Malaria is one of the most lethal infectious diseases in tropical regions. Among the five protozoan species of Plasmodium that infect humans, P. falciparum parasites cause the highest and most severe disease burden, accounting for 90% of total deaths¹. The current 'gold standard' for malaria diagnosis is the microscopic examination of Giemsa-stained blood smears². Depending on experience and the availability of fresh reagents, a microscopist may be able to identify as low as 0.001% parasitemia from a thick-film blood smear^{3, 4}. Thin blood smears provide much clearer images but suffer from reduced sensitivity^{5, 6}, whereas PCR-based diagnostic approaches are often not practical for on-site application⁵. In addition, both PCR and rapid diagnostic tests (lateral-flow immunoassay of parasite-specific proteins) are unable to provide quantitative analysis on the parasitemia level^{3, 5} (Supplementary Table 1).

Here we report a benchtop MRR system that is capable of rapid detection and quantification of the parasite load during the early stage of infection (Fig. 1). Traditional equipment for nuclear magnetic resonance (NMR) spectroscopy is expensive and bulky and thus difficult to adapt for on-site diagnostic tools. Recently, however, Weissleder and co-workers scaled down the typical NMR system for point-of-care medical diagnostics^{7, 8}. Based on this work, we previously developed a low-cost, portable, field-programmable gate array–based system and have demonstrated that one can indirectly deduce the relative bulk magnetic susceptibility of blood cells within a few minutes of signal acquisition time^{9, 10, 11}, without expensive and extensive immunolabeling procedures.

In this work, we exploit the presence of the hemozoin crystallites formed within the erythrocyte as early as in the ring stage^{12, 13} as a natural magnetic label for MRR detection of Plasmodium spp.–infected red blood cells (iRBCs) (Fig. 1c). During the intraerythrocytic cycle, malaria parasites consume hemoglobin and release free heme. Free heme is immediately converted into an insoluble crystallite known as hemozoin. The formation of hemozoin represents a simple transformation in the redox state of the RBCs, from ferrous Fe²⁺ (diamagnetic state) into ferric Fe³⁺ (paramagnetic state), which causes a major change in bulk magnetic susceptibility of the iRBCs^{14, 15} and induces measurable changes in the magnetic resonance relaxation of nearby protons^{16, 17}.

With an inexpensive portable benchtop system and minimal sample preparation, we demonstrate that the system is able to detect P. falciparum infection to as low as 0.0002% parasitemia (P < 0.0003) using 750 nl of cultured blood, in less than 5 min. We further demonstrate using an in vivo P. berghei mouse infection model a strong correlation (R² > 0.98) between the transverse relaxation rate (R₂ index) measured by MRR and blood smear microscopy results, allowing reliable estimation of parasitemia to as low as 0.0001% (<10 parasites μl⁻¹, P < 0.05). In order to compensate for field inhomogeneity within the sample, we use standard Carr-Purcell-Meiboom-Gill (CPMG) spin-echo with high radiofrequency nutation field at ultrashort echo technique (UET)¹⁸ to determine the R₂ index (Fig. 1a). The MRR technique outperformed blood smear technique in a blinded test conducted especially for the early stage of infection.

MRR detection and quantification of cultured P. falciparum 3D7 iRBCs

We synchronized P. falciparum 3D7 parasites from cultured blood to obtain highly enriched ring-stage parasites free of any residual hemozoin by using sorbitol treatment¹⁹ and magnetic separation²⁰ (Online Methods and Supplementary Fig. 1). We prepared and measured three independent samples (A, B and C) with different purities of ring-stage iRBCs (Fig. 2a). The normalized ΔR₂ (R_2(iRBC) – R_2(RBC)) indicated a strong correlation (R² > 0.95) for all three samples (Supplementary Fig. 2). R_2(RBC) represents the baseline correction of the uninfected RBCs.

Sample A recorded as low as 0.00025% parasitemia (12.5 iRBCs μl⁻¹), which corresponds fewer than ten iRBCs inside our probe (~300 nl active volume within the coil's detection region). Samples for parasitemia level below 0.00025% (P < 0.001) were indistinguishable from uninfected RBCs. Sample B and Sample C recorded a limit of detection of 0.0002% (P < 0.0003) and 0.0007% (P < 0.05), respectively. The limit of detection of the current MRR setup for P. falciparum infection is, therefore, on the order of 0.0001%. Samples A, B and C reflected the parasite stages observed in the peripheral blood of infected subjects in actual clinical settings²¹.

