Abstract:
Total of 104 countries and territories were considered malaria endemic in 2013, where 97 in malaria transmission phase and 7 with prevention phase. Severity of malaria can be estimated, as globally 3.4 billion individuals are at malarial risk and 207 million cases occurred with 627000 deaths in 2012. Parasite of plasmodium genus caused malaria in animals, birds and more specifically five of them i.e. p. ovale, p. knowlesi, p. falciparum, plasmodium, p. malariae and p. vivax effect humans. Severe malarial anaemia (SA), respiratory distress (RD) and cerebral malaria (CM) are common malarial syndromes. Malarial retinopathy is associated with retinal changes such as retinal whitening, hemorrhages, vessels discoloration and papilloedema occurred as a result of malaria and obsevered in more than 50% patients. Retinal changes in severe malaria such as hemorrhage are observed over 130 years and unique retinal signs first described in 1993 in Africa. This research focuses on reliable extraction of hemorrhages and retinal whitening which are major signs of malarial retinopathy. The proposed system detects all possible hemorrhages for grading of input retinal image using image processing and machine learning techniques. The retinal whitening is detected by extracting different image level features and using support vector machine. The validity of proposed system is tested using locally gathered retinal images and results show the significance of proposed system.
