Tile-based microscopic image processing for malaria screening using a deep learning approach.

BACKGROUND: Manual microscopic examination remains the golden standard for malaria diagnosis. But it is laborious, and pathologists with experience are needed for accurate diagnosis. The need for computer-aided diagnosis methods is driven by the enormous workload and difficulties associated with manual microscopy based examination. While the importance of computer-aided diagnosis is increasing at an enormous pace, fostered by the advancement of deep learning algorithms, there are still challenges in detecting small objects such as malaria parasites in microscopic images of blood films. The state-of-the-art (SOTA) deep learning-based object detection models are inefficient in detecting small objects accurately because they are underrepresented on benchmark datasets. The performance of these models is affected by the loss of detailed spatial information due to in-network feature map downscaling. This is due to the fact that the SOTA models cannot directly process high-resolution images due to their low-resolution network input layer. METHODS: In this study, an efficient and robust tile-based image processing method is proposed to enhance the performance of malaria parasites detection SOTA models. Three variants of YOLOV4-based object detectors are adopted considering their detection accuracy and speed. These models were trained using tiles generated from 1780 high-resolution P. falciparum-infected thick smear microscopic images. The tiling of high-resolution images improves the performance of the object detection models. The detection accuracy and the generalization capability of these models have been evaluated using three datasets acquired from different regions. RESULTS: The best-performing model using the proposed tile-based approach outperforms the baseline method significantly (Recall, [95.3%] vs [57%] and Average Precision, [87.1%] vs [76%]). Furthermore, the proposed method has outperformed the existing approaches that used different machine learning techniques evaluated on similar datasets. CONCLUSIONS: The experimental results show that the proposed method significantly improves P. falciparum detection from thick smear microscopic images while maintaining real-time detection speed. Furthermore, the proposed method has the potential to assist and reduce the workload of laboratory technicians in malaria-endemic remote areas of developing countries where there is a critical skill gap and a shortage of experts.

Positron emission tomography and magnetic resonance imaging of the brain in experimental human malaria, a prospective cohort study.

Cerebral malaria is the most serious manifestation of severe falciparum malaria. Sequestration of infected red blood cells and microvascular dysfunction are key contributing processes. Whether these processes occur in early stage disease prior to clinical manifestations is unknown. To help localize and understand these processes during the early stages of infection, we performed 18-F fluorodeoxyglucose positron emission tomography/magnetic resonance imaging in volunteers with Plasmodium falciparum induced blood stage malaria (IBSM) infection, and compared results to individuals with P. vivax infection, in whom coma is rare. Seven healthy, malaria-naïve participants underwent imaging at baseline, and at early symptom onset a median 9 days following inoculation (n = 4 P. falciparum, n = 3 P. vivax). Participants with P. falciparum infection demonstrated marked lability in radiotracer uptake across all regions of the brain, exceeding expected normal variation (within subject coefficient of variation (wCV): 14.4%) compared to the relatively stable uptake in participants with P. vivax infection (wCV: 3.5%). No consistent imaging changes suggestive of microvascular dysfunction were observed in either group. Neuroimaging in early IBSM studies is safe and technically feasible, with preliminary results suggesting that differences in brain tropism between P. falciparum and P. vivax may occur very early in infection.

Label-free imaging and classification of live P. falciparum enables high performance parasitemia quantification without fixation or staining.

Manual microscopic inspection of fixed and stained blood smears has remained the gold standard for Plasmodium parasitemia analysis for over a century. Unfortunately, smear preparation consumes time and reagents, while manual microscopy is skill-dependent and labor-intensive. Here, we demonstrate that deep learning enables both life stage classification and accurate parasitemia quantification of ordinary brightfield microscopy images of live, unstained red blood cells. We tested our method using both a standard light microscope equipped with visible and near-ultraviolet (UV) illumination, and a custom-built microscope employing deep-UV illumination. While using deep-UV light achieved an overall four-category classification of Plasmodium falciparum blood stages of greater than 99% and a recall of 89.8% for ring-stage parasites, imaging with near-UV light on a standard microscope resulted in 96.8% overall accuracy and over 90% recall for ring-stage parasites. Both imaging systems were tested extrinsically by parasitemia titration, revealing superior performance over manually-scored Giemsa-stained smears, and a limit of detection below 0.1%. Our results establish that label-free parasitemia analysis of live cells is possible in a biomedical laboratory setting without the need for complex optical instrumentation. We anticipate future extensions of this work could enable label-free clinical diagnostic measurements, one day eliminating the need for conventional blood smear analysis.

