and each of the three pathologies separately. The area under the receiver operating characteristic curve (AUC) was calculated to assess the performance.

Results: The cross-validation AUC result on the development dataset was 0.976, 0.931, 0939, and 0.938, and the AUC result on the holdout testing set was 0.978, 0.969, 0.941, and 0.975, for identifying normal macula, MH, ME, and AMD, respectively.

Conclusions: The proposed automated data-driven method successfully identified various macular pathologies (all AUC > 0.94). This method may effectively identify the discriminative features without relying on a potentially error-prone segmentation module.

Figure 1.

Stages of our approach. Morphologic op., morphologic operations; LBP, local binary patterns; PCA, principle component analysis; SVM, support vector machine.

Figure 2.

Venn diagram of labeling agreement among the three experts on all macular categories in database A. The actual scan numbers and the percentage are both shown. E1, E2, E3, the three experts.

Figure 3.

The number of cases and representative examples in database A where all three experts, two experts, or only one expert gave “positive” labels for the presence of normal macula and each pathology. Note that images in the first two columns were defined as positive while the ones in the last column were regarded as negative in our majority-opinion–based ground truth. The images without total agreement usually contain early pathologies that were subtle and occupied small areas.

Figure 4.

Examples of the aligned retinal images and their Canny edge maps derived under different edge-detection thresholds t for each macular category. The smaller the value of t, the more edges are retained.

Figure 5.

ROC curve of one run of 10-fold cross-validation on all images in dataset A. The best feature setting for each macular pathology was used. Feature setting: TS (t = 0.4), S (t = 0.4), TS (t = 0.4), and TS (t = 0.2) for NM, MH, ME, and AMD, respectively.

Figure 6.

AUC results with respect to a varied training set size from dataset A. For each training fold, 10%, 20%, …, 100% of the positive and negative subjects were sampled and used for training, whereas the testing fold was unchanged. Feature setting: TS (t = 0.4), S (t = 0.4), TS (t = 0.4), and TS (t = 0.2) for NM, MH, ME, and AMD, respectively.

Figure 7.

ROC curve of testing on dataset B, based on the pathology classifiers trained using images from dataset A. The ground truth for this experiment was defined by the consensus of the two experts (experts 1 and 2) on both datasets. The statistics of pathology distribution are shown in Table 6. The feature and parameter setting for each pathology was determined using dataset A only. Feature setting: TS (t = 0.4), S (t = 0.4), TS (t = 0.4), and TS (t = 0.2) for NM, MH, ME, and AMD, respectively.