New Technology Quickly Diagnoses Malaria

The gold standard for malaria diagnosis is manual microscopic evaluation of Giemsa stained blood smears; however, the utility of this approach is limited by the skill of an expert microscopist. Further, both the staining process and microscopic examination can be time consuming.

A computerized method has been developed that relies on computers and light-based holographic scans correctly identified malaria-infected cells in a blood sample and this technique does not require any human intervention and could form the basis for a rapid field test for malaria.


A multidisciplinary team of scientists from Duke University (Durham, NC, USA) collected whole blood samples from healthy donors and red blood cells (RBCs) were isolated and purified. RBCs were infected with a Plasmodium falciparum, and synchronized. During the 48-hour life cycle, infected RBCs were isolated from the general RBC population by magnetic sorting via a magnetic cell separation system (MACS, Miltenyi Biotec, Bergisch Gladbach, Germany) to separate uninfected RBCs from those containing parasites. The team used quantitative phase spectroscopy system (QPS) to image red blood cells. Refocused images are analyzed to extract 23 morphological descriptors based on the phase information.

Machine learning algorithms were used to distinguish uninfected RBCs from three different hemozoin containing stages of P. falciparum infected RBCs (early trophozoite–ET, late trophozoite–LT, and schizont–S). All of the classification methods have higher specificities compared to their sensitivities when distinguishing uninfected from infected RBCs for all three stages of infection. The specificities ranged from 98.4% for LT with the early stage of infection (ET) to 100% for the best performing method (LDC) for both LT and S stages.

The authors concluded that one of the main strengths of using machine learning algorithms to analyze the extracted parameters is that the identification of RBC infection will be based on quantified metrics and pre-built classifiers that requires minimal operator training. In order to enable automated imaging in the future, a microfluidic device with controlled flow rates can be combined with the analysis approach that would allow high throughput.

Adam Wax, PhD, a professor of biomedical engineering who helped pioneer the technology, said, “With this technique, the path is there to be able to process thousands of cells per minute. That’s a huge improvement to the 40 minutes it currently takes a field technician to stain, prepare and read a slide to personally look for infection,” The study was published on September 16, 2016, in the journal Public Library of Science ONE.

Related Links:
Duke University
Miltenyi Biotec