Adaptive adversarial neural networks for the analysis of lossy and domain-shifted datasets of medical images.

In machine learning for image-based medical diagnostics, supervised convolutional neural networks are typically trained with large and expertly annotated datasets obtained using high-resolution imaging systems. Moreover, the network's performance can degrade substantially when applied to a dataset with a different distribution. Here, we show that adversarial learning can be used to develop high-performing networks trained on unannotated medical images of varying image quality. Specifically, we used low-quality images acquired using inexpensive portable optical systems to train networks for the evaluation of human embryos, the quantification of human sperm morphology and the diagnosis of malarial infections in the blood, and show that the networks performed well across different data distributions. We also show that adversarial learning can be used with unlabelled data from unseen domain-shifted datasets to adapt pretrained supervised networks to new distributions, even when data from the original distribution are not available. Adaptive adversarial networks may expand the use of validated neural-network models for the evaluation of data collected from multiple imaging systems of varying quality without compromising the knowledge stored in the network.

Brain Magnetic Resonance Imaging Reveals Different Courses of Disease in Pediatric and Adult Cerebral Malaria.

BACKGROUND: Cerebral malaria is a common presentation of severe Plasmodium falciparum infection and remains an important cause of death in the tropics. Key aspects of its pathogenesis are still incompletely understood, but severe brain swelling identified by magnetic resonance imaging (MRI) was associated with a fatal outcome in African children. In contrast, neuroimaging investigations failed to identify cerebral features associated with fatality in Asian adults. METHODS: Quantitative MRI with brain volume assessment and apparent diffusion coefficient (ADC) histogram analyses were performed for the first time in 65 patients with cerebral malaria to compare disease signatures between children and adults from the same cohort, as well as between fatal and nonfatal cases. RESULTS: We found an age-dependent decrease in brain swelling during acute cerebral malaria, and brain volumes did not differ between fatal and nonfatal cases across both age groups. In nonfatal disease, reversible, hypoxia-induced cytotoxic edema occurred predominantly in the white matter in children, and in the basal ganglia in adults. In fatal cases, quantitative ADC histogram analyses also demonstrated different end-stage patterns between adults and children: Severe hypoxia, evidenced by global ADC decrease and elevated plasma levels of lipocalin-2 and microRNA-150, was associated with a fatal outcome in adults. In fatal pediatric disease, our results corroborate an increase in brain volume, leading to augmented cerebral pressure, brainstem herniation, and death. CONCLUSIONS: Our findings suggest distinct pathogenic patterns in pediatric and adult cerebral malaria with a stronger cytotoxic component in adults, supporting the development of age-specific adjunct therapies.

Malaria Screener: a smartphone application for automated malaria screening.

BACKGROUND: Light microscopy is often used for malaria diagnosis in the field. However, it is time-consuming and quality of the results depends heavily on the skill of microscopists. Automating malaria light microscopy is a promising solution, but it still remains a challenge and an active area of research. Current tools are often expensive and involve sophisticated hardware components, which makes it hard to deploy them in resource-limited areas. RESULTS: We designed an Android mobile application called Malaria Screener, which makes smartphones an affordable yet effective solution for automated malaria light microscopy. The mobile app utilizes high-resolution cameras and computing power of modern smartphones to screen both thin and thick blood smear images for P. falciparum parasites. Malaria Screener combines image acquisition, smear image analysis, and result visualization in its slide screening process, and is equipped with a database to provide easy access to the acquired data. CONCLUSION: Malaria Screener makes the screening process faster, more consistent, and less dependent on human expertise. The app is modular, allowing other research groups to integrate their methods and models for image processing and machine learning, while acquiring and analyzing their data.

Real-time cholesterol sorting in Plasmodium falciparum-erythrocytes as revealed by 3D label-free imaging.

