Multispectral imaging of human blood media applied to malaria diagnostics

Abstract

Malaria is a serious disease whose confirmation in laboratories, usually by optical microscopy, remains a challenge. The process involved in preparation of blood films and their examination in the conventional microscopes is time consuming, and the results may vary depending on the expertise of the examining technician, putting the lives of patients at risk. Other laboratory techniques for confirming malarial infection are expensive and may involve procedures that require extra training of personnel. In this thesis, a method for rapid detection of Plasmodia (malaria parasites) in unstained thin blood smears has been developed. The method is based on microscopically imaging red blood cells using different wavelengths of light in the UV, visible and NIR region for illumination (an emerging field known as multispectral imaging microscopy). Imaging was accomplished by use of an optical microscope modified by replacing its tungsten light source with a set of light emitting diodes (LEDs) whose emission spectra were centered at 375 nm, 400 nm, 435 nm, 470 nm, 525 nm, 590 nm, 625 nm, 660 nm, 700 nm, 750 nm, 810 nm, 850 nm and 940 nm. These LEDs were made to illuminate the samples from different orientations to obtain transmittance, reflectance, and dark-field images to reveal different optical properties of the imaged sample. The microscope was fitted with a monochrome CMOS camera that recorded the intensity of light emanating from the specimen as gray-level images. Using in vitro cultures of red blood cells infected with Plasmodium falciparum prepared as thin smears and without any stain, congruent intensity images were recorded at different wavele infected and non-infected red-blood cells. Results obtained using Principal Components Analysis (PCA) and Hierarchical Cluster Analysis (HCA) show that detection and identification of malaria parasites in multispectral images is possible in the 375-940 nm range and relies on presence of hemozoin (a pigment generated by the parasites when they digest haemoglobin). Artificial Neural Network (ANN) model shows promising results in accurately discriminating infected red blood cells from non-infected ones and in determining parasite burden (parasitemia) in unstained human blood. The developed method is rapid, with the whole diagnostic process taking approximately 10 minutes as opposed to the conventional microscopy which takes about 30 minutes.