Hematology analyzers are essential tools for detecting abnormal blood cell populations, which can indicate hematological malignancies and other disorders. This study evaluates the performance of the HORIBA Yumizen H2500, along with its middleware Yumizen P8000, in identifying blasts, atypical lymphocytes, and immature granulocytes. Key diagnostic metrics such as sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), were analyzed to determine the diagnostic accuracy of the analyzer.
The Yumizen H2500, part of HORIBA's high-throughput hematology analyzer lineup, offers several notable features:
The performance validation report2 for the Yumizen H1500 and Yumizen H2500 demonstrated high correlation between the Yumizen H2500 and established systems such as the Sysmex XE-2100 and XN-20 for various parameters of WBC, RBC, Hemoglobin, HCT, MCV and platelet and reticulocyte count (r2~0.95). This study was conducted on 1200 patient records. It showed comparable or superior performance of the Yumizen H2500 in comparison to the Sysmex XN-20 for all measured parameters.
The objective of this study was to assess the performance of the HORIBA Yumizen H2500 hematology analyzer in detecting abnormal cell populations, specifically:
By analyzing true positives (TP), false positives (FP), true negatives (TN), and false negatives (FN), the study assesses the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall efficiency of the HORIBA Yumizen H2500.
A dataset of 407 samples, reduced to 377 distinct cases, was analyzed. It consisted of samples from a hospital laboratory which were flagged by their LIS for manual review due to various considerations such as abnormal scattergram, leukocyte count, patient history or specific physician requests. Consequently, the dataset included both pathological and non-pathological samples. Each sample underwent both manual review and automated flagging by the Yumizen H2500 analyzer. True positives (TP), false positives (FP), true negatives (TN), and false negatives (FN) were calculated to assess diagnostic performance metrics. The study assessed the sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV) and efficiency of the device.
The following flags were considered for the Yumizen H2500 analysis and were matched with manual microscopy findings: a. Blasts suspicion, b. Abnormal matrix, c. Atypical lymphocytes suspicion, d. Immature cells. Additionally, for the calculation of performance based on “any flag” the following flags were also included: e. WBC suspicion, f. Bad separation.
The various diagnostic accuracy variables are defined as follows:
Sensitivity (True Positive Rate)
Sensitivity measures the ability of a test to correctly identify those who have the disease (true positives).
♦ Sensitivity=True Positives (TP) / [True Positives (TP) + False Negatives (FN)]
A high sensitivity means the test is good at detecting the disease, minimizing false negatives.
Specificity (True Negative Rate)
Specificity measures the ability of a test to correctly identify those who do not have the disease (true negatives).
♦ Specificity=True Negatives (TN) / [True Negatives (TN) + False Positives (FP)]
A high specificity means the test is good at ruling out the disease, minimizing false positives.
Positive Predictive Value (PPV)
PPV measures how likely it is that a person who tested positive actually has the disease.
♦ PPV=True Positives (TP) / [True Positives (TP) + False Positives (FP)]
A high PPV means that most positive test results are correct, reducing false alarms.
Negative Predictive Value (NPV)
NPV measures how likely it is that a person who tested negative truly does not have the disease.
♦ NPV=True Negatives (TN) / [True Negatives (TN) + False Negatives (FN)]
A high NPV means that most negative test results are correct, minimizing missed diagnoses.
Efficiency (Overall Accuracy)
Efficiency measures the overall correctness of the test.
♦ Efficiency=[True Positives (TP) + True Negatives (TN)] / Total Cases (TP + TN + FP + FN)
High efficiency means that the test is reliable overall, balancing both sensitivity and specificity.
In the table below, true positives are instances where the sample has blasts, atypical lymphocytes or immature granulocytes confirmed manually, and the machine has flagged these appropriately.
For blasts detection, achieving high sensitivity necessitates including the Abnormal matrix flag. This requires acceptance of a considerable number of False Positives and lowers the machine’s specificity for blast detection. This trade-off is commonly observed in hematology analyzers.
Table 1 summarizes the performance of the HORIBA Yumizen H2500, with its middleware Yumizen P8000, in flagging hematological malignancies. The results demonstrate high sensitivity across various performance parameters evaluated. Specifically, blast detection achieved a sensitivity of 93% and a specificity of 55%. For the detection of atypical lymphocytes, sensitivity was 78% and specificity 72%. With respect to immature granulocytes, sensitivity was 85% and specificity 79%.
Table 1: Performance of Yumizen H2500 in flagging abnormal cells in hematological malignancies
| Variables | TP | FP | TN | FN | Sensitivity | Specificity | PPV | NPV | Efficiency |
| Corresponding flag | |||||||||
| Blast detection | 14 | 162 | 200 | 1 | 93% | 55% | 8 | 100 | 57% |
| Atypical lymphocytes | 31 | 93 | 244 | 9 | 78% | 72% | 25% | 96% | 73% |
| Immature granulocytes | 61 | 65 | 240 | 11 | 85% | 79% | 48 | 96 | 80% |
| Any flag | |||||||||
| All of the above (together) | 101 | 150 | 111 | 15 | 87% | 43% | 40 | 88 | 56% |
*Note: True Positives are those where any of the included abnormal cells (blasts, atypical lymphocytes or immature granulocytes) are found in the manual slide review and the machine has flagged these.