P. falciparum 3D7 detection and measurement in continuous culture

We tracked the daily growth of P. falciparum 3D7 parasites in continuous culture and evaluated the R₂ index during the first few cycles. We inoculated healthy donor blood with highly synchronized ring-stage parasites (94% ring, 6% early trophozoite, 0% schizont) to obtain an initial parasitemia of 0.5 iRBCs μl⁻¹. We took ten samples per time point every 24 h (Fig. 2b). The R₂ index displayed a steady increase over time reflecting the increase in parasite levels in the culture. Furthermore, in every 24-h interval, the late-stage trophozoites had a higher R₂ index as compared to the ring-stage counterpart for the same cycle (Fig. 2c).

48 h (day 2) after the initial inoculation in the first ring cycle, the R₂ readings showed a pronounced shift (P < 0.0001) from the control readings. Theoretically, with an expected five- to tenfold increase per cycle¹⁹, by day 2 the parasitemia levels would have grown up to about 3 iRBCs μl⁻¹ (0.00006%). From ten samples, the mean reading for day 2 was 8.06 ± 0.30 s⁻¹ as compared to the mean control readings of 7.20 ± 0.15 s⁻¹, which is the healthy uninfected baseline of RBCs. In fact, the receiver operating characteristic (ROC) on day 1 was already indicative of a good diagnostic performance (area under the curve (AUC) > 0.80) and an excellent diagnostic performance beyond day 2 (Fig. 2d and Supplementary Fig. 3). An AUC value of 0.80–1.00 is considered good and excellent accuracy, 0.70–0.80 is fair, and below 0.60 is poor (Supplementary Fig. 3). The AUC values from day 1 to day 8 were 0.80, 1.00, 0.96, 1.00, 1.00, 1.00, 1.00 and 1.00.

In vivo detection of P. berghei infection in BALB/c mice

We used the murine malaria parasite P. berghei, which is a good model of human P. falciparum infection, for the in vivo validation of this approach²². We intraperitoneally infected BALB/c mice (n = 25) with ~1 × 10⁴ P. berghei parasites (denoted as day 0) (Online Methods and Supplementary Fig. 4a). We allowed the infection to establish itself for 1 d. From day 2 to day 13 post-infection, we analyzed tail blood from the mice each day both by MRR measurements and by thin-smear microscopy (Fig. 3a). P. berghei infections in laboratory rodents tend to be asynchronous. Therefore, for P. berghei the measured R₂ index reflects the signal from a mixture of parasite stages instead of the more synchronized ring stages of P. falciparum found in human host.

In general, the R₂ index of the mice and parasitemia levels increased with the number of post-infection days, with a slight fluctuation in R₂ index in the early days of infection (Fig. 3b). Reliable counting of parasitemia using Giemsa-stained smears was, however, only possible from day 6 onwards (Supplementary Table 2) and showed that the parasitemia levels gradually increased and reached approximately 25% by the end of the experiment on day 13, consistent with the reports that BALB/c mice survive through the early phase of infection and eventually die of anemia and high parasitemia²³. Remarkably, there was a strong linear relationship (R² > 0.98) between the measured R₂ index and parasitemia, as determined by blood-smear counts (Fig. 3c). We performed ROC analysis to investigate the accuracy of differentiating infected mice from uninfected mice throughout the experiment from day 2 to day 13 (Supplementary Fig. 5 and Supplementary Table 3). We further established the limit of detection of P. berghei infection, which we found to be as low as 0.0001% parasitemia (Supplementary Figs. 6 and 7 and Supplementary Table 4).

Blinded test: early stage of P. berghei infection

In order to evaluate the performance of MRR as a diagnostics test, we carried out a blinded study focusing on the early stage (day 1 to day 6) of P. berghei infection in BALB/c mice. We randomly infected BALB/c mice (n = 58) intraperitoneally with asynchronously mixed stages of P. berghei parasites in three different sizes of inoculum (1 × 10⁷, 1 × 10⁴ and zero as control). We first asked five experienced microscopists and a well-trained MRR technician, who were blinded to whether the mice were preinfected and to the inoculum size, to independently and rapidly screen for parasites using both blood-smear technique and MRR (Online Methods and Supplementary Fig. 4d). At the end of the experiment, we cross-checked the results and classified them as true positive, true negative, false positive and false negative (Table 1).

