An intelligent diagnostic system for malaria
Malaria is a serious health problem, causing many deaths and morbidity cases throughout the world particularly in Africa and south Asia. In 2013, there were about 198 million cases of malaria and an estimation of 584,000 deaths recorded globally, mostly among African children. Malaria is caused by...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Subjects: | |
| Online Access: | http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/1/Page%201-24.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/2/Full%20text.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| id |
my-unimap-61534 |
|---|---|
| record_format |
uketd_dc |
| institution |
Universiti Malaysia Perlis |
| collection |
UniMAP Institutional Repository |
| language |
English |
| advisor |
Prof. Dr. Mohd Yusoff Mashor |
| topic |
Malaria Malaria diagnosis Malarsia parasite Intelligent diagnostic system |
| spellingShingle |
Malaria Malaria diagnosis Malarsia parasite Intelligent diagnostic system Aimi Salihah, Abdul Nasir An intelligent diagnostic system for malaria |
| description |
Malaria is a serious health problem, causing many deaths and morbidity cases throughout the world particularly in Africa and south Asia. In 2013, there were about 198 million cases
of malaria and an estimation of 584,000 deaths recorded globally, mostly among African children. Malaria is caused by infection of red blood cells with protozoan parasite of the
genus Plasmodium. Plasmodium Falciparum and Plasmodium Vivax are the two main
species that have caused the most malaria infections worldwide. Malaria can become lifethreatening if it is not treated immediately. Until now, microscopy-based diagnosis still remains the most widely used approaches for malaria diagnosis. However, this subjective evaluation procedure is time consuming, labour intensive and requires special training.
Thus, this research has developed an intelligent diagnostic system for malaria which consists of image processing and intelligent classifier for the purpose of malaria diagnosis. A 3-stage classification of intelligent diagnostic system can be used as an early detection
for malaria based on the classification of blood samples between normal and malaria on the
first stage, and further classify the malaria sample as either P. falciparum or P. vivax
species on the second stage, along with its four different life-cycle stages which are young
trophozoite, mature trophozoite, schizont and gametocyte on the final stage. In order to
perform the diagnosis process, the blood images were processed with various image
processing techniques such as contrast enhancement and image segmentation for obtaining
a fully segmented malaria parasite. As for contrast enhancement, this study proposed
modified global and modified linear contrast stretching based on total pixel approach, as
well as modified global and modified linear contrast stretching based on pixel level
approach. After the image has been enhanced, the malaria image was segmented using
different types of clustering algorithms. This included the used of the proposed enhanced kmeans
clustering. The combination between contrast enhancement and image segmentation
have provided good segmented malaria parasite. Later, various features such as size, shape
and colour based features were extracted from the segmented malaria parasite. These
features were fed as inputs to the three different classifiers which are multilayered
perceptron (MLP) neural network trained by Levenberg-Marquardt (LM) algorithm, singlehidden
layer feed forward neural network (SLFN) trained by online sequential extreme
learning machine (OS-ELM) algorithm and random forest. The MLP network trained by
LM algorithm has been proven to be the best with the highest classification performance as
compared to others. Overall, the intelligent diagnostic system for malaria that has been
developed using MLP network trained by LM algorithm is capable to perform the detection
process by classifying a total of 1800 images consisting of malaria and normal blood
images with testing accuracy, sensitivity and specificity of 95.28%, 96.06% and 86.00%,
respectively. As for the diagnosis process, the system has classified a total of 1453 malaria
images (accuracy of 90.81%) correctly into P. falciparum and P. vivax species, along with
their four life-cycle stages. Thus, the proposed intelligent diagnostic system for malaria
parasites is capable to perform the detection of malaria parasites, and then further diagnose
the detected malaria parasites into its species and life-cycle stages. |
| format |
Thesis |
| author |
Aimi Salihah, Abdul Nasir |
| author_facet |
Aimi Salihah, Abdul Nasir |
| author_sort |
Aimi Salihah, Abdul Nasir |
| title |
An intelligent diagnostic system for malaria |
