Automated Morphologic Differentiation Between Iron Deficiency Anemia and Thalassemia

Abstract

Background: Iron deficiency anemia (IDA) and hemoglobinopathies (HbP) are two frequent conditions characterized by microcytemia. Published criteria/scores discriminating these conditions with hematology analyzer parameters are not fully satisfactory. Although patients with HbP have been reported to have more red blood cells (RBC) with a target cell (TC) morphology than patients with IDA, obtaining TC percentages remains a time-consuming task since at least 1000 RBC must be examined.

Methods: Using the Mindray CAL 8000 2.0 0111 and MC-80 module, 152 microcytic samples from 51 patients with IDA and 101 with HbP were analyzed. Data from RBC parameters used in published scores were collected, as well as the percentages of target cells automatically provided by the MC-80.

Results: Patients with IDA had significantly lower median hemoglobin level, red blood cell numbers, and mean corpuscular hemoglobin concentration than those with HbP, yet had more microcytes. Using TC percentages provided by the MC-80 module, receiving operator characteristic curves identified this parameter as the most discriminant to segregate patients with IDA or HbP. With a 0.4% threshold, this yielded a 74.2% sensitivity and 86.3% specificity, confirming that patients with HbP have significantly higher TC percentages.

Conclusion: The automated identification and enumeration of abnormal RBC performed by Mindray MC-80 rapidly provides TC percentages, allowing for a fast discrimination between IDA and HbP in samples with microcytosis, orienting early towards confirmatory tests for these disorders. Moreover, this study confirms TC, which can obviously be obtained through other methods, as a robust parameter in this context.

Keywords: AI‐aided morphology; red blood cells; target cells.

FIGURE 1

Result of the morphological identification of target cells by MC‐80. The top image shows how target cells are highlighted on demand (here 2 target cells). The bottom image shows the result of MC‐80 analysis, in this case 16 cells were observed for a percentage of 0.5%.

FIGURE 2

Discrimination between IDA (n = 51) and HbP (n = 101) based on the percentage of target cells identified by the MC‐80. The dotted line indicates the segregating 0.4% threshold. The ordinate presents percentages on a logarithmic scale. Box‐whisker graphs indicate the median and 25/75 percentiles. Student t‐test confirmed a statistically significant difference with p < 0.0001.