Not unexpectedly, in the early days of the infection, blood-smear evaluation performed poorly with mean (sensitivity of 74.9%, specificity of 82.0% and accuracy of 75.5%) throughout the six days of observation (Table 1) and had great variability between microscopists (Fig. 4a). The accuracy of the microscopy technique improved after day 3 of infection when parasitemia increased above 0.01% (Supplementary Table 5). In contrast, MRR measurements showed a better diagnostic performance with mean (97.9% sensitivity, 90% specificity, and 96.9% accuracy) throughout the six days of observation (Table 1).

Finally, the blood-smear slides were once again cross-checked carefully without time constraints for detailed quantitative analysis (Fig. 4b and Supplementary Figs. 8 and 9). The reconstructed R₂ index against the post-infection days mapped similar trends of parasitemia levels as evaluated by blood-smear microscopy (Fig. 4c). The MRR measurements also provided new insights into the pathophysiology of early post-infection days in the rodent mouse model (Figs. 3 and 4). We detected a critical event at around day 4 post-infection (1 × 10⁴ inoculum), where the median R₂ index was slightly lower than the R₂ index on day 3 (Fig. 4b). We observed this result consistently in multiple experiments and also, albeit less pronounced, and on day 2 when we used a higher initial inoculum (1 × 10⁷ inoculum). The parasite multiplication rate slowed down by day 4 and day 3, for inocula of 1 × 10⁴ and 1 × 10⁷, respectively (Supplementary Table 5d,e).

The rapid, sensitive and reliable early detection of Plasmodium-infected subjects is a key component of the global effort to eliminate malaria. Here we have demonstrated that hemozoin crystallites, a metabolic byproduct of malaria parasites known to be present in all stages and all human-infecting strains of malaria¹⁵, can be used as a natural biomarker for malaria diagnostics. Our new magnetic susceptibility classification, the R₂ index, measured with MRR techniques, correlates well with iRBC load, and may be a valuable metric for clinical prognosis, monitoring of drug resistance, hospitalization criteria and assessing severity of infection.

The higher sensitivity of the MRR shown in this work can be attributed to a few factors. First, the micron-sized radiofrequency detection coil, which produces strong radiofrequency nutation fields per unit current²⁴, markedly improved the signal-to-noise ratio and volume sensitivity. Second, advances in microelectronics have enabled faster radiofrequency switching and higher sampling rates of analog-to-digital converters, which has been crucial for the ultrashort echo technique^{18, 24} in MRR spectroscopy adopted here. This technique retrieves spin-echoes at very short echo time (10–100 μs) over a long array of pulses, therefore increasing the signal-to-noise ratio of tissues and cells of very short relaxation rates or in the presence of strong inhomogeneous fields¹⁸. We have shown that concentrating the RBCs via standard centrifugation and exploiting UET have given a signal sensitivity enhancement, compared with work reported by Karl et al.²⁵ (Supplementary Fig. 10 and Supplementary Discussion).

Our system offers highly sensitive label-free detection (<10 parasites μl⁻¹), is less prone to human error, uses a minimal blood volume (a few microliters), requires minimal sample preparation, can be used in the field and has an ultralow cost per assay (less than ten cents). Given these advantages, we believe that MRR can be an alternative to traditional malaria diagnostics such as Giemsa blood-smear microscopy and dipsticks. Currently, electronics and polarizing magnets are the main cost driver of the MRR system. Yet, we¹⁰ and others^{7, 26} have shown that an entire MRR system (along with inexpensive magnets) can be packaged onto a single chip preprogrammed with radio-frequency pulse sequences, which will make it cheaper (less than $2,000), smaller and easier to use in point-of-care settings^{7, 26}.

Integrated sample preparation and detection.

For parasite detection, highly synchronized ring-stage P. falciparum and P. berghei parasites (Fig. 1d, Supplementary Figs. 1a and 4) are mixed well at room temperature via repeated pipetting before being transferred into a microcapillary tube (Drummond Scientific Co., Broomall, PA, USA) via capillary action. Repeated pipetting ensures that any deoxygenated hemoglobin (deoxy-Hb), which is in the form of paramagnetic state will be converted into oxygenated-hemoglobin (oxy-Hb), which is in the diamagnetic state, and thereby create a universal baseline for all other measurements^{16, 17}. From our experience, the residual fraction of deoxy-Hb drawn directly from subjects is easily converted to oxy-Hb upon exposure to the ambient air. It is also worth noting that the trace amount of methemoglobin (metHb) present in human blood physiologically is also in a paramagnetic state. The ferric state (Fe³⁺) in metHb, however, cannot be reduced back to its original ferrous state (Fe²⁺) through physical mixing such as pipetting. Since the current protocol is not designed to remove the trace amount of metHb in whole blood, any unusually high amount of metHb may introduce false-positive signal, or extreme case of patient-to-patient variation.