Cholesterol, a necessary component of animal cell membranes, is also needed by the lethal human malaria parasite Plasmodium falciparum. Because P. falciparum lacks a cholesterol synthesis pathway and malaria patients have low blood cholesterol, we speculated that it scavenges cholesterol from them in some way. We used time-lapse holotomographic microscopy to observe cholesterol transport in live P. falciparum parasites and structurally investigate erythrocyte membranes, both during and after P. falciparum invasion of human erythrocytes. After P. falciparum initially acquired free cholesterol or inner erythrocytic membrane-derived cholesterol, we observed budding lipid membranes elongating into the cytosol and/or membrane segments migrating there and eventually fusing with the parasite membranes, presumably at the parasitophorous vacuole membrane (PVM). Finally, the cholesterol-containing segments were seen to surround the parasite nucleus. Our imaging data suggest that a novel membrane transport system operates in the cytosol of P. falciparum-infected erythrocytes as a cholesterol import system, likely between the PVM and the erythrocyte membrane, and that this transportation process occurs during the live erythrocyte stages of P. falciparum.

Molecular Imaging of a Zirconium-89 Labeled Antibody Targeting Plasmodium falciparum-Infected Human Erythrocytes.

PURPOSE: Nuclear imaging is an important preclinical research tool to study infectious diseases in vivo and could be extended to investigate complex aspects of malaria infections. As such, we report for the first time successful radiolabeling of a novel antibody specific to Plasmodium-infected erythrocytes (IIIB6), its in vitro assessment and molecular imaging in nude mice. PROCEDURES: In vitro confocal microscopy was used to determine the stage-specificity of Plasmodium-infected erythrocytes recognised by IIIB6. To enable micro-positron emission tomography (PET)/X-ray computed tomography (CT) imaging, IIIB6 was conjugated to Bz-DFO-NCS and subsequently radiolabeled with zirconium-89. Healthy nude mice were injected with [89Zr]IIIB6, and pharmacokinetics and organ uptake were monitored over 24 h. This was followed by post-mortem animal dissection to determine the biodistribution of [89Zr]IIIB6. RESULTS: IIIB6 recognised all the relevant stages of Plasmodium falciparum-infected erythrocytes (trophozoites, schizonts and gametocytes) that are responsible for severe malaria pathology. [89Zr]IIIB6-radiolabeling yields were efficient at 84-89 %. Blood pool imaging analysis indicated a pharmacological half-life of 9.6 ± 2.5 h for [89Zr]IIIB6. The highest standard uptake values were determined at 2-6 h in the liver followed by the spleen, kidneys, heart, stomach and lung, respectively. Minimal activity was present in muscle and bone tissues. CONCLUSION: In vitro characterization of IIIB6 and pharmacokinetic characterization of [89Zr]IIIB6 revealed that this antibody has potential for future use in Plasmodium-infected mouse models to study malaria in a preclinical in vivo setting with PET/CT imaging.

Severe malaria.

Malaria represents a medical emergency. Without rapid diagnosis and treatment, it can progress and lead to severe complications and, eventually, death. Severe malaria is almost always caused by Plasmodium falciparum. Here, we present an image showing a set of hematological findings associated with severe malaria, highlighting the importance of a correct morphological diagnosis.

Reduced Cardiac Index Reserve and Hypovolemia in Severe Falciparum Malaria.

BACKGROUND: Impaired microvascular perfusion is central to the development of coma and lactic acidosis in severe falciparum malaria. Refractory hypotension is rare on admission but develops frequently in fatal cases. We assessed cardiac function and volume status in severe falciparum malaria and its prognostic significance. METHODS: Patients with severe (N = 101) or acute uncomplicated falciparum malaria (N = 83) were recruited from 2 hospitals in India and Bangladesh, and healthy participants (N = 44) underwent echocardiography. RESULTS: Patients with severe malaria had 38% shorter left ventricular (LV) filling times and 25% shorter LV ejection times than healthy participants because of tachycardia; however, stroke volume, LV internal diameter in diastole (LVIDd), and LV internal diameter in systole (LVIDs) indices were similar. A low endocardial fraction shortening (eFS) was present in 17% (9 of 52) of severe malaria patients. Adjusting for preload and afterload, eFS was similar in health and severe malaria. Fatal cases had smaller baseline LVIDd and LVIDs indices, more collapsible inferior vena cavae (IVC), and higher heart rates than survivors. The LVIDs and IVC collapsibility were independent predictors for mortality, together with base excess and Glasgow Coma Scale. CONCLUSIONS: Patients with severe malaria have rapid ejection of a normal stroke volume. Fatal cases had features of relative hypovolemia and reduced cardiac index reserve.

