Image processing algorithms for medical applications

Abstract

Image processing algorithms plays an important role in the image processing world. By using applying various algorithm to noisy images, the overall quality of the image can be substantially improved. The presence of noise in medical images may affect the doctor’s ability to diagnose a patient. In rare cases, a misdiagnosis may endanger the patient’s life, hence the importance of noise removal. In the project, four noise reduction filters (Mean, Gaussian, Wiener and Median filter) are simulated and tested. Four image quality metrics (Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM) and Perception based Image Quality Evaluator (PIQE)) is used for checking the post-filter image quality. This project is also testing on four different medical image types (X-Ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET)) from different scanning techniques. The different image types are artificially corrupted with four different types of noises (Gaussian, Poisson, Salt-and-Pepper and Speckle noise). In order to reach a fair conclusion, twelve images from each type of medical image are simulated, totaling to forty-eight images. By comparing the average values from each of the four quality metrics, a conclusion for the most effective filter can be derived.