Once the blood sample is transferred to the capillary tube, one end of the tube is sealed with an inert tube sealant (Critoseal, Krackeler Scientific Inc., Albany, NY, USA), and the tube is centrifuged with a microcentrifuge (Sorvall Legend Micro 21, Waltham, MA, USA) at 6,000g for 3 min. The microcapillary tube is then transferred to the MRR detection coil (Fig. 1d). MRR detection was carried out on the concentrated (near 100% hematocrit) band of RBCs, which allows us to remove the possible confounding effect of variations in blood hematocrit number among individuals. This strategy enhanced the MRR sensitivity, in contrast to the work by Karl and co-workers²⁵, where centrifugation is not used. We discuss the effect of centrifugation on the sensitivity of detection (Supplementary Fig. 10 and Supplementary Discussion).

For the case of P. berghei iRBCs, which have less density^{27, 28, 29, 30} than non-infected RBCs, an additional step was added to lyse the cell and submerge the hemozoin to the bottom of the tubes, upon centrifugation. Two techniques can be used to lyse the cells: Giemsa incubation (Supplementary Fig. 4a) and saponin cell-lysis method (Supplementary Fig. 4d). The whole MRR assay, including both sample preparation (less than 5 min for the case of P. falciparum and less than 10 min for the case of P. berghei) and detection stage (less than a minute), can be done in approximately 10 min. In addition, the whole process is designed to minimize sample loss and avoid contamination and can easily be adapted to process multiple samples at the same time.

MRR measurement and detection (P. falciparum).

We performed ¹H MRR measurements of bulk red blood cells at the resonance frequency of 21.65 MHz inside a portable permanent magnet (Metrolab Instruments, Plan-les-Ouates, Switzerland), B_o = 0.5 T, with a bench-top type console (Kea Magritek, Wellington, New Zealand). We constructed a single resonance proton MRR probe with detection coil of 900-μm inner diameter to accommodate the MRR sample microcapillary tube (o.d.: 900 μm, i.d.: 550 μm) (Drummond Scientific Co., Broomall, PA) for a detection region of approximately 300 nl. Otherwise, we used heparinized microcapillary tubes (o.d.: 1,500 μm, i.d.: 950 μm) (22-260-950, Fisherbrand, Waltham, MA, USA) for a 1-mm microcoil RF-probe. The smaller diameter tube were used for works as shown in Figures 2a and 3, whereas the bigger diameter tubes were used for works shown in Figures 2b and 4. As the filling factors differ, the y axes are not comparable.

We mounted the electronic parts and coil on a single printed circuit board (Fig. 1d). We measured the transverse relaxation rates, R₂ by standard Carr-Purcell-Meiboom-Gill (CPMG) train pulses (60 μs of interecho time) consisting of 5,000 echoes. We acquired a total of 48 scans for signal averaging unless mentioned otherwise. We measured all samples at room temperature. We acquired all data five times and reported as means ± s.e.m. We maintained the transmitter power output at 12.5 mW for a single 90° pulse of pulse length 14 μs, which corresponds to nutation frequency of 17.9 kHz. We set a recycle delay of 1 s, which is sufficient to allow all the spins to return to thermal equilibrium, between each pulse. For experiments with magnetic beads suspended in DI water, we used longer recycle delay of 30 s. We used the same parameters throughout this work except otherwise mentioned throughout the manuscript.

In vitro culture of P. falciparum.

We used P. falciparum 3D7 strain (MR4, Manassas, VA, USA) in this study. Parasites were cultured in RPMI medium 1640 (Invitrogen, Carlsbad, CA, USA) supplemented with 0.3 g of L-glutamine, 5 g of AlbuMAX II (Invitrogen), 2 g NaHCO₃, and 0.05 g of hypoxanthine (Sigma-Aldrich, St. Louis, USA) dissolved in 1 ml of 1m NaOH, together with 1 ml 10 mg ml⁻¹ of gentamicin (Invitrogen). We synchronized the parasites at ring stage using 2.5% D-sorbitol to maintain a synchronous culture. We incubated the cultures at 37 °C after gassing with a 5% CO₂, 3% O₂ and 92% N₂ gas mixture, and their hematocrit maintained at 2.5%. We harvested highly synchronized ring-stage iRBCs (more than 90%). We obtained whole blood from healthy donors and pelleted down the RBCs for parasite culture. We treated the RBC pellet with citrate phosphate dextrose adenine (CPDA) for 3 days before washing it three times with RPMI 1640 and storing it for use (Fig. 2a).

