Computer Vision Malaria Diagnostic Systems-Progress and Prospects.

Accurate malaria diagnosis is critical to prevent malaria fatalities, curb overuse of antimalarial drugs, and promote appropriate management of other causes of fever. While several diagnostic tests exist, the need for a rapid and highly accurate malaria assay remains. Microscopy and rapid diagnostic tests are the main diagnostic modalities available, yet they can demonstrate poor performance and accuracy. Automated microscopy platforms have the potential to significantly improve and standardize malaria diagnosis. Based on image recognition and machine learning algorithms, these systems maintain the benefits of light microscopy and provide improvements such as quicker scanning time, greater scanning area, and increased consistency brought by automation. While these applications have been in development for over a decade, recently several commercial platforms have emerged. In this review, we discuss the most advanced computer vision malaria diagnostic technologies and investigate several of their features which are central to field use. Additionally, we discuss the technological and policy barriers to implementing these technologies in low-resource settings world-wide.

Evaluation of the Parasight Platform for Malaria Diagnosis.

The World Health Organization estimates that nearly 500 million malaria tests are performed annually. While microscopy and rapid diagnostic tests (RDTs) are the main diagnostic approaches, no single method is inexpensive, rapid, and highly accurate. Two recent studies from our group have demonstrated a prototype computer vision platform that meets those needs. Here we present the results from two clinical studies on the commercially available version of this technology, the Sight Diagnostics Parasight platform, which provides malaria diagnosis, species identification, and parasite quantification. We conducted a multisite trial in Chennai, India (Apollo Hospital [n = 205]), and Nairobi, Kenya (Aga Khan University Hospital [n = 263]), in which we compared the device to microscopy, RDTs, and PCR. For identification of malaria, the device performed similarly well in both contexts (sensitivity of 99% and specificity of 100% at the Indian site and sensitivity of 99.3% and specificity of 98.9% at the Kenyan site, compared to PCR). For species identification, the device correctly identified 100% of samples with Plasmodium vivax and 100% of samples with Plasmodium falciparum in India and 100% of samples with P. vivax and 96.1% of samples with P. falciparum in Kenya, compared to PCR. Lastly, comparisons of the device parasite counts with those of trained microscopists produced average Pearson correlation coefficients of 0.84 at the Indian site and 0.85 at the Kenyan site.

Computer-vision-based technology for fast, accurate and cost effective diagnosis of malaria.

BACKGROUND: Microscopy has long been considered to be the gold standard for diagnosis of malaria despite the introduction of newer assays. However, it has many challenges like requirement of trained microscopists and logistic issues. A vision based device that can diagnose malaria, provide speciation and estimate parasitaemia was evaluated. METHODS: The device was evaluated using samples from 431 consented patients, 361 of which were initially screened by RDT and microscopy and later analysed by PCR. It was a prospective, non-randomized, blinded trial. Quantification of parasitaemia was performed by two experienced technicians. Samples were subjected to diagnosis by Sight Dx digital imaging scanning. RESULTS: The sensitivity and specificity of the SightDx P1 device for analysed samples were found to be 97.05 and 96.33%, respectively, when compared to PCR. When compared to microscopy, sensitivity and specificity were found to be 94.4 and 95.6%, respectively. The device was able to speciate 73.3% of the PCR Plasmodium falciparum positive samples and 91.4% of PCR Plasmodium vivax positive samples. CONCLUSION: The ability of the device to detect parasitaemia as compared with microscopy, was within 50% in 71.3% of cases and demonstrated a correlation coefficient of 0.89.