Regarding overall performance, the Yumizen H2500 demonstrated sensitivity of 87% and efficiency of 56% when considering any flag. Notably, it achieved its highest sensitivity (93%) in blast detection. For atypical lymphocytes, sensitivity was 78% along with a strong negative predictive value of 96%. Overall efficiency ranged from 56% to 80%, indicating that the analyzer’s performance varies based on the specific type of abnormal cell detected.
The sensitivity for detecting atypical lymphocytes is 78%, accompanied by a strong negative predictive value (96%), suggesting that the analyzer is more effective at excluding malignancies than confirming them. Overall efficiency ranges from 56% to 80%, indicating that the analyzer’s performance varies based on the specific type of abnormal cell being detected.
When examining the overall performance using "any flag", the Yumizen H2500 exhibits high sensitivity (87%), making it a strong candidate as a primary screening tool for hematological malignancies. Its sensitivity ranges from 78% to 93%, indicating a robust ability to detect cases with abnormal cells, ensuring that very few true positives are missed. This high sensitivity is especially valuable for serious conditions where early and accurate detection is crucial and missed diagnoses can have significant consequences.
Lower specificity and positive predictive value, especially in blast detection, point to a higher rate of false positives. It is worthwhile mentioning here that true positives are those where the sample has blasts confirmed manually after the machine has raised the flag, as specified earlier. In the case of Yumizen H2500, since we are considering the abnormal matrix flag, we get a high number of false positives, reducing the specificity of the machine. One of the analyzer's key strengths is its high negative predictive value (NPV ≥ 88%). This means that when the Yumizen H2500 does not flag a sample, it is usually accurate - making it particularly effective at ruling out disease. This high NPV across the board supports the analyzer's role as a reliable tool for excluding hematologic malignancies.
Efficiency, reflecting overall accuracy, is moderate at approximately 56% to 80%. This is largely attributed to the lower specificity, as the broad "any flag" criteria often trigger false positives. Consequently, while the analyzer reduces the likelihood of missed diagnoses, it does require a notable number of manual slide confirmations.
The Yumizen H2500 has demonstrated especially strong sensitivity in detecting blasts and immature granulocytes, both vital for diagnosing acute leukemia and infections. Nevertheless, its lower specificity may result in over-flagging, requiring manual confirmation.
Previous studies have highlighted the clinical value of the Yumizen H2500. For instance, a diagnostic accuracy study comparing it to the DxH 800, DxH 900, and XN-9000 found the Yumizen H2500 to be more dependable for abnormal cell detection due to its high sensitivity and specificity, along with a low turnaround time.3
Another validation report on both the Yumizen H1500 and H2500 showed strong correlation (r² ≈ 0.95) with established analyzers like the Sysmex XE-2100 and XN-20 across multiple blood parameters, including WBC, RBC, hemoglobin, HCT, MCV, platelet, and reticulocyte counts—based on data from 1200 patient samples.
Additionally, comparative analysis with the Sysmex XN 3000 highlighted consistently strong correlations (r > 0.9) for several important parameters,4 highlighting the Yumizen H2500’s reliability and clinical utility across multiple parameters, including white blood cells, red blood cells and hemoglobin.5
The dataset analyzed in this study was notably enriched with abnormal or particularly challenging samples. Even samples reported as normal by the manual reviewer often exhibited features that would typically prompt a laboratory to flag them for slide review. These same attributes may have led to the Yumizen H2500 flagging the sample as well. In a standard laboratory setting with a broader, unselected sample pool, both the specificity and the Negative Predictive Value (NPV) of the analyzer are expected to surpass the values reported here. This is because NPV (and PPV) are influenced by pre-test probability. In a workload consisting largely of normal samples the NPV will naturally be higher than in this abnormality-rich dataset.
Overall, the findings highlight the value of the Yumizen H2500 as a screening tool, capable of minimizing the risk of missed diagnoses. Nonetheless, final confirmation by microscopy remains essential. The study reinforces the potential for such systems to streamline laboratory workflows by reducing unnecessary slide reviews when paired with high negative predictive value.
The Yumizen H2500 hematology analyzer demonstrates high diagnostic sensitivity across a range of abnormal cell populations, making it particularly effective for screening hematologic malignancies. Although its specificity is moderate, the analyzer’s high negative predictive value positions it as an excellent front-line tool for routine hematology diagnostics.
Albert J. de Graaf1,2
Arjan de Mare1,3
Shubham Rastogi4
Jennita Slomp1,5
1 Unilabs Diagnostics B.V., Enschede, Netherlands
2 Saxenburgh Medisch Centrum, Hardenberg, Netherlands
3 Ziekenhuisgroep Twente, Almelo, Netherlands
4 HORIBA ABX SAS, Montpellier, France
5 Medisch Spectrum Twente, Enschede, Netherland
如您有任何疑问，请在此留下详细需求信息，我们将竭诚为您服务。