| title_short |
An intelligent diagnostic system for malaria |
| title_full |
An intelligent diagnostic system for malaria |
| title_fullStr |
An intelligent diagnostic system for malaria |
| title_full_unstemmed |
An intelligent diagnostic system for malaria |
| title_sort |
intelligent diagnostic system for malaria |
| granting_institution |
Universiti Malaysia Perlis (UniMAP) |
| granting_department |
School of Mechatronics Engineering |
| url |
http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/1/Page%201-24.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/2/Full%20text.pdf |
| _version_ |
1747836843124064256 |
| spelling |
my-unimap-615342019-08-22T06:48:58Z An intelligent diagnostic system for malaria Aimi Salihah, Abdul Nasir Prof. Dr. Mohd Yusoff Mashor Malaria is a serious health problem, causing many deaths and morbidity cases throughout the world particularly in Africa and south Asia. In 2013, there were about 198 million cases of malaria and an estimation of 584,000 deaths recorded globally, mostly among African children. Malaria is caused by infection of red blood cells with protozoan parasite of the genus Plasmodium. Plasmodium Falciparum and Plasmodium Vivax are the two main species that have caused the most malaria infections worldwide. Malaria can become lifethreatening if it is not treated immediately. Until now, microscopy-based diagnosis still remains the most widely used approaches for malaria diagnosis. However, this subjective evaluation procedure is time consuming, labour intensive and requires special training. Thus, this research has developed an intelligent diagnostic system for malaria which consists of image processing and intelligent classifier for the purpose of malaria diagnosis. A 3-stage classification of intelligent diagnostic system can be used as an early detection for malaria based on the classification of blood samples between normal and malaria on the first stage, and further classify the malaria sample as either P. falciparum or P. vivax species on the second stage, along with its four different life-cycle stages which are young trophozoite, mature trophozoite, schizont and gametocyte on the final stage. In order to perform the diagnosis process, the blood images were processed with various image processing techniques such as contrast enhancement and image segmentation for obtaining a fully segmented malaria parasite. As for contrast enhancement, this study proposed modified global and modified linear contrast stretching based on total pixel approach, as well as modified global and modified linear contrast stretching based on pixel level approach. After the image has been enhanced, the malaria image was segmented using different types of clustering algorithms. This included the used of the proposed enhanced kmeans clustering. The combination between contrast enhancement and image segmentation have provided good segmented malaria parasite. Later, various features such as size, shape and colour based features were extracted from the segmented malaria parasite. These features were fed as inputs to the three different classifiers which are multilayered perceptron (MLP) neural network trained by Levenberg-Marquardt (LM) algorithm, singlehidden layer feed forward neural network (SLFN) trained by online sequential extreme learning machine (OS-ELM) algorithm and random forest. The MLP network trained by LM algorithm has been proven to be the best with the highest classification performance as compared to others. Overall, the intelligent diagnostic system for malaria that has been developed using MLP network trained by LM algorithm is capable to perform the detection process by classifying a total of 1800 images consisting of malaria and normal blood images with testing accuracy, sensitivity and specificity of 95.28%, 96.06% and 86.00%, respectively. As for the diagnosis process, the system has classified a total of 1453 malaria images (accuracy of 90.81%) correctly into P. falciparum and P. vivax species, along with their four life-cycle stages. Thus, the proposed intelligent diagnostic system for malaria parasites is capable to perform the detection of malaria parasites, and then further diagnose the detected malaria parasites into its species and life-cycle stages. Universiti Malaysia Perlis (UniMAP) 2015 Thesis en http://dspace.unimap.edu.my:80/xmlui/handle/123456789/61534 http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/1/Page%201-24.pdf 75105b2980b98eb4e02b4636b0340b52 http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/2/Full%20text.pdf 784bfb56f6ca012a09c176f26edd7b95 http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/61534/3/license.txt 8a4605be74aa9ea9d79846c1fba20a33 Malaria Malaria diagnosis Malarsia parasite Intelligent diagnostic system School of Mechatronics Engineering |