A field evaluation of the Coulter STKS

Machine learning algorithms for the detection of spurious white blood cell differentials due to erythrocyte lysis resistance

AIMS

Red blood cell (RBC) lysis resistance interferes with white blood cell (WBC) count and differential; still, its detection relies on the identification of an abnormal scattergram, and this is not clearly adverted by specific flags in the Beckman-Coulter DXH-800. The aims were to analyse precisely the effect of RBC lysis resistance interference in WBC counts, differentials and cell population data (CPD) and then to design, develop and implement a novel diagnostic machine learning (ML) model to optimise the detection of samples presenting this phenomenon.

METHODS

WBC counts, differentials and CPD from 232 patients (anaemia or liver disease) were compared with 100 healthy controls (HC) using analysis of variance. The data were analysed after a corrective action, and the analyser differentials were also compared with the digital leucocyte differentials. The ML support vector machine (SVM) algorithm was trained with 70% of the samples (n=233) and the 30% remaining (n=99) were employed exclusively during the validation phase.

RESULTS

We identified that impedance WBC was not affected by the RBC lysis resistance interference while the DXH-800 differentials overestimated lymphoid subpopulations (17.6%), sometimes even yielding spurious lymphocytosis, and the latter were corrected when sample dilution was performed. The ML-SVM algorithm allowed the classification of the pathological groups when compared with HC with validation accuracies corresponding to 97.98%, 100% and 88.78% for the global, anaemia and liver disease groups, respectively.

CONCLUSIONS

The proposed algorithm has an impressive discriminatory potential and its application would be a valuable support system to detect spurious results due to RBC lysis resistance.

Validation and implementation of an algorithm for reporting the automated absolute neutrophil count from selected flagged specimens

Samples that are flagged by automated cell counters and, therefore, require a time-consuming microscopic review cause unacceptable wait times for patients in hematology/oncology clinics. We used a set of 518 samples to validate that 5 flags on Sysmex XE/XT instruments (Sysmex, Kobe, Japan) could safely be ignored when the absolute neutrophil count (ANC) was the primary clinical question. The R(2) between automated and manual ANCs was 96.9% for samples triggering non-ANC flags, with 1 clinically significant discrepant sample out of 296 (0.3%). A new test code allowing clinicians to specify "ANC-only" was implemented, and these non-ANC specific flags were disregarded. The new order set was used in 16.3% of patients. Automated reporting of the ANC in selected flagged samples reduced the review rate by 60% and decreased the turnaround time by 100 minutes. This approach to automatically report the ANC in selected flagged specimens in which the ANC is the primary clinical interest safely decreases the turnaround time for many ANC samples.

Quality specification in haematology: the automated blood cell count

BACKGROUND

Quality specifications for automated blood cell counts include topics that go beyond the traditional analytic stage (imprecision, inaccuracy, quality control) and extend to pre- and post-analytic phases.

METHODS

In this review pre-analytic aspects concerning the choice of anticoagulants, maximum conservation times and differences between storage at room temperature or at 4 degrees C are considered. For the analytic phase, goals for imprecision and bias obtained with various approaches (ratio to biologic variation, state of the art, specific clinical situations) are evaluated. For the post-analytic phase, medical review criteria (algorithm, decision limit and delta check) and the structure of the report (general part and comments), which constitutes the formal act through which a laboratory communicates with clinicians, are considered.

RESULTS

K2EDTA is considered the anticoagulant of choice for automated cell counts. Regarding storage, specimens should be analyzed as soon as possible. Storage at 4 degrees C may stabilize specimens from 24 to 72 h when complete blood count (CBC) and differential leucocyte count (DLC) is performed. For precision, analytical goals based on the state of the art are acceptable while for bias this is satisfactory only for some parameters.

CONCLUSIONS

In haematology quality specifications for pre- and analytical phases are important, but the review criteria and the quality of the report play a central role in assuring a definite clinical value.

