Geometric mean filter with grayscale morphological method to enhance the RNFL thickness in the SD-OCT images

Abstract

Alzheimer disease plays an important role in day-to-day life. Alzheimer disease is mainly affects the creative thinking, memory and other day-to-day activities. In general, Alzheimer disease influences more on the people who are in the age group of 80-year-90. A recent report from The Hindu states that nearly 50 lakhs of people are affected by Alzheimer disease in India. The essential goal of Optical Coherence Tomography (OCT) imaging technology is to identify the Retinal Nerve Fiber Layer (RNFL) thickness that originally detects the Alzheimer disease status in patients. Spectral-domain (SD) OCT devices have produces OCT images in greater speed; hence, fluctuation of sampling will be high. Open-top hat filter dose not perform well when over fluctuation of sampling of input OCT images. Geometric mean filter is not much affected by fluctuation of sampling. Hence, geometric mean filter with grayscale morphological method is used in this paper to remove the noises as well as enhance the RNFL thickness in the input SD-OCT images. The proposed Geometric mean filter with grayscale morphological operation method is comparatively analyzed with the our other existing approaches proposed earlier namely, various image enhancement approaches such as Otsu’s method with Geometric mean filter, Switching Median Filter (SMF) with morphological operation, and Open Top Hat filter with Goldilocks method. The experimental results prove that the proposed geometric mean filter with grayscale morphological performed well when compared other approaches. The proposed algorithm is comparatively analyzed with the help of Sensitivity, Specificity, Accuracy, Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR).