Visible Spectrum-based Classification of Malaria Blood Samples on Handheld Spectrometer

Malaria diagnosis using computational approach offers a potential solution to the shortcomings of labour intensive, manual microscopic examination. In this paper, an alternative diagnostic method is introduced for malaria diagnosis directly from blood serum. A light spectrometer prototype is developed which can capture electromagnetic spectrum data (340-850 nm) from blood serum. Unlike manual microscopy where sample preparation or blood smear is a methodical process, the proposed device works on the serum which is quite easy to extract. The proposed device is a low-cost design which can be employed in the field for mass screening. The essential electronic components like light source, micro-spectrometer sensor, microcontroller board, etc. are fabricated into a small 100$\times$85$\times$80 mm3 handheld model. A dedicated stand-alone software and smartphone app are developed to communicate, data acquisition, plot and classify spectral data. For classification, Artificial Neural Network (ANN) and Support Vector Machine (SVM) classifiers are adopted. Both classifiers perform satisfactorily with an accuracy of 100% to identify malaria-infected blood samples. The proposed spectrometer will be a logical alternative for first-aid malaria diagnosis.