A Red Blood Cell Recognition System of Anemia Based on Blood smear Images.
碩士 === 國立中興大學 === 資訊管理學系所 === 101 === Anemia comes from being enable to produce enough hemoglobin by the body itself, and let outer red blood cells (RBCs) blood volume lower than normal value. In blood testing, evaluating whether a person has anemia usually uses hematology analyzer to analysis some...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2013
|
Online Access: | http://ndltd.ncl.edu.tw/handle/22038716073211562912 |
id |
ndltd-TW-101NCHU5396025 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-101NCHU53960252015-10-13T22:35:50Z http://ndltd.ncl.edu.tw/handle/22038716073211562912 A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. 以血液抹片為基礎之貧血症紅血球細胞辨識系統 Ming-Jun Su 蘇明俊 碩士 國立中興大學 資訊管理學系所 101 Anemia comes from being enable to produce enough hemoglobin by the body itself, and let outer red blood cells (RBCs) blood volume lower than normal value. In blood testing, evaluating whether a person has anemia usually uses hematology analyzer to analysis some factors, such as Hemoglobin (Hb), Red blood cell (RBC), mean corpuscular volume (MCV), etc. However, hematology analyzer is very expensive for general clinics. In order to solve this problem, this thesis proposes a new estimation method based on 2-dimensions blood smear images to develop a mechanism that can help doctors to do diagnosis. In the blood smear images, the normal diameter of single RBC is in the range from 6um to 9um. The RBCs with diameter smaller than 6um are called “Small blood type anemia” and those with diameter larger than 9um are called “Big blood type anemia.” The proposed method judges whether the red blood cells belong to one of five anemia blood classes according to the shape of single RBC in blood smear images. These five classes are Acanthocytes, Schistocytes, Round, Oval, and Teardrop. There are four steps in our proposed method. The first step combines Top Hat Transform, Bottom Hat Transform, and Otsu’s thresholding to segment initial RBC regions. The second step uses circular detecting method and adaptive k-means to obtain the single normal RBC region. The third step deletes the region which is small than -times single normal RBC mean area. The region which is larger than -times single normal RBC mean area is treated as overlapping region. The mathematical morphology is used to segment the multiple overlapping RBCs into single RBC. The fourth step classifies the segmentation results into one of five anemia blood classes. Moreover, a genetic algorithm is used to determine the proper parameters used in our proposed method. Yung-Kuan Chan 詹永寬 2013 學位論文 ; thesis 54 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立中興大學 === 資訊管理學系所 === 101 === Anemia comes from being enable to produce enough hemoglobin by the body itself, and let outer red blood cells (RBCs) blood volume lower than normal value. In blood testing, evaluating whether a person has anemia usually uses hematology analyzer to analysis some factors, such as Hemoglobin (Hb), Red blood cell (RBC), mean corpuscular volume (MCV), etc. However, hematology analyzer is very expensive for general clinics. In order to solve this problem, this thesis proposes a new estimation method based on 2-dimensions blood smear images to develop a mechanism that can help doctors to do diagnosis. In the blood smear images, the normal diameter of single RBC is in the range from 6um to 9um. The RBCs with diameter smaller than 6um are called “Small blood type anemia” and those with diameter larger than 9um are called “Big blood type anemia.” The proposed method judges whether the red blood cells belong to one of five anemia blood classes according to the shape of single RBC in blood smear images. These five classes are Acanthocytes, Schistocytes, Round, Oval, and Teardrop. There are four steps in our proposed method. The first step combines Top Hat Transform, Bottom Hat Transform, and Otsu’s thresholding to segment initial RBC regions. The second step uses circular detecting method and adaptive k-means to obtain the single normal RBC region. The third step deletes the region which is small than -times single normal RBC mean area. The region which is larger than -times single normal RBC mean area is treated as overlapping region. The mathematical morphology is used to segment the multiple overlapping RBCs into single RBC. The fourth step classifies the segmentation results into one of five anemia blood classes. Moreover, a genetic algorithm is used to determine the proper parameters used in our proposed method.
|
author2 |
Yung-Kuan Chan |
author_facet |
Yung-Kuan Chan Ming-Jun Su 蘇明俊 |
author |
Ming-Jun Su 蘇明俊 |
spellingShingle |
Ming-Jun Su 蘇明俊 A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. |
author_sort |
Ming-Jun Su |
title |
A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. |
title_short |
A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. |
title_full |
A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. |
title_fullStr |
A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. |
title_full_unstemmed |
A Red Blood Cell Recognition System of Anemia Based on Blood smear Images. |
title_sort |
red blood cell recognition system of anemia based on blood smear images. |
publishDate |
2013 |
url |
http://ndltd.ncl.edu.tw/handle/22038716073211562912 |
work_keys_str_mv |
AT mingjunsu aredbloodcellrecognitionsystemofanemiabasedonbloodsmearimages AT sūmíngjùn aredbloodcellrecognitionsystemofanemiabasedonbloodsmearimages AT mingjunsu yǐxuèyèmǒpiànwèijīchǔzhīpínxuèzhènghóngxuèqiúxìbāobiànshíxìtǒng AT sūmíngjùn yǐxuèyèmǒpiànwèijīchǔzhīpínxuèzhènghóngxuèqiúxìbāobiànshíxìtǒng AT mingjunsu redbloodcellrecognitionsystemofanemiabasedonbloodsmearimages AT sūmíngjùn redbloodcellrecognitionsystemofanemiabasedonbloodsmearimages |
_version_ |
1718078674193874944 |