In Vivo Imaging of the Buccal Mucosa Shows Loss of the Endothelial Glycocalyx and Perivascular Hemorrhages in Pediatric Plasmodium falciparum Malaria.

Severe malaria is mostly caused by Plasmodium falciparum, resulting in considerable, systemic inflammation and pronounced endothelial activation. The endothelium forms an interface between blood and tissue, and vasculopathy has previously been linked with malaria severity. We studied the extent to which the endothelial glycocalyx that normally maintains endothelial function is involved in falciparum malaria pathogenesis by using incident dark-field imaging in the buccal mucosa. This enabled calculation of the perfused boundary region, which indicates to what extent erythrocytes can permeate the endothelial glycocalyx. The perfused boundary region was significantly increased in severe malaria patients and mirrored by an increase of soluble glycocalyx components in plasma. This is suggestive of a substantial endothelial glycocalyx loss. Patients with severe malaria had significantly higher plasma levels of sulfated glycosaminoglycans than patients with uncomplicated malaria, whereas other measured glycocalyx markers were raised to a comparable extent in both groups. In severe malaria, the plasma level of the glycosaminoglycan hyaluronic acid was positively correlated with the perfused boundary region in the buccal cavity. Plasma hyaluronic acid and heparan sulfate were particularly high in severe malaria patients with a low Blantyre coma score, suggesting involvement in its pathogenesis. In vivo imaging also detected perivascular hemorrhages and sequestering late-stage parasites. In line with this, plasma angiopoietin-1 was decreased while angiopoietin-2 was increased, suggesting vascular instability. The density of hemorrhages correlated negatively with plasma levels of angiopoietin-1. Our findings indicate that as with experimental malaria, the loss of endothelial glycocalyx is associated with vascular dysfunction in human malaria and is related to severity.

Portable bright-field, fluorescence, and cross-polarized microscope toward point-of-care imaging diagnostics.

Emerging technologies are enabling the feasibility of new types of point-of-care diagnostic devices. A portable, multimodal microscopy platform intended for use in remote diagnostic applications is presented. Use of such a system could bring high-quality microscopy to field use for diseases such as malaria, allowing better diagnostic and surveillance information to be gathered. The microscope was designed using off-the-shelf components and a manual filter selection to generate bright-field, fluorescent, and cross-polarized images of samples mounted to microscopy slides. Design parameters for the system are discussed, and characterization is performed using standardized imaging targets, multimodal phantoms, and blood smears simulating those used in malaria diagnosis. The microscope is shown to be able to image below element 9-3 of a 1951 U.S. Air Force target, indicating that the system is capable of resolving features < 775 nm. Morphological indicators of Plasmodium falciparum can be visualized in images from each modality and combined into high-contrast composite images. To optimize parasitic feature contrast across all three imaging modes, several different staining techniques were compared, with results indicating that use of a single nucleic acid binding fluorophore is preferable.

The Diversity of the Plasmodium falciparum K13 Propeller Domain Did Not Increase after Implementation of Artemisinin-Based Combination Therapy in Uganda.

Artemisinin-based combination therapies (ACTs) are the standard of care to treat uncomplicated falciparum malaria. However, resistance to artemisinins, defined as delayed parasite clearance after therapy, has emerged in Southeast Asia, and the spread of resistance to sub-Saharan Africa could have devastating consequences. Artemisinin resistance has been associated in Southeast Asia with multiple nonsynonymous single nucleotide polymorphisms (NS-SNPs) in the propeller domain of the gene encoding the Plasmodium falciparum K13 protein (K13PD). Some K13PD NS-SNPs have been seen in Africa, but the relevance of these mutations is unclear. To assess whether ACT use has selected for specific K13PD mutations, we compared the K13PD genetic diversity in clinical isolates collected before and after the implementation of ACT use from seven sites across Uganda. We detected K13PD NS-SNPs in 16 of 683 (2.3%) clinical isolates collected between 1999 and 2004 and in 26 of 716 (3.6%) isolates collected between 2012 and 2016 (P = 0.16), representing a total of 29 different polymorphisms at 27 codons. Individual NS-SNPs were usually detected only once, and none were found in more than 0.7% of the isolates. Three SNPs (C469F, P574L, and A675V) associated with delayed clearance in Southeast Asia were seen in samples collected between 2012 and 2016, each in a single isolate. No differences in diversity following implementation of ACT use were found at any of the seven sites, nor was there evidence of selective pressures acting on the locus. Our results suggest that selection by ACTs is not impacting on K13PD diversity in Uganda.

