Machine learning based endothelial cell image analysis of patients undergoing descemet membrane endothelial keratoplasty surgery

OBJECTIVES

In this study, we developed a machine learning approach for postoperative corneal endothelial cell images of patients who underwent Descemet's membrane keratoplasty (DMEK).

METHODS

An AlexNet model is proposed and validated throughout the study for endothelial cell segmentation and cell location determination. The 506 images of postoperative corneal endothelial cells were analyzed. Endothelial cell detection, segmentation, and determining of its polygonal structure were identified. The proposed model is based on the training of an R-CNN to locate endothelial cells. Next, by determining the ridges separating adjacent cells, the density and hexagonality rates of DMEK patients are calculated.

RESULTS

The proposed method reached accuracy and F1 score rates of 86.15 % and 0.857, respectively, which indicates that it can reliably replace the manual detection of cells in vivo confocal microscopy (IVCM). The AUC score of 0.764 from the proposed segmentation method suggests a satisfactory outcome.

CONCLUSIONS

A model focused on segmenting endothelial cells can be employed to assess the health of the endothelium in DMEK patients.

CellsDeepNet: A Novel Deep Learning-Based Web Application for the Automated Morphometric Analysis of Corneal Endothelial Cells

The quantification of corneal endothelial cell (CEC) morphology using manual and semi-automatic software enables an objective assessment of corneal endothelial pathology. However, the procedure is tedious, subjective, and not widely applied in clinical practice. We have developed the CellsDeepNet system to automatically segment and analyse the CEC morphology. The CellsDeepNet system uses Contrast-Limited Adaptive Histogram Equalization (CLAHE) to improve the contrast of the CEC images and reduce the effects of non-uniform image illumination, 2D Double-Density Dual-Tree Complex Wavelet Transform (2DDD-TCWT) to reduce noise, Butterworth Bandpass filter to enhance the CEC edges, and moving average filter to adjust for brightness level. An improved version of U-Net was used to detect the boundaries of the CECs, regardless of the CEC size. CEC morphology was measured as mean cell density (MCD, cell/mm2), mean cell area (MCA, μm2), mean cell perimeter (MCP, μm), polymegathism (coefficient of CEC size variation), and pleomorphism (percentage of hexagonality coefficient). The CellsDeepNet system correlated highly significantly with the manual estimations for MCD (r = 0.94), MCA (r = 0.99), MCP (r = 0.99), polymegathism (r = 0.92), and pleomorphism (r = 0.86), with p < 0.0001 for all the extracted clinical features. The Bland–Altman plots showed excellent agreement. The percentage difference between the manual and automated estimations was superior for the CellsDeepNet system compared to the CEAS system and other state-of-the-art CEC segmentation systems on three large and challenging corneal endothelium image datasets captured using two different ophthalmic devices.

An automatic approach for cell detection and segmentation of corneal endothelium in specular microscope

PURPOSE

Specular microscopy is an indispensable tool for clinicians seeking to monitor the corneal endothelium. Automated methods of determining endothelial cell density (ECD) are limited in their ability to analyze images of poor quality. We describe and assess an image processing algorithm to analyze corneal endothelial images.

METHODS

A set of corneal endothelial images acquired with a Konan CellChek specular microscope was analyzed using three methods: flex-center, Konan Auto Tracer, and the proposed method. In this technique, the algorithm determines the region of interest, filters the image to differentiate cell boundaries from their interiors, and utilizes stochastic watershed segmentation to draw cell boundaries and assess ECD based on the masked region. We compared ECD measured by the algorithm with manual and automated results from the specular microscope.

RESULTS

We analyzed a total of 303 images manually, using the Auto Tracer, and with the proposed image processing method. Relative to manual analysis across all images, the mean error was 0.04% in the proposed method (p = 0.23 for difference) whereas Auto Tracer demonstrated a bias towards overestimation, with a mean error of 5.7% (p = 2.06× 10^-8). The relative mean absolute errors were 6.9% and 7.9%, respectively, for the proposed and Auto Tracer. The average time for analysis of each image using the proposed method was 2.5 s.

CONCLUSION

We demonstrate a computationally efficient algorithm to analyze corneal endothelial cell density that can be implemented on devices for clinical and research use.

Fully convolutional architecture vs sliding-window CNN for corneal endothelium cell segmentation

Background

Corneal endothelium (CE) images provide valuable clinical information regarding the health state of the cornea. Computation of the clinical morphometric parameters requires the segmentation of endothelial cell images. Current techniques to image the endothelium in vivo deliver low quality images, which makes automatic segmentation a complicated task. Here, we present two convolutional neural networks (CNN) to segment CE images: a global fully convolutional approach based on U-net, and a local sliding-window network (SW-net). We propose to use probabilistic labels instead of binary, we evaluate a preprocessing method to enhance the contrast of images, and we introduce a postprocessing method based on Fourier analysis and watershed to convert the CNN output images into the final cell segmentation. Both methods are applied to 50 images acquired with an SP-1P Topcon specular microscope. Estimates are compared against a manual delineation made by a trained observer.

Results

U-net (AUC=0.9938) yields slightly sharper, clearer images than SW-net (AUC=0.9921). After postprocessing, U-net obtains a DICE=0.981 and a MHD=0.22 (modified Hausdorff distance), whereas SW-net yields a DICE=0.978 and a MHD=0.30. U-net generates a wrong cell segmentation in only 0.48% of the cells, versus 0.92% for the SW-net. U-net achieves statistically significant better precision and accuracy than both, Topcon and SW-net, for the estimates of three clinical parameters: cell density (ECD), polymegethism (CV), and pleomorphism (HEX). The mean relative error in U-net for the parameters is 0.4% in ECD, 2.8% in CV, and 1.3% in HEX. The computation time to segment an image and estimate the parameters is barely a few seconds.

Conclusions

Both methods presented here provide a statistically significant improvement over the state of the art. U-net has reached the smallest error rate. We suggest a segmentation refinement based on our previous work to further improve the performance.