Differential counting of blood leukocytes using automated microscopy and a decision support system based on artificial neural networks--evaluation of DiffMaster Octavia

From microscope to micropixels: A rapid review of artificial intelligence for the peripheral blood film

Artificial intelligence (AI) and its application in classification of blood cells in the peripheral blood film is an evolving field in haematology. We performed a rapid review of the literature on AI and peripheral blood films, evaluating the condition studied, image datasets, machine learning models, training set size, testing set size and accuracy. A total of 283 studies were identified, encompassing 6 broad domains: malaria (n = 95), leukemia (n = 81), leukocytes (n = 72), mixed (n = 25), erythrocytes (n = 15) or Myelodysplastic syndrome (MDS) (n = 1). These publications have demonstrated high self-reported mean accuracy rates across various studies (95.5% for malaria, 96.0% for leukemia, 94.4% for leukocytes, 95.2% for mixed studies and 91.2% for erythrocytes), with an overall mean accuracy of 95.1%. Despite the high accuracy, the challenges toward real world translational usage of these AI trained models include the need for well-validated multicentre data, data standardisation, and studies on less common cell types and non-malarial blood-borne parasites.

Artificial Intelligence-Based Management of Adult Chronic Myeloid Leukemia: Where Are We and Where Are We Going?

Artificial intelligence (AI) is emerging as a discipline capable of providing significant added value in Medicine, in particular in radiomic, imaging analysis, big dataset analysis, and also for generating virtual cohort of patients. However, in coping with chronic myeloid leukemia (CML), considered an easily managed malignancy after the introduction of TKIs which strongly improved the life expectancy of patients, AI is still in its infancy. Noteworthy, the findings of initial trials are intriguing and encouraging, both in terms of performance and adaptability to different contexts in which AI can be applied. Indeed, the improvement of diagnosis and prognosis by leveraging biochemical, biomolecular, imaging, and clinical data can be crucial for the implementation of the personalized medicine paradigm or the streamlining of procedures and services. In this review, we present the state of the art of AI applications in the field of CML, describing the techniques and objectives, and with a general focus that goes beyond Machine Learning (ML), but instead embraces the wider AI field. The present scooping review spans on publications reported in Pubmed from 2003 to 2023, and resulting by searching "chronic myeloid leukemia" and "artificial intelligence". The time frame reflects the real literature production and was not restricted. We also take the opportunity for discussing the main pitfalls and key points to which AI must respond, especially considering the critical role of the 'human' factor, which remains key in this domain.

Digital assessment of peripheral blood and bone marrow aspirate smears

The diagnosis of benign and neoplastic hematologic disorders relies on analysis of peripheral blood and bone marrow aspirate smears. As demonstrated by the widespread laboratory adoption of hematology analyzers for automated assessment of peripheral blood, digital analysis of these samples provides many significant benefits compared to relying solely on manual review. Nonetheless, analogous instruments for digital bone marrow aspirate smear assessment have yet to be clinically implemented. In this review, we first provide a historical overview detailing the implementation of hematology analyzers for digital peripheral blood assessment in the clinical laboratory, including the improvements in accuracy, scope, and throughput of current instruments over prior generations. We also describe recent research in digital peripheral blood assessment, particularly in the development of advanced machine learning models that may soon be incorporated into commercial instruments. Next, we provide an overview of recent research in digital assessment of bone marrow aspirate smears and how these approaches could soon lead to development and clinical adoption of instrumentation for automated bone marrow aspirate smear analysis. Finally, we describe the relative advantages and provide our vision for the future of digital assessment of peripheral blood and bone marrow aspirate smears, including what improvements we can soon expect in the hematology laboratory.

Multicentric evaluation of the variability of digital morphology performances also respect to the reference methods by optical microscopy

INTRODUCTION

Despite the important diagnostic role of peripheral blood morphology, cell classification is subjective. Automated image-processing systems (AIS) provide more accurate and objective morphological evaluation. The aims of this multicenter study were the evaluation of the intra and inter-laboratory variation between different AIS in cell pre-classification and after reclassification, compared with manual optical microscopy, the reference method.