MRI demonstrates glutamine antagonist-mediated reversal of cerebral malaria pathology in mice.

The deadliest complication of Plasmodium falciparum infection is cerebral malaria (CM), with a case fatality rate of 15 to 25% in African children despite effective antimalarial chemotherapy. No adjunctive treatments are yet available for this devastating disease. We previously reported that the glutamine antagonist 6-diazo-5-oxo-l-norleucine (DON) rescued mice from experimental CM (ECM) when administered late in the infection, a time by which mice had already suffered blood-brain barrier (BBB) dysfunction, brain swelling, and hemorrhaging. Herein, we used longitudinal MR imaging to visualize brain pathology in ECM and the impact of a new DON prodrug, JHU-083, on disease progression in mice. We demonstrate in vivo the reversal of disease markers in symptomatic, infected mice following treatment, including the resolution of edema and BBB disruption, findings usually associated with a fatal outcome in children and adults with CM. Our results support the premise that JHU-083 is a potential adjunctive treatment that could rescue children and adults from fatal CM.

Convolutional neural network-based malaria diagnosis from focus stack of blood smear images acquired using custom-built slide scanner.

The present paper introduces a focus stacking-based approach for automated quantitative detection of Plasmodium falciparum malaria from blood smear. For the detection, a custom designed convolutional neural network (CNN) operating on focus stack of images is used. The cell counting problem is addressed as the segmentation problem and we propose a 2-level segmentation strategy. Use of CNN operating on focus stack for the detection of malaria is first of its kind, and it not only improved the detection accuracy (both in terms of sensitivity [97.06%] and specificity [98.50%]) but also favored the processing on cell patches and avoided the need for hand-engineered features. The slide images are acquired with a custom-built portable slide scanner made from low-cost, off-the-shelf components and is suitable for point-of-care diagnostics. The proposed approach of employing sophisticated algorithmic processing together with inexpensive instrumentation can potentially benefit clinicians to enable malaria diagnosis.

Improving vector-borne pathogen surveillance: A laboratory-based study exploring the potential to detect dengue virus and malaria parasites in mosquito saliva.

BACKGROUND & OBJECTIVES: Vector-borne pathogen surveillance programmes typically rely on the collection of large numbers of potential vectors followed by screening protocols focused on detecting pathogens in the arthropods. These processes are laborious, time consuming, expensive, and require screening of large numbers of samples. To streamline the surveillance process, increase sample throughput, and improve cost-effectiveness, a method to detect dengue virus and malaria parasites (Plasmodium falciparum) by leveraging the sugar-feeding behaviour of mosquitoes and their habit of expectorating infectious agents in their saliva during feeding was investigated in this study. METHODS: Dengue virus 2 (DENV-2) infected female Aedes aegypti mosquitoes and P. falciparum infected female Anopheles stephensi mosquitoes were allowed to feed on honey coated Flinders Technical Associates -FTA® cards dyed with blue food colouring. The feeding resulted in deposition of saliva containing either DENV-2 particles or P. falciparum sporozoites onto the FTA card. Nucleic acid was extracted from each card and the appropriate real-time PCR (qPCR) assay was run to detect the pathogen of interest. RESULTS: As little as one plaque forming unit (PFU) of DENV-2 and as few as 60 P. falciparum parasites deposited on FTA cards from infected mosquitoes were detected via qPCR. Hence, their use to collect mosquito saliva for pathogen detection is a relevant technique for vector surveillance. INTERPRETATION & CONCLUSION: This study provides laboratory confirmation that FTA cards can be used to capture and stabilize expectorated DENV-2 particles and P. falciparum sporozoites from infectious, sugar-feeding mosquitoes in very low numbers. Thus, the FTA card-based mosquito saliva capture method offers promise to overcome current limitations and revolutionize traditional mosquito-based pathogen surveillance programmes. Field testing and further method development are required to optimize this strategy.