Protocol to prepare highly synchronized ring-stage sample, which mimics the actual patient's peripheral blood sample.

We obtained P. falciparum–infected RBCs of 10% parasitemia level from cultured parasites, as described above. We spun down these iRBCs in a microcentrifuge at 1,000g for about 3 min, washed the pellet three times with isotonic PBS solution, and re-suspended in PBS solution (Supplementary Fig. 1a). The PBS solution was previously bubbled with ambient air for a few minutes so that all the hemoglobin will be converted into oxy-hemoglobin states. For each sample, we determined the relative ratios of parasite stages by standard Giemsa-stained blood microscopy. We prepared samples of various parasitemia levels by spiking iRBCs into uninfected RBCs and subsequently diluted to various parasitemia levels ranging from 0.0002% to 6.5%.

Differential counting of Giemsa-stained blood smears of P. falciparum–infected RBCs.

Differential counting of Giemsa-stained blood smears of the different parasite stages of the highly synchronized ring-stage culture (sample A) indicated 92.1% rings, 7.3% early trophozoites, 0.6% late trophozoites and 0% schizonts³¹. We serially diluted the ring-stage parasites with uninfected RBCs obtained from a healthy donor to obtain different final parasite concentrations (0%, 0.00025%, 0.0005%, 0.001%, 0.0025%, 0.005%, 0.01%, 0.05%, 0.1%, 0.5%, 1%, 5%). We measured the transverse relaxation rates of proton MRR, R₂ of the various infection levels, R_2(iRBC), which reflect the instantaneous parasite load of the sample (Fig. 2a).

Magnetic separation.

We further purified the parasite cultures, which consist of highly purified ring stage²⁰, with MACS system (25 LD columns, Miltenyi Biotec, Bergisch Gladbach, Germany). We preloaded the LD columns with rinsing buffer (0.5% BSA in 1× PBS), which was held with Quadro MACS magnetic support. The blood was then passed through the column. The nonmagnetic parts (uninfected RBCs) and iRBCs with low magnetic susceptibility, especially the ring stage, will be able to pass through the column without being trapped. Most of the late stages (trophozoite and schizont) and suspending hemozoin from the previous generation will be trapped inside the column (Supplementary Fig. 1).

Protocol for day-to-day P. falciparum cell-culture growth with MRR measurement.

We cultured P. falciparum 3D7 strain (MR4) with the protocol as described above. We used D-sorbitol and magnetic separation techniques to produce highly synchronized ring-stage cultures. We spiked highly synchronized ring-stage cultures (94% ring, 6% early trophozoite, 0% schizont) into a healthy donor's blood to form an initial parasitemia of 0.5 parasitized RBCs per μl (0.5 pRBCs μl⁻¹), denoted as day 0. We carried out magnetic separations every 48 h to maintain the synchrony and remove the gametocytes of the previous generation. We performed MRR measurements every 24 h (Fig. 2b). 10 samplings, which consist of roughly 4 μl packed RBCs, for the readings of each time point were taken. We used a heparinized tube (22-260-950, Fisherbrand) to sample out the blood from culture and performed standard microcentrifuge (3,000g, 1 min) to separate the packed RBCs and the media. A microcoil probe (o.d.: 1,550 μm), which accommodate 4 μl volume of packed RBCs, is used in this work. For parasitemia of >0.01%, thick- and thin-smear microscopy techniques were used for verification purposes. 50 fields of more than 5,000 RBCs were counted to estimate the parasitemia (Fig. 2b,c).

MRR measurement and detection of P. berghei ANKA (day-to-day growth monitoring).