Evaluation of the coulter LH 750 haematology analyzer compared with flow cytometry as the reference method for WBC, platelet and nucleated RBC count

The Coulter LH 750 is a new haematology analyser with several new features: a count of nucleated red blood cells (NRBCs), automated WBC correction in presence of a flag indicating a cellular interference and a lower incidence of platelet or WBC interference flags when compared with the GEN.S, our current instrument. We had three main goals in our study: evaluating the LH 750 WBC counts when a GEN.S flag suggests a risk of WBC interference, ascertaining whether the platelet counts not flagged by the LH 750 were accurately assessed in samples flagged by the GEN.S and evaluating the NRBC assay provided by the LH 750. Flow cytometry, using CD45 and CD41, respectively for WBC and platelet labelling, was used as a reference method to assess the accuracy of the LH 750 counts. NRBC were identified by double labelling with propidium iodide (PI) and CD45, NRBCs being CD45-/PI+. A significant relationship was found between LH 750 and flow cytometric WBC counts, whether a WBC correction was made by the LH 750 (r = 0.9809, n = 54) or not (r = 0.9901, n = 23). A highly significant relationship was observed for platelets not only in the range from 0 to 450 x 10(9)/l (r = 0.981, n = 108) but also in cases of thrombocytopenia (range: 0-80 x 10(9)/l; r = 0.956, n = 51). In samples with NRBCs, the NRBC percentages given by the LH 750 and by flow cytometry were highly correlated (r = 0.977, n = 60) and WBC counts were accurate. In conclusion, the reduction in flagging by the LH 750, the accuracy of the results, and the availability of a NRBC count, constitute major advantages.

Utility of automated counting to determine absolute neutrophil counts and absolute phagocyte counts for pediatric cancer treatment protocols

BACKGROUND

Absolute neutrophil counts (ANCs) and absolute phagocyte counts (APCs) are used to guide cancer treatment. Although automated counting could replace manual counting, data showing correlations are lacking. By analyzing blood samples from children undergoing cancer treatment, the authors determined whether ANCs and APCs obtained by automated methods correlated positively with ANCs and APCs obtained manually.

METHODS

The authors analyzed 3640 consecutive peripheral blood samples. Leukocyte counts determined by Beckman-Coulter Gen-S or HmX analyzers (Beckman-Coulter, Miami, FL) were used to calculate counts obtained by automated or manual methods. Automated differential counts were obtained by automated analyzers and manual differential counts were performed by medical technologists. Counts underwent linear regression analysis. The authors evaluated 5 cutoff values for ANCs and APCs commonly used in decision-making related to cancer treatment: 300/muL, 500/muL, 750/muL, 1000/muL, and 1500/muL. Manually determined ANCs and APCs served as standards to determine the sensitivity, specificity, positive and negative predictive values, and kappa coefficient for automated counting.

RESULTS

R(2) values were 0.81 for ANCs determined by manual and automated methods and 0.84 for APCs determined by both methods. The specificity of the automated method was > 90% for all ranges of ANCs and APCs, except one (APCs < 300/muL). There was excellent agreement (kappa > 0.9) between ANCs determined by manual and automated methods and APCs calculated by both methods.

CONCLUSIONS

Automated methods of determining ANCs and APCs for children undergoing cancer treatment were reliable and can replace manual counting. Blood smear examination to validate ANCs and APCs determined by automated methods was needed only in selected cases.

Evaluation of a hematology analyzer with canine and feline blood

A semiautomatic electronic blood cell counter (Sysmex F-800:Toa Medical Electronics Europa Gmbh, Hamburg, Germany) was evaluated using canine and feline blood, following the International Committee for Standardization in Hematology protocol (ICSH, 1984). Precision and overall reproducibility were acceptable for all the parameters studied except for the feline platelet count, in which overlapping of erythrocyte and platelet populations prohibited determination of an accurate platelet count. Since carry-over from canine hematocrit values and platelet counts and from feline hematocrit values was unsatisfactory, the use of a blank diluent sample between different analyses was necessary. Linearity of the analyzer was acceptable in the studied range. Thirty canine and feline blood samples were analyzed using the Sysmex F-800 and a manual method. Correlations between both methods were acceptable for all the parameters, except for feline platelet count and erythrocyte indices for both species. In the storage study, red blood cell count and hemoglobin concentration were the parameters with the longest stability (72 hours at 4 degrees C and 25 degrees C) in both species. A statistically significant increase in MCV was obtained at 12 hours post-extraction in canine samples stored at 25 degrees C and at 24 hours in refrigerated samples. Feline leucocyte counts showed a downward trend at 12 hours post-extraction at both temperatures. Canine platelet count decreased significantly at 6 hours post-extraction in samples stored at 4 degrees C. During the evaluation period, Sysmex F-800 was user friendly and appeared well suited for routine canine and feline blood cell analysis.

