Comparison of 21 artificial intelligence algorithms in automated diabetic retinopathy screening using handheld fundus camera.

BACKGROUND: Diabetic retinopathy (DR) is a common complication of diabetes and may lead to irreversible visual loss. Efficient screening and improved treatment of both diabetes and DR have amended visual prognosis for DR. The number of patients with diabetes is increasing and telemedicine, mobile handheld devices and automated solutions may alleviate the burden for healthcare. We compared the performance of 21 artificial intelligence (AI) algorithms for referable DR screening in datasets taken by handheld Optomed Aurora fundus camera in a real-world setting. PATIENTS AND METHODS: Prospective study of 156 patients (312 eyes) attending DR screening and follow-up. Both papilla- and macula-centred 50° fundus images were taken from each eye. DR was graded by experienced ophthalmologists and 21 AI algorithms. RESULTS: Most eyes, 183 out of 312 (58.7%), had no DR and mild NPDR was noted in 21 (6.7%) of the eyes. Moderate NPDR was detected in 66 (21.2%) of the eyes, severe NPDR in 1 (0.3%), and PDR in 41 (13.1%) composing a group of 34.6% of eyes with referable DR. The AI algorithms achieved a mean agreement of 79.4% for referable DR, but the results varied from 49.4% to 92.3%. The mean sensitivity for referable DR was 77.5% (95% CI 69.1-85.8) and specificity 80.6% (95% CI 72.1-89.2). The rate for images ungradable by AI varied from 0% to 28.2% (mean 1.9%). Nineteen out of 21 (90.5%) AI algorithms resulted in grading for DR at least in 98% of the images. CONCLUSIONS: Fundus images captured with Optomed Aurora were suitable for DR screening. The performance of the AI algorithms varied considerably emphasizing the need for external validation of screening algorithms in real-world settings before their clinical application.

What is already known on this topic? Diabetic retinopathy (DR) is a common complication of diabetes. Efficient screening and timely treatment are important to avoid the development of sight-threatening DR. The increasing number of patients with diabetes and DR poses a challenge for healthcare.What this study adds? Telemedicine, mobile handheld devices and artificial intelligence (AI)-based automated algorithms are likely to alleviate the burden by improving efficacy of DR screening programs. Reliable algorithms of high quality exist despite the variability between the solutions.How this study might affect research, practice or policy? AI algorithms improve the efficacy of screening and might be implemented to clinical use after thorough validation in a real-life setting.

Towards a point-of-care strip test to diagnose sickle cell anemia.

A rapid test to identify patients with sickle cell disease could have important benefits in low-resource settings. Sickle cell anemia (SCA) affects about 300,000 newborns each year, the majority of whom are born in sub-Saharan Africa. Low-cost therapies are available to treat SCA, but most countries in sub-Saharan Africa lack robust neonatal screening programs needed to identify patients in need of treatment. To address this need, we developed and evaluated a competitive lateral flow assay that identifies patients with SCA (genotype HbSS) in 15 minutes using undiluted whole blood. A small volume of blood (0.5 µL- 3 µL) is mixed with antibody-coated blue latex beads in a tube and applied to the strip. Strips are then placed in a well of running buffer and allowed to run for 10 minutes. Laboratory evaluation with samples containing different proportions of hemoglobin A (HbA) and hemoglobin S (HbS) indicated that the test should enable identification of SCA patients but not persons with sickle cell trait (SCT). We evaluated the test using 41 samples from individuals with SCA, SCT, and normal blood. With visual inspection or quantitative analysis, we found a 98% accuracy when differentiating SCA from normal and SCT samples as a group (90% sensitivity and 100% specificity for identifying SCA). This work demonstrates important steps towards making a lateral flow test for hemoglobinopathies more appropriate for point-of-care use; further work is needed before the test is appropriate for clinical use.

Preparation, characterisation and application of europium(III) chelate-dyed polystyrene-acrylic acid nanoparticle labels.

Preparation, characterisation and application of europium(III) chelate-incorporated polystyrene-acrylic acid (AAc) nanoparticle labels featuring diverse AAc proportions is described. Emulsion copolymerisation of styrene and AAc was used to synthesise uniform-sized nanoparticles, approximately 50 nm in diameter. The structural, fluorescence and functional properties of the nanoparticles were characterised to obtain the optimal polymer composition of particulate labels. The AAc content had only a delicate effect on the fluorescence of the chelate and structural characteristics of the particles. The fluorescence spectra or lifetime of the incorporated europium(III) chelate were not notably affected, and all the particles were analogous in size, had monomodal size distributions and good colloidal stability. However, the AAc content affected strongly on the stability of the incorporated dye and functionality of the labels. Nanoparticles having up to 5.4 mass% of AAc were stable and applicable in high-sensitivity assays, where low detection limit and variation were achieved. The nanoparticles possessing 7.2 or 11.5 mass% of AAc lost remarkably the dye content during storage influencing their usability as labels in assays. Overall, the characterisation and employment results achieved evidenced the importance of the structural composition of nanoparticles, moreover, the knowledge of the effects of structural changes is utilisable in the development of improved nanoparticle labels and assay applications.

