Image processing algorithms for medical applications

• Main Author: Sharveena D/O Mohan
• Other Authors: Mohammed Yakoob Siyal
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; Image processing
• Online Access: https://hdl.handle.net/10356/176777
• Institution: Nanyang Technological University

Image processing algorithms are significant in medical image processing. There are various algorithms that improves the overall quality of the medical scans and assist practitioners in making accurate diagnosis, providing patient the right course of medical treatment. Algorithms found in medical imaging processing includes but not limited to, noise removal and tumour detection. Noise is a significant challenge faced in medical imaging as it can corrupt the image. If there is too much noise present in the image, the scans might be blurry hence, it will be difficult for practitioners to make a diagnosis. This can lead to inaccurate diagnosis which is dangerous for the patient. Thus, noise removal is a necessary feature during image processing. This project aims to experiment on how different filters can effectively remove noise present in an image and how these filters affect abnormality detection. This will be conducted using a Graphic User Interface (GUI) which will be implemented on MATLAB. Four different imaging modalities will be used to conduct the experiment. The imaging modalities used are Ultrasound, Computed Tomography (CT), Magnetic Resonance Imaging (MRI) and X-Ray. The four noises used in this project are, Salt and Pepper, Speckle, Gaussian, and Poisson noise. The four filters used to remove the different noises are, Median, Mean, Gaussian, and Wiener filter. Performance Evaluation will be used to mathematically compute the effectiveness of the four filters. The performance evaluation used in this project are, PSNR, MSE and SSIM. Various edge detection techniques are included to identify the edges of the features of the image to detect any anomaly. Lastly, an automated tumour detection function is included in the GUI to test how accurate the program can detect the tumours present in the images for each imaging modality.