Changes in automated complete blood cell count and differential leukocyte count results induced by storage of blood at room temperature

OBJECTIVE

To delineate changes that occur in various parameters of automated complete blood cell count (CBC) and differential leukocyte count (differential) on prolonged storage of blood at room temperature.

DESIGN

A CBC and an automated differential were performed on the Coulter Gen.S on 40 K(3) (tripotassium ethylenediamine-tetraacetate) EDTA-anticoagulated blood specimens once daily, specimen volume permitting, for 3 to 7 days. Specimens were kept at room temperature throughout the study. The results were tabulated using a personal computer with Excel software. Percent change or absolute difference from the initial value for each parameter for each subsequent day of the study period was calculated.

RESULTS

Among the CBC parameters, hemoglobin, red blood cell count, and mean corpuscular hemoglobin were stable for the duration of the study (7 days), white blood cell count was stable for at least 3 days (up to 7 days, if the count was within or above the normal range), and platelet count was stable for at least 4 days (up to 7 days, if the count was within or above the normal range). The mean corpuscular volume, mean platelet volume, hematocrit, and red blood cell distribution width each increased, and the mean corpuscular hemoglobin concentration decreased from day 2 onward. Among the differential parameters, the relative percentages and absolute numbers of neutrophils, lymphocytes, and eosinophils tended to increase, whereas those of monocytes trended downward over time. Limited data on basophils did not reveal an appreciable change.

CONCLUSION

Blood specimens stored at room temperature for more than 1 day (up to 3 days or possibly longer) were found to be acceptable with some limitations for CBC but not for the differential.

The white blood cell differential: three methods compared

The analysis of the automated blood cell count is an essential tool in haematological diagnostics. However, in the case of the white blood cell differential the microscopy method, although tedious, often serves as reference. We evaluated the ABX Pentra 120 Retic haematology analyser in comparison to the Coulter STKS haematology system and the microscopy method with respect to accuracy, precision and reliability. We compared 308 samples (239 samples from adults and 69 from children) including patients with oncological diseases. The comparison of the white blood cell differential revealed strong correlations between the results obtained with the ABX Pentra 120 Retic and the microscopy method, the Coulter STKS and the microscopy method and both automated methods (values of paediatric samples in parentheses; neutrophils: rs > or = 0.933 (rs > or = 0.951), lymphocytes: rs > or = 0.907 (rs > or = 0.945), monocytes: rs > or = 0.584 (rs > or = 0.459) and eosinophils: rs > or = 0.963 (rs > or = 0.966)). The analytical performance of automatic analysers for the detection of the morphological "left shift" was determined for all samples in comparison to the microscopical white blood cell differential. The sensitivity, specificity and efficiency depended strongly on the chosen threshold levels and were different for both analysers. The sensitivity for flagging a left shift increased with an increasing proportion of neutrophil bands, metamyelocytes, myelocytes and promyelocytes. Our study suggests that the ABX Pentra 120 Retic haematology analyser, as well as the Coulter STKS haematology system are useful tools for routine analysis in haematology.

Cytofluorometric methods for assessing absolute numbers of cell subsets in blood. European Working Group on Clinical Cell Analysis

The enumeration of absolute levels of cells and their subsets in clinical samples is of primary importance in human immunodeficiency virus (HIV)+ individuals (CD4+ T- lymphocyte enumeration), in patients who are candidates for autotransplantation (CD34+ hematopoietic progenitor cells), and in evaluating leukoreduced blood products (residual white blood cells). These measurements share a number of technical options, namely, single- or multiple-color cell staining and logical gating strategies. These can be accomplished using single- or dual-platform counting technologies employing cytometric methods. Dual-platform counting technologies couple the percentage of positive cell subsets obtained by cytometry and the absolute cell count obtained by automated hematology analyzers to derive the absolute value of such subsets. Despite having many conceptual and technical limitations, this approach is traditionally considered as the reference method for absolute cell count enumeration. As a result, the development of single-platform technologies has recently attracted attention with several different technical approaches now being readily available. These single-platform approaches have less sources of variability. A number of reports clearly demonstrate that they provide better coefficients of variation (CVs) in multicenter studies and a lower chance to generate aberrant results. These methods are therefore candidates for the new gold standard for absolute cell assessments. The currently available technical options are discussed in this review together with the results of some cross-comparative studies. Each analytical system has its own specific requirements as far as the dispensing precision steps are concerned. The importance of precision reverse pipetting is emphasized. Issues still under development include the establishment of the critical error ranges, which are different in each test setting, and the applicability of simplified low-cost techniques to be used in countries with limited resources.

