Abstract:Background: A method for accurate immunohematology testing by fluorescence cytometry (FC) was previously described. Nevertheless, the use of vacuum filtration to wash RBCs and a standard-flow cytometer for data acquisition hindered efforts to incorporate this method into an automated platform. Study design and methods: A modified procedure was developed that used low-speed centrifugation of 96-well filter plates for RBC staining. Small-footprint benchtop capillary cytometers (PCA and PCA-96, Guava Technologies, Inc.) were used for data acquisition. Authentic clinical samples from hospitalized patients were tested for ABO group and the presence of D antigen (n = 749) as well as for the presence of RBC alloantibodies (n = 428). Challenging samples with mixed-field reactions and weak antibodies were included. Results were compared to those obtained by column agglutination technology (CAT), and discrepancies were resolved by standard tube methods. Detailed investigations of FC sensitivity and reproducibility were also performed. Results: The modified FC method with the PCA determined the correct ABO group and D type for 98.7 percent of 520 samples, compared to 98.8 percent for CAT (p > 0.05). No-type-determined (NTD) rates were 1.2 percent for both methods. In testing for unexpected alloantibodies, FC determined the correct result for 98.6 percent of 215 samples, compared to 96.3 percent for CAT (p > 0.05). When samples were automatically acquired in the 96-well plate format with the PCA-96, 98.7 percent of 229 samples had correct ABO group and D type determined by FC, compared to 97.4 percent for CAT (p > 0.05). NTD rates were 0.9 and 2.6 percent, respectively. Antibody screens were accurate for 99.1 percent of 213 samples with the PCA-96, compared to 99.5 percent for CAT (p > 0.05). Further investigations demonstrated that FC with the PCA-96 was better than CAT at detecting weak anti-A (p < 0.0001) and alloantibodies. Conclusions: An improved method for FC immunohematology testing has been described. This assay was comparable in accuracy to standard CAT techniques, but had better sensitivity for detecting weak antibodies and was superior in detecting mixed-field reactions (p < 0.005). The FC method demonstrated excellent reproducibility. The compatibility of this assay with the PCA-96 capillary cytometer with plate-handling capabilities should simplify development of a completely automated platform.

Utility of Faster R-CNN in methodological comparison and evaluation of reticulocytes

Experimental evaluation of deep learning method in reticulocyte enumeration in peripheral blood

Impact of Staining Methods and Human Factors on Accuracy of Manual Reticulocyte Enumeration

Performance Verification and Interference Factors of SYSMEX XN-9000 Hematology Analyzer for the Determinations of Reticulocyte Count and Parameters According to CAP Accreditation Requirements

Automated Blood Cell Detection and Counting via Deep Learning for Microfluidic Point-of-Care Medical Devices

Compare the accuracy and precision of Coulter LH780, Mindray BC-6000 Plus, and Sysmex XN-9000 with the international reference flow cytometric method in platelet counting

Using Convolutional Neural Networks for Determining Reticulocyte Percentage in Cats

Automatic retinal image diagnosis system for mass health screenings

Automated reticulocyte response is a good predictor of bone-marrow recovery in pediatric malignancies

Comparison of peripheral blood automated hematopoietic progenitor cell count and flow cytometric CD34+ cell count

Artificial intelligence system of faster region-based convolutional neural network surpassing senior radiologists in evaluation of metastatic lymph nodes of rectal cancer

Improved method for fluorescence cytometric immunohematology testing

Comparison of the conventional tube and erythrocyte-magnetized technology in titration of red blood cell alloantibodies

Fast and accurate automated recognition of the dominant cells from fecal images based on Faster R-CNN

Antinuclear Antibodies Detection: A Comparative Study Between Automated Recognition and Conventional Visual Interpretation

Deep learning ResNet34 model-assisted diagnosis of sickle cell disease via microcolumn isoelectric focusing

Automated Diabetic Retinopathy Grading and Lesion Detection Based on the Modified R-FCN Object-Detection Algorithm

Automatic Identification of Human Erythrocytes in Microscopic Fecal Specimens

Detection of nucleated red blood cells using the Mindray BC-6800Plus hematology analyzer: a clinical performance evaluation

Automatic model for cervical cancer screening based on convolutional neural network: a retrospective, multicohort, multicenter study

Detecting Anemia from Retinal Fundus Images