We extracted approximately 20 μl of whole blood from the mouse tail (Fig. 3). We extracted uninfected blood samples from the healthy mice (as control), whereas we extracted the infected blood samples from day 2 upon P. berghei intraperitoneal infection. We checked the hematocrit level of the sample by standard centrifugation technique. Blood extracted from mouse was mixed well in ambient air via repeated pipetting. We pipetted approximately 750 nl of RBCs via microcapillary tubes and incubated it with 10 μl of Giemsa solution for approximately 5 min. This is to allow all P. berghei–parasitized RBCs, which have density less than the uninfected RBCs^{27, 28}, to settle down in the same band upon centrifugation (Supplementary Fig. 4b). We then centrifuged the microcapillary tube at 6,000g for 3 min to separate the plasma from the RBCs, and finally slotted the tube into the rf-probe for MRR measurements (Supplementary Fig. 5). We carried out daily MRR measurements from day 2 onwards. We performed ¹H MRR measurements of bulk RBCs with the same parameters as that for P. falciparum experiments.

Mice and P. berghei ANKA parasite.

Male BALB/c mice of 6–8 weeks old were obtained from Sembawang Laboratory Animal Center, National University of Singapore, and subsequently bred under specific pathogen–free (SPF) conditions at Nanyang Technological University Animal Holding Unit. We infected the donor mice with cryopreserved stocks of P. berghei ANKA strain by intraperitoneal injection. We carried out subsequent passages to maintain the parasites, and we monitored the parasitemia levels by thin blood smears stained with Giemsa. For the experiment summarized in Figure 3, we used a batch of 25 healthy mice (n = 25) to determine the control baseline of R₂ reading. We collected 20 μl tail blood with heparin (Sigma-Aldrich) from one mouse by tail nick. We injected the same batch of mice (n = 25) intraperitoneally with 1 × 10⁴ P. berghei–parasitized RBCs. We did the samplings as described from day 2 to day 13 post-infection, with five mice per group each day. We quantified the parasitemia levels using thin blood films stained with Giemsa. We took at least 5 fields, and microscopic analyses were performed by two different individuals. For day 2 to day 5, the parasitemia levels were too low to be determined accurately by thin-smear technique.

Early stage of P. berghei ANKA infection in mice (blinded test).

Male BALB/c mice of 6–8 weeks old were obtained from InVivos Pte Ltd, Singapore, and subsequently bred under SPF conditions at Nanyang Technological University Animal Holding Unit. We infected the donor mice with cryopreserved stocks of P. berghei ANKA strain by intraperitoneal injection. We carried out subsequent passages to maintain the parasites and monitored parasitemia levels by thin blood smears stained with Giemsa. We used fifty-eight healthy mice (n = 58) to determine the control baseline of R₂ readings (Fig. 4). We collected 30 μl of tail blood with heparin (Sigma-Aldrich) from one mouse by tail nick. We injected the same batch of mice intraperitoneally with 1 × 10⁴ (n = 30) or 1 × 10⁷ (n = 18) P. berghei–parasitized RBCs on day 0. We did the samplings of ten mice per day by extracting 600–800 μl of whole blood from each euthanized mouse via cardiac puncture, ranging from day 1 to day 6.

The extracted blood was tested blindly by MRR and blood smear Giemsa microscopy simultaneously. We checked the hematocrit level of each mouse by standard centrifugation technique and resuspended into 26.6% hematocrit level (centrifugation 6,000g for 1 min). We mixed 5 μl of 1% Saponin (Sigma- Aldrich) into 15 μl of iRBCs to lyse the RBC membrane, while leaving the parasites intact. We immediately extracted 20 μl of blood with microcapillary tubes (22-260-950, Fisherbrand) and sealed with an inert tube sealant (Critoseal, Krackeler Scientific Inc.) to prevent further oxidation which may result in False Positive. After incubation in Saponin solution, P. berghei parasites submerged down into a single narrow band at the bottom of the microcapillary tubes upon centrifugation (3,000g for 1 min). Finally, we slotted the microcapillary tube into the rf-probe (o.d.: 1,550 μm) for MRR measurements. We regulated the temperature of the MRR system at 26 °C. We carried out daily MRR measurements from day 1 to day 6. We examined Giemsa-stained thin blood smears under microscope to determine the parasitemia level. We checked up to 100 oil-immersion fields (or 15,000 RBCs) for negative confirmation and early days of post-infection.

Data analysis for blind test.

