当前位置: X-MOL 学术IETE J. Res. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Automatic Detection of Malaria Infected Erythrocytes Based on the Concavity Point Identification and Pseudo-Valley Based Thresholding
IETE Journal of Research ( IF 1.3 ) Pub Date : 2020-07-16 , DOI: 10.1080/03772063.2020.1787238
Manish Sharma 1 , Salam Shuleenda Devi 1 , Rabul Hussain Laskar 1
Affiliation  

Automatic malaria diagnosis requires the segmentation of the erythrocytes from the microscopic blood smears images. In this paper, the system for the analysis of the erythrocyte to detect the malaria infection has been proposed. The challenging issues are: preprocessing, foreground extraction, clump erythrocyte segmentation and infected erythrocyte detection. The non-uniform illumination and other distortions of the acquired microscopic images are corrected by using median filtering technique. The foreground regions, i.e. erythrocyte regions are segmented from the other blood components in the smears by using Otsu’s thresholding along with morphological filtering. The clump erythrocytes are segmented based on the inherent geometrical property of the concavity point detection of the clump which is independent of any geometrical deformation or image acquisition errors. Further, the segmented erythrocytes are analysed to detect the infected erythrocytes using thresholding technique based on the “pseudo-valley” concept. The experimental analyses of the proposed method are carried out on three different sets of database. An accuracy of 97.95% was observed for the erythrocyte segmentation, with an improvement in accuracy of 8.76% and 0.87% as compared to classical watershed and marker controlled watershed segmentation technique respectively. Moreover, the infected erythrocyte was detected with an accuracy of 88.57%.



中文翻译:

基于凹点识别和基于伪谷阈值的疟疾感染红细胞自动检测

自动疟疾诊断需要从显微血涂片图像中分割出红细胞。在本文中,提出了用于检测疟疾感染的红细胞分析系统。具有挑战性的问题是:预处理、前景提取、红细胞团块分割和感染红细胞检测。利用中值滤波技术对采集到的显微图像的光照不均匀等畸变进行校正。前景区域,通过使用 Otsu 的阈值和形态学过滤,将红细胞区域从涂片中的其他血液成分中分离出来。基于团块的凹点检测的固有几何特性对团块红细胞进行分割,该几何特性独立于任何几何变形或图像采集误差。此外,使用基于“伪谷”概念的阈值技术分析分割的红细胞以检测感染的红细胞。在三组不同的数据库上对所提出的方法进行了实验分析。观察到红细胞分割的准确率为 97.95%,与经典分水岭和标记控制的分水岭分割技术相比,准确度分别提高了 8.76% 和 0.87%。而且,

更新日期:2020-07-16
down
wechat
bug