METHODS

Six peripheral blood samples were included in this study, for each sample, 70 May-Grunwald and Giemsa stained PB smears were prepared from each specimen and 10 slides were delivered to the seven laboratories involved. Smears were processed by both optical microscopy (OM) and AIS. In addition, the assessment times of both methods were recorded.

RESULTS

Within-laboratory Reproducibility ranged between 4.76% and 153.78%; between-laboratory Precision ranged between 2.10% and 82.2%, while Total Imprecision ranged between 5.21% and 20.60%. The relative Bland Altman bias ranged between -0.01% and 20.60%. The mean of assessment times were 326 ± 110 s and 191 ± 68 s for AIS post reclassification and OM, respectively.

CONCLUSIONS

AIS can be helpful when the number of cell counted are low and can give advantages in terms of efficiency, objectivity and time saving in the morphological analysis of blood cells. They can also help in the interpretation of some morphological features and can serve as learning and investigation tools.

Establishment of Review Criteria Coordinating With the Automated Digital Cell Morphology Identification System in a Specialized Women’s and Children’s Hospital

Objective

We aimed to establish appropriate review criteria for blood cell analysis in a specialized women’s and children’s hospital. Also, the CellaVision DI-60, was developed as one of the automated digital cell morphology analyzer, we evaluated if it was shown to be most effective under the certain review criteria.

Methods

A total of 2890 blood samples were detected to optimize the previously established review criteria for women and children with the Sysmex XE-2100. A total of 623 samples were used to validate the criteria.

Results

The microscopic-review rate based on the initial review criteria was 51.0%. After optimization, it was reduced to 17.3% and the false-negative rate was 3.85%. There was > 80% consistency between manual review results and CellaVision DI-60 preclassification when samples triggered the platelet- or red cell-related rules. The sensitivity for abnormalities (immature granulocytes, nucleated red blood cells) of reclassification was 90% to 100% and the false-negative rate was < 5%. However, direct microscopic review was required when the “Blasts/AbnLympho?” and “Atypical Lympho?” flags were triggered.

Conclusion

Specialized review criteria are needed for women and children. An automated morphology identification system might help to improve the review criteria.

Rise of the Machines: Artificial Intelligence and the Clinical Laboratory

BACKGROUND

Artificial intelligence (AI) is rapidly being developed and implemented to augment and automate decision-making across healthcare systems. Being an essential part of these systems, laboratories will see significant growth in AI applications for the foreseeable future.

CONTENT

In laboratory medicine, AI can be used for operational decision-making and automating or augmenting human-based workflows. Specific applications include instrument automation, error detection, forecasting, result interpretation, test utilization, genomics, and image analysis. If not doing so today, clinical laboratories will be using AI routinely in the future, therefore, laboratory experts should understand their potential role in this new area and the opportunities for AI technologies. The roles of laboratorians range from passive provision of data to fuel algorithms to developing entirely new algorithms, with subject matter expertise as a perfect fit in the middle. The technical development of algorithms is only a part of the overall picture, where the type, availability, and quality of data are at least as important. Implementation of AI algorithms also offers technical and usability challenges that need to be understood to be successful. Finally, as AI algorithms continue to become available, it is important to understand how to evaluate their validity and utility in the real world.

SUMMARY

This review provides an overview of what AI is, examples of how it is currently being used in laboratory medicine, different ways for laboratorians to get involved in algorithm development, and key considerations for AI algorithm implementation and critical evaluation.

Comparison of the effects of deferasirox, deferoxamine, and combination of deferasirox and deferoxamine on an aplastic anemia mouse model complicated with iron overload

Background and aim

Iron overload is commonly observed during the course of aplastic anemia (AA), which is believed to aggravate hematopoiesis, cause multiple organ dysfunction, lead to