We evaluated the performance of MRR against blood smear microscopy technique. The variables measured were the number of true positives (TP), number of true negatives (TN), number of false positives (FP), and number of false negatives (FN). Sensitivity was calculated as TP/(TP + FN), specificity was calculated as TN/(TN + FP), the positive predictive value (PPV) was calculated as TP/(TP + FP), and the negative predictive value (NPV) was calculated as TN/(FN + TN). With the assumption that all the intraperitoneal infection were successful, and if the mean R₂ measured was higher than the 75th percentile range of the baseline healthy mice and was statistically significant (P < 0.05) in Student's t-test, it was considered as true positive. Otherwise, if the mean R₂ measured lies below 75th percentile or was statistically not significant (P ≥ 0.05) in Student's t-test as compared to the baseline healthy mice, it was considered as true negative. false positive is the event where the mouse is detected with infection, but was not pre-infected. False negative is the event where the mouse is detected as no infection, but was actually intraperitoneally infected (Table 1).

Ethics statement.

This study was carried out in strict accordance with the recommendations of the NACLAR (National Advisory Committee for Laboratory Animal Research) guidelines under the Animal & Birds (Care and Use of Animals for Scientific Purposes) Rules of Singapore. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the Nanyang Technological University of Singapore (Approval number: ARFSBS/NIE A002). All efforts were made to minimize the suffering of the mice. The use of human blood was approved by the domain-specific review board of Nanyang Technological University (IRB number: NTU-IRB 11/12/2011). Blood component collection service was provided by Blood Transfusion Service and Blood Donation Centre of National University Hospital. All individuals gave informed consent.