Higher incidence of elevated body temperature or increased C-reactive protein level in asthmatic children showing transient reduction of theophylline metabolism

The authors investigated whether theophylline metabolism is decreased in asthmatic patients and what condition may be related to its reduction. Fifty-two children with asthma were given 15 mg/kg/day aminophylline intravenously at a constant rate. Blood and spot urine samples were collected at 24 hours, 48 hours, and 72 hours after beginning infusion. The ratio of plasma theophylline concentration at 72 hours to that at 24 hours (C72h/C24h) varied from 0.42 to 1.51 (average 0.894). Plasma theophylline concentration of patients with lower C72h/C24h than average reduced significantly, while the concentration of those with higher C72h/C24h remained unchanged. The urinary ratio of the sum of the metabolites to theophylline was significantly increased in the patients with the lower ratio. Among the demographic characteristics examined, significant difference was found only in the incidence of patients with C-reactive protein (CRP) of 0.5 mg/dl or greater or patients with a fever of 37.5 degrees C or greater when admitted. Acute febrile illness accompanied by increased CRP level may affect theophylline metabolism.

Preliminary evaluation of the new hematology analyzer COULTER GEN-S in a university hospital

The COULTER GEN-S system (COULTER Corp, Miami, USA) is an automated hematology instrument that is designed to provide a complete hematological profile including white blood cells (WBC), complete blood count (CBC) differential count (diff) and the reticulocyte parameters. It was evaluated in our laboratory over a one month period. A preliminary study was performed using the GEN-S software revision 1D. The evaluation had two purposes: 1) evaluation of the GEN-S specifications; 2) comparison of its analytical performance with the hematology analyzer currently used in our laboratory. The first part of the evaluation showed that the COULTER GEN-S is reproducible, has linearity beyond the specifications given by the manufacturer and produces stable results up to 48 hours after blood collection. The evaluation of analytical performance included: 1) a comparison between the GEN-S CBC and diff numerical results and the COULTER STKS (COULTER Corp, Miami, USA). These comparisons showed that the results given by the two methods are similar and suggested that the COULTER GEN-S could replace the current hematology instrument in use in our laboratory; 2) a performance analysis, to measure the system's ability to detect morphologic abnormalities, when compared to the reference blood smear examination. This part of the evaluation was performed on both normal and abnormal samples. The results of this analysis showed that the sensitivity of the GEN-S was excellent, especially regarding blast cells, immature granulocytes, nucleated red blood cells (NRBCs) and platelet clumps when using the complete suspect flagging system of the instrument. In addition, the use of the review criteria in our laboratory allowed to detect all hematological diseases. The overall false negative rate was 0.9%. Thus we consider that the COULTER GEN-S system is suited for use in medium-to large hospital laboratories which perform more than 100 CBC/day. Overall, the instrument had excellent performance, with a throughput of more than 100 samples/h. Its user friendly workstation has complete patient data management, which is in compliance with good laboratory practices in France.

Epidemiology studies in immunotoxicity evaluations

Studies in humans designed to detect immunomodulation from exposure to xenobiotics present challenging problems to epidemiologists and immunotoxicologists. Exposed and control groups must be carefully selected, exposure to the xenobiotic must be sufficiently high and well-documented, and the referent group should be as similar as possible to the exposed. Immune markers/functional tests in an individual may be influenced by sunlight exposure, medication, illness and use of recreational drugs; all of these potential confounding factors must be addressed. Sample acquisition is usually performed at sites geographically distant from the controlled environment of an investigator's laboratory, yielding an assortment of new problems that would not occur in clinical or hospital situations. Regulations and guidelines concerning the transport of biological samples and potential hazards of HIV and HBV exposures to personnel must be adapted to field conditions. Since the application of immunotoxicological techniques to populations exposed to xenobiotics is relatively new, and the ability to measure an increasing number of immune biomarkers of activation, suppression, autoimmunity or hypersensitivity is rapidly expanding, there are difficulties in the interpretation of statistically positive results (sometimes within the normal range) and their potential health significance. Finally, both biological and methodological factors complicate the assessment of dose-response/concentration effect relationships in human immunotoxicity studies, and traditional dose-response relationships may not always be present.