Integrated quantitative phase and birefringence microscopy for imaging malaria-infected red blood cells.

A dual-modality birefringence/phase imaging system is presented. The system features a crystal retarder that provides polarization mixing and generates two interferometric carrier waves in a single signal spectrum. The retardation and orientation of sample birefringence can then be measured simultaneously based on spectral multiplexing interferometry. Further, with the addition of a Nomarski prism, the same setup can be used for quantitative differential interference contrast (DIC) imaging. Sample phase can then be obtained with two-dimensional integration. In addition, birefringence-induced phase error can be corrected using the birefringence data. This dual-modality approach is analyzed theoretically with Jones calculus and validated experimentally with malaria-infected red blood cells. The system generates not only corrected DIC and phase images, but a birefringence map that highlights the distribution of hemozoin crystals.

ARAM: an automated image analysis software to determine rosetting parameters and parasitaemia in Plasmodium samples.

BACKGROUND: Rosetting is associated with severe malaria and a primary cause of death in Plasmodium falciparum infections. Detailed understanding of this adhesive phenomenon may enable the development of new therapies interfering with rosette formation. For this, it is crucial to determine parameters such as rosetting and parasitaemia of laboratory strains or patient isolates, a bottleneck in malaria research due to the time consuming and error prone manual analysis of specimens. Here, the automated, free, stand-alone analysis software automated rosetting analyzer for micrographs (ARAM) to determine rosetting rate, rosette size distribution as well as parasitaemia with a convenient graphical user interface is presented. METHODS: Automated rosetting analyzer for micrographs is an executable with two operation modes for automated identification of objects on images. The default mode detects red blood cells and fluorescently labelled parasitized red blood cells by combining an intensity-gradient with a threshold filter. The second mode determines object location and size distribution from a single contrast method. The obtained results are compared with standardized manual analysis. Automated rosetting analyzer for micrographs calculates statistical confidence probabilities for rosetting rate and parasitaemia. RESULTS: Automated rosetting analyzer for micrographs analyses 25 cell objects per second reliably delivering identical results compared to manual analysis. For the first time rosette size distribution is determined in a precise and quantitative manner employing ARAM in combination with established inhibition tests. Additionally ARAM measures the essential observables parasitaemia, rosetting rate and size as well as location of all detected objects and provides confidence intervals for the determined observables. No other existing software solution offers this range of function. The second, non-malaria specific, analysis mode of ARAM offers the functionality to detect arbitrary objects. CONCLUSIONS: Automated rosetting analyzer for micrographs has the capability to push malaria research to a more quantitative and statistically significant level with increased reliability due to operator independence. As an installation file for Windows © 7, 8.1 and 10 is available for free, ARAM offers a novel open and easy-to-use platform for the malaria community to elucidate resetting.

Acute hemorrhagic necrotizing pancreatitis in falciparum malaria.

Malaria is a pathology caused by a parasite called Plasmodium, characteristic of tropical countries. The most frequent symptomatology includes cerebral malaria, jaundice, convulsive crisis, anemia, hypoglycemia, kidney failure and metabolic asidosis, among others. We are presenting the case of a patient diagnosed with malaria who suffered from acute necrotizing hemorrhagic pancreatitis and evolved poorly, as an example of this combination of symptoms, rarely found in our country.

Transient lesion in the splenium of the corpus callosum in acute uncomplicated falciparum malaria.

Patients with acute uncomplicated Plasmodium falciparum malaria have no evident neurologic disorder, vital organ dysfunction, or other severe manifestations of infection. Nonetheless, parasitized erythrocytes cytoadhere to the endothelium throughout their microvasculature, especially within the brain. We aimed to determine if 3 Tesla magnetic resonance imaging studies could detect evidence of cerebral abnormalities in these patients. Within 24 hours of admission, initial magnetic resonance imaging examinations found a lesion with restricted water diffusion in the mid-portion of the splenium of the corpus callosum of 4 (40%) of 10 male patients. The four patients who had a splenial lesion initially had evidence of more severe hemolysis and thrombocytopenia than the six patients who had no apparent abnormality. Repeat studies four weeks later found no residua of the lesions and resolution of the hematologic differences. These observations provide evidence for acute cerebral injury in the absence of severe or cerebral malaria.