2017/08/17 by Arash Mehrjou, Mehrjou, Arash
Biochemistry, Genetics and Molecular Biology · Computer Science · Engineering · #Digital Imaging for Blood Diseases #FOS: Computer and information sciences #Identification and Quantification in Food #Image Processing Techniques and Applications #Machine Learning (stat.ML)
paper · pdf · doi:10.48550/arxiv.1708.05200
openalex publication_date 2017/08/17 · openalex created_date 2022/09/30 · openalex updated_date 2026/07/28
Malaria is a serious infectious disease that is responsible for over half\nmillion deaths yearly worldwide. The major cause of these mortalities is late\nor inaccurate diagnosis. Manual microscopy is currently considered as the\ndominant diagnostic method for malaria. However, it is time consuming and prone\nto human errors. The aim of this paper is to automate the diagnosis process and\nminimize the human intervention. We have developed the hardware and software\nfor a cost-efficient malaria diagnostic system. This paper describes the\nmanufactured hardware and also proposes novel software to handle parasite\ndetection and life-stage identification. A motorized microscope is developed to\ntake images from Giemsa-stained blood smears. A patch-based unsupervised\nstatistical clustering algorithm is proposed which offers a novel method for\nclassification of different regions within blood images. The proposed method\nprovides better robustness against different imaging settings. The core of the\nproposed algorithm is a model called Mixture of Independent Component Analysis.\nA manifold based optimization method is proposed that facilitates the\napplication of the model for high dimensional data usually acquired in medical\nmicroscopy. The method was tested on 600 blood slides with various imaging\nconditions. The speed of the method is higher than current supervised systems\nwhile its accuracy is comparable to or better than them.\n