Evaluation of a haematological analyser (Sysmex F-800) with equine blood

A semiautomatic electronic blood cell counter (Sysmex F-800) was evaluated with equine blood, according to the protocol of the International Committee for Standardization in Haematology (ICSH, 1984). The precision and overall reproducibility were acceptable for all the parameters studied except for the platelet count, in which a coefficient of variation of 18.8% and 21.7% was obtained for within and between batch precision and 26.76% for overall reproducibility. Carry-over for the haematocrit value and platelet count was unsatisfactory, thus the use of a blank diluent sample between different analyses was necessary. Linearity of the analyser was acceptable in the range studied. A total of 100 equine blood samples were studied with the Sysmex F-800 and the manual method. The correlations between both Methods were acceptable for all the parameters. In the storage study, the red blood cell count and the haemoglobin concentration were the parameters with longest stability (72 h) at 4 degrees C and 25 degrees C. Platelet count decreased significantly at 3 h post-extraction in those samples stored at 4 degrees C. During the evaluation period, Sysmex F-800 was easy to handle and it is well suited for routine equine blood cell analysis.

Evaluation of automated basophil counting by using fluorescence-labelled monoclonal antibodies

The shortcomings of current methods of basophil enumeration detract from the clinical value of the basophil count. Moreover, sophisticated and costly techniques of automated basophil counting hardly can be validated for lack of a suitable reference method. We investigated whether a flow cytometric technique using double staining with fluorescence-labelled monoclonal antibodies (mAb) CD45-FITC and CD14-PE on a Coulter Epics Profile II could be used to evaluate basophil counting performance of hematology analyzers. The technique was compared with the 800-cell manual differential, the Coulter STKS, and the Cobas Argos 5 Diff. Precision: STKS, Argos and Profile II showed a precision analogous to a 2,173, 2,250-, and 14,705-cell differential, respectively, illustrating the superiority of automated methods. Accuracy (150 normal and abnormal samples): Using the Profile II as reference the STKS showed a notably weaker correlation than the Argos (r = 0.581 and 0.718, respectively), although this difference was nearly concealed when the imprecise manual differential served as reference (r = 0.517 and 0.562, respectively). The Profile II correlated relatively well with the manual differential (r = 0.730). Analyzing 137 healthy adult subjects, we obtained a reference range of 0.33 to 1.35% (0.020 to 0.102 x 10(9) basophils/L) for the mAb-based method. These data would recommend mAb-based basophil counting as a valuable tool for instrument evaluation. However, an observed bias of 0.09% against the manual differential suggests that modifications are necessary before this technique can be considered as new reference method.

Peripheral blood monocyte counting: towards a new reference method

Flow cytometric enumeration of monocytes stained with fluorescence-labelled monoclonal antibodies has been proposed as a possible reference method for monocyte counting. We compared precision and accuracy of monocyte counting of the Coulter STKS, the Cobas Argos 5 Diff, the 800-cell manual differential, and the Coulter Epics Profile II flow cytometer using double-staining with fluorescence-labelled monoclonal antibodies (CD45-FITC and CD14-PE). Precision: STKS, Argos and Profile II achieved a precision analogous to a 3423-, 1298-, and 11089-cell differential, respectively, confirming the superiority of automated methods. Accuracy (136 normal and abnormal samples): Correlation of automated methods with the manual differential was good (STKS: r = 0.934, Argos 5 Diff: r = 0.808, Profile II: r = 0.924; Spearman's rank correlation coefficient). The mean relative STKS monocyte result was 0.52 +/- 1.63% (mean +/- SD) higher than the manual differential, whereas the Argos 5 Diff results were 1.22 +/- 2.51% lower (p < 0.001). Profile II results showed a small bias against the manual differential (-0.18 +/- 1.44%, p < 0.05). Analysing 135 healthy adult subjects on the Profile II, males were found to have a higher mean monocyte count (relative count: 6.95 +/- 1.43% vs. 5.86 +/- 0.98%; absolute count: 0.48 +/- 0.15 x 10(9)/l vs. 0.39 +/- 0.11 x 10(9)/l, p < 0.001) and a higher and wider normal range than females (relative count: 4.97 to 9.78% vs. 4.26 to 7.81%, absolute count: 0.30 to 0.84 x 10(9)/l vs. 0.25 to 0.65 x 10(9)/l). Flow cytometry based on fluorescence-labelled monoclonal antibodies for monocyte enumeration seems an efficient tool to evaluate the monocyte counting performance of haematology analysers and an ideal successor to the manual differential as reference method for monocyte counting.