Synthesis, characterization, and application of Eu(III), Tb(III), Sm(III), and Dy(III) lanthanide chelate nanoparticle labels.

Preparation and characterization of europium(III), terbium(III), samarium(III), and dysprosium(III) polystyrene nanoparticle labels with lanthanide-specific fluorescence properties has been presented. Emulsion copolymerization of styrene and acrylic acid was used to synthesize uniform-sized nanoparticles approximately 45 nm in diameter. Europium(III) and samarium(III) lanthanides were chelated with 2-naphthoyltrifluoroacetone and trioctylphosphine oxide to dye the spherical particles, whereas terbium(III) and dysprosium(III) chelate complexes contained a newly synthesized ligand, 4-(2,4,6-tridecyloxyphenyl)pyridine-2,6-dicarboxylic acid. The fluorescence properties of the four lanthanides-including a wide Stokes shift, a narrow emission peak, and long fluorescence lifetime-were retained despite the incorporation into the nanoparticles. Furthermore, the nanoparticles, containing more than 1000 lanthanide chelates, were detectable at label concentrations 3 orders of magnitude lower than the corresponding soluble lanthanide chelate labels. The applicability of the labels prepared was demonstrated by a heterogeneous sandwich-type immunoassay for human prostate-specific antigen, where the lowest limits of detection of 1.6, 2.4, 10.1, and 114.2 ng/L were achieved using europium(III), terbium(III), samarium(III), and dysprosium(III) nanoparticles, respectively. The spectral and functional properties of the lanthanide-embedded polystyrene nanoparticles developed here suggest that the technology is applicable for high-sensitivity multicolor assays.

Immunoassay of total prostate-specific antigen using europium(III) nanoparticle labels and streptavidin-biotin technology.

Nanoparticle labels conjugated with biomolecules are used in a variety of different assay applications. We investigated the possibility of using europium(III)-labeled 68-nm nanoparticles coated with monoclonal antibodies or streptavidin (SA) to detect prostate-specific antigen (PSA) in serum. The selection of a suitable antibody pair and interference caused by the combination of nanoparticle label and structurally complex analyte were of special interest. A set of antibodies recognizing different epitope areas of PSA was mapped to find the optimal antibody pair for the immunometric nanoparticle-based assay. Different assay configurations were tested to obtain a good correlation with a conventional method based on biotinylated detection antibodies and europium(III) chelate-labeled streptavidin. Monoclonal capture antibody 5E4 was covalently coated on a microtitration well surface; biotinylated 5H6 monoclonal antibody (Mab) was used for detection, and europium(III)-labeled streptavidin-coated nanoparticles were utilized for signal generation. Total PSA concentrations were determined from a panel of male serum samples to test the developed assay. The correlation of the nanoparticle-based and reference assays was good; y=0.9844x-0.1252, R2=0.98, n=27; and the lowest limit of detection of the assay (LLD=0.83 ng/l) was 35-fold lower than for the reference method. The assay application presented here, where a structurally complex analyte is detected, combines the exceptionally high affinity of streptavidin-biotin technology and the high specific activity of long lifetime fluorescence nanoparticle labels. The general characteristics of this combination should permit the development of various immunoassay applications featuring high sensitivity, rapidity, and low consumption of reagents.

Highly sensitive immunoassay of free prostate-specific antigen in serum using europium(III) nanoparticle label technology.

BACKGROUND: Recent proceedings in utilization of europium(III) chelate-dyed polystyrene nanoparticles as labels have combined the advantages of an enhanced monovalent binding affinity and a high specific activity of nanoparticle-antibody bioconjugate. Our objective was to evaluate the performance of the nanoparticle label technology with biological samples in an immunoassay of free prostate-specific antigen (PSA-F) using a standard microtitration well platform. METHODS: Long-lifetime luminescent europium(III)-chelate nanoparticles, 107 nm in diameter, were coated with a PSA-F specific monoclonal antibody. The two-step noncompetitive immunoassay was performed in a microtitration well coated with a second monoclonal antibody. The signal of the surface-bound nanoparticle-antibody bioconjugates was measured directly from the bottom of the well using a standard time-resolved plate fluorometer. RESULTS: The detection limit (mean + 2SD) of the nanoparticle-based PSA-F assay was 0.21 ng/l using a 20-microl sample volume. The assay response was linear up to 5 microg/l, and the functional sensitivity was approximately 0.5 ng/l. The within-run imprecision for spiked serum samples at concentrations 0.0005-0.5 microg/l was 6.4-21.8%, and the within-run and between-run imprecisions for serum samples at concentrations 0.2-2.5 microg/l were 3.4-7.2% and 4.4-7.6%, respectively. The concentrations obtained from serum samples correlated well with the reference immunoassay; slope = 1.018 +/- 0.018; intercept = 0.012 +/- 0.021 microg/l; S(y/x) = 0.112 microg/l; r = 0.993; n = 51. CONCLUSIONS: The developed method demonstrated acceptable performance characteristics allowing clinical studies utilizing patient samples with extremely low concentrations of PSA-F. The present assay detected PSA-F in most of samples from prostatectomized men and in few samples from healthy women that were nondetectable according to the reference immunoassay.