1. WHO. World Malaria Report 2013 (World Health Organization, Geneva, 2013).
2. Warhurst, D.C. & Williams, J. ACP Broadsheet no 148. July 1996. Laboratory diagnosis of malaria. J. Clin. Pathol. 49, 533–538 (1996).
   • CAS
   • PubMed
   • Article
3. Amexo, M., Tolhurst, R., Barnish, G. & Bates, I. Malaria misdiagnosis: effects on the poor and vulnerable. Lancet 364, 1896–1898 (2004).
   • PubMed
   • Article
4. Caramello, P., Lucchini, A., Savoia, D. & Gioannini, P. Rapid diagnosis of malaria by use of fluorescent probes. Diagn. Microbiol. Infect. Dis. 17, 293–297 (1993).
   • CAS
   • PubMed
   • Article
5. Moody, A. Rapid diagnostic tests for malaria parasites. Clin. Microbiol. Rev. 15, 66–78 (2002).
   • CAS
   • ISI
   • PubMed
   • Article
6. Makler, M.T., Ries, L., Ries, J., Horton, R. & Hinrichs, D. Detection of Plasmodium falciparum infection with the fluorescent dye, benzothiocarboxypurine. Am. J. Trop. Med. Hyg. 44, 11–16 (1991).
   • CAS
   • PubMed
7. Lee, H., Sun, E., Ham, D. & Weissleder, R. Chip–NMR biosensor for detection and molecular analysis of cells. Nat. Med. 14, 869–874 (2008).
   • CAS
   • ISI
   • PubMed
   • Article
8. Lee, H., Yoon, T.-J., Figueiredo, J.-L., Swirski, F.K. & Weissleder, R. Rapid detection and profiling of cancer cells in fine-needle aspirates. Proc. Natl. Acad. Sci. USA 106, 12459–12464 (2009).
   • PubMed
   • Article
9. Peng, W.K. & Han, J. Biosensor, palm-sized device and method based on Magnetic Resonance Relaxometry. WO Patent WO2012118442 (2012).
10. Peng, W.K., Chen, L. & Han, J. Development of miniaturized, portable magnetic resonance relaxometry system for point-of-care medical diagnosis. Rev. Sci. Instrum. 83, 095115 (2012).
    • CAS
    • PubMed
    • Article
11. Vo, N.N. et al. Highly integrated, low cost, palm-top sized magnetic resonance relaxometry system for rapid blood screening. IFMBE Proc. 43, 558–561 (2014).
12. Sullivan, D. Jr. Hemozoin: a biocrystal synthesized during the degradation of hemoglobin. Biopolym. Online doi:10.1002/3527600035.bpol9007 (15 January 2005).
    • Article
13. Sullivan, D.J., Gluzman, I.Y. & Goldberg, D.E. Plasmodium hemozoin formation mediated by histidine-rich proteins. Science 271, 219–222 (1996).
    • CAS
    • ISI
    • PubMed
    • Article
14. Hackett, S., Hamzah, J., Davis, T.M. & St Pierre, T.G. Magnetic susceptibility of iron in malaria-infected red blood cells. Biochim. Biophys. Acta 1792, 93–99 (2009).
    • CAS
    • PubMed
    • Article
15. Bradley, D. Malaria. in Oxford Textbook of Medicine (ed. Weatherall, D.J.) 835–863 (Oxford Medical Publishers, 2003).
16. Meyer, M.E., Yu, O., Eclancher, B., Grucker, D. & Chambron, J. NMR relaxation rates and blood oxygenation level. Magn. Reson. Med. 34, 234–241 (1995).
    • CAS
    • ISI
    • PubMed
    • Article
17. Thulborn, K.R., Waterton, J.C., Matthews, P.M. & Radda, G.K. Oxygenation dependence of the transverse relaxation-time of water protons in whole-blood at high-field. Biochim. Biophys. Acta 714, 265–270 (1982).
    • CAS
    • ISI
    • PubMed
    • Article
18. Eccles, C. Low field NMR methods and applications. in Encyclopedia of Spectroscopy and Spectrometry (ed. Lindon, J.C.) 1357–1371 (Academic Press, Oxford, 2010).
19. Lambros, C. & Vanderberg, J.P. Synchronization of Plasmodium falciparum erythrocytic stages in culture. J. Parasitol. 65, 418–420 (1979).
    • CAS
    • ISI
    • PubMed
    • Article
20. Ribaut, C. et al. Concentration and purification by magnetic separation of the erythrocytic stages of all human Plasmodium species. Malar. J. 7, 45 (2008).
    • PubMed
    • Article
21. Silamut, K. et al. A quantitative analysis of the microvascular sequestration of malaria parasites in the human brain. Am. J. Pathol. 155, 395–410 (1999).
    • CAS
    • ISI
    • PubMed
    • Article
22. Goodman, A.L. et al. The utility of Plasmodium berghei as a rodent model for anti-merozoite malaria vaccine assessment. Sci. Rep. 3, 1706 (2013).
    • CAS
    • PubMed
    • Article
23. Yañez, D.M., Manning, D.D., Cooley, A.J., Weidanz, W.P. & Van Der Heyde, H. Participation of lymphocyte subpopulations in the pathogenesis of experimental murine cerebral malaria. J. Immunol. 157, 1620–1624 (1996).
    • CAS
    • ISI
    • PubMed
24. Webb, G.A. Radiofrequency microcoils in magnetic resonance. Prog. Nucl. Magn. Reson. Spectrosc. 31, 1–42 (1997).
    • Article
25. Karl, S., Gutiérrez, L., House, M.J., Davis, T.M. & Pierre, T.G.S. Nuclear magnetic resonance: a tool for malaria diagnosis? Am. J. Trop. Med. Hyg. 85, 815–817 (2011).
    • PubMed
    • Article
26. Takeda, K. A highly integrated FPGA-based nuclear magnetic resonance spectrometer. Rev. Sci. Instrum. 78, 033103 (2007).
    • CAS
    • PubMed
    • Article
27. Saul, A., Myler, P., Elliott, T. & Kidson, C. Purification of mature schizonts of Plasmodium falciparum on colloidal silica gradients. Bull. World Health Organ. 60, 755–759 (1982).
    • CAS
    • PubMed
28. Tosta, C.E., Sedegah, M., Henderson, D. & Wedderburn, N. Plasmodium yoelii and Plasmodium berghei: Isolation of infected erythrocytes from blood by colloidal silica gradient centrifugation. Exp. Parasitol. 50, 7–15 (1980).
    • CAS
    • PubMed
    • Article
29. Rüssmann, L., Jung, A. & Heidrich, H.-G. The use of Percoll gradients, elutriator rotor elution, and mithramycin staining for the isolation and identification of intraerythrocytic stages of Plasmodium berghei. Z. Parasitenkd. 66, 273–280 (1982).
    • PubMed
    • Article
30. Eling, W. Ficoll fractionation for the separation of parasitized erythrocytes from malaria infected blood. Bull. World Health Organ. 55, 105–114 (1977).
    • CAS
    • PubMed
31. Crewe, W. Basic Malaria Microscopy 2nd edn. 219 (World Health Organization, 2010).

Download references

This work is supported by the National Research Foundation Singapore under its SMART Centre, BioSystems and Micromechanics IRG and Infectious Disease IRG. W.K.P. acknowledges support from the SMART Postdoctoral Research Fellows Programme and SMART Ignition Grant (ING 11025-BIO(IGN)) and K.R. Roy for culturing and preparing the sample parasites. T.F.K. and C.S.N. acknowledge financial support from the Singapore-MIT Alliance (SMA) Graduate Fellowship.