Purpose:
In order to improve the accuracy in the manual measurements of the choroidal thickness and to facilitate further the development of automatic choroidal segmentation, we aimed to design and validate algorithm for automated enhanced visualization of the choroidal boundaries in images obtained via optical coherence tomography (OCT) by the method of enhanced depth imaging (EDI).
Methods:
Images were obtained with Heidelberg Spectralis OCT using EDI mode. Fast programming languages were used to implement an algorithm. Various approaches were tested on representing set of images to refine the selection properties and minimize possible image deterioration from digital processing. Fifty images from fifty eyes were used to validate the algorithm. A graphical user interface (GUI) was designed for facile manipulation and improved visualization. The tests were performed on computer with Intel® Core™2 Quad Processor under Microsoft Windows 7 OS. The enhanced images generated by the new algorithm were evaluated by two masked observers and compared to the original non-enhanced images. Only one image from each data set was used. Twenty five images were regraded to asses the reproducibility.
Results:
Fast algorithm for automatic enhancement of images was generated. The enhanced images were presented side by side with the original image in specially designed GUI. There was 95.5% agreement between the observers and 98% within the observers regarding the superiority of the delineation in the enhanced images. Reproducibility of the generated by the algorithm images was tested on half of the images. There was 95% agreement between the observers. Each image was generated for approximately 1 second.
Conclusions:
Robust automatic algorithm for enhanced choroidal boundary delineation in OCT images was designed and validated. The improved boundary contour generated by our software may significantly contribute in improvement of the accuracy and reproducibility of the manual measurements of the choroidal thickness as required by clinical trials and facilitate development of automatic choroidal segmentation algorithms.
Keywords: choroid • imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)