Lymphocyte subset determination using a hematology analyzer

Anti-CD4 antibody (T4)-coated microspheres were used to label CD4 cells in whole blood. The mixture was lysed and analyzed by a modified Coulter VCS hematology analyzer, which differentiated microsphere-labeled cells by a change in Coulter volume, conductance, and light scatter. %CD3+/CD4+ fluorescent values from a profile were compared to %CD4 values using the VCS-microsphere method. CD3 gating was used to exclude CD4+ monocytes from the 90LS-FALS lymphocyte gate. The results correlated well (R = 0.996). The percentage of CD4+ lymphocytes from profile scatterplots and VCS scatterplots showed a line of regression close to the equivalence line (n = 76, slope = 0.96) when CD3 gating was used for the profile. These results suggest that CD3 gating, though necessary for 90LS-FALS scatterplots, may not be necessary for volume-conductance-light scatterplots.

Evaluation of the Roche Cobas Argos 5Diff automated haematology analyser with comparison to a Coulter STKS

The performance of the Roche Cobas Argos 5Diff (Argos) automated haematology analyser was evaluated by comparison to manual blood film examination and a Coulter STKS (STKS) analyser. The Argos demonstrated excellent between and inter-batch imprecision for all parameters, except the MCHC, and good linearity for Hb, WBC and platelet count (PLT). After an initial fall the PLT, results were stable for up to six h at 18 degrees C in EDTA(K3) after which an increasing proportion of cells were classified as lymphocytes. Results of 239 patient samples analysed on both instruments, compared by linear regression, gave excellent correlation (r2 > 0.90) for most parameters with the exceptions of the MCHC (0.317), eosinophils% (0.756), monocytes% (0.48) and basophils% (0.002). 'Flagging' of cellular abnormalities by the Argos resulted in excellent sensitivity (97.5%), specificity (93.2%) and efficiency/agreement (93.2%), with fewer false positive and negative results than the STKS, although these differences were not statistically significant. The performance characteristics of the Argos were comparable to those of the STKS with a possible improvement in its flagging abilities.

Evaluation and use of the white blood cell differential provided by the Coulter STKS in a children's hospital

The Coulter STKS was evaluated in a children's hospital, in order to (a) compare the WBC differential given by the instrument to a 400 cell visual differential (reference method); (b) evaluate the sensitivity and specificity of the alarm system, and (c) provide data concerning the use and interpretation of results in children. 653 blood samples were collected. The Coulter STKS results were studied in 523 patients having no morphological abnormalities in the blood smears, separated into subgroups according to the presence of STKS alarms and according to age. The results were found accurate both in STKS negative and STKS positive patients (i.e., those with alarms: 'Blasts', Imm Gran 2, Variant Lymph, NRBC, review slide). Negative STKS results had the same accuracy in all age groups, except in neonates where slide review must be systematically performed. The instrument exhibited a good sensitivity of the suspect flags studied (91.4%), with a lower specificity (72%) reflecting the number of false positive results found in our group, probably due to the cytological features particular to children. However, it was shown that the numerical results given by the Coulter STKS in positive patients could be taken into account, provided that a scan of the blood smear was negative for morphological WBC abnormalities.

Diagnostic performance: a comparative study of the leukocyte differential count on four automated haematology analysers

The diagnostic performance of the haematology analysers, Technicon H1, Sysmex NE 8000, Coulter STKS and Sequoia Cell Dyn 3000 was determined by comparing their results for automated differential counting with the result obtained for 400 cells by the manual microscopic method. These comparative studies were performed on a group of adults, containing both normal individuals (n = 150) and those affected by various pathologies (n = 113). The number of morphologically false negatives is low for all the systems (from 1.8% for the Cell Dyn 3000 to 6.2% for the NE 8000), while the number of distributionally false negatives was slightly higher (from 4.4% for the Cell Dyn 3000 to 7% for the H1 and NE 8000). The morphological flags, though useful for improving the diagnostic performance of the instruments, show a rather modest sensitivity when taken individually: immature granulocytes/bands from 71% for the STKS to 43% for the NE 8000; blasts from 66.6% for the H1 to 53.3% for the STKS, atypical lymphocytes from 59% for the H1 to 13.6% for the NE 8000; erythroblasts from 42.8% for the STKS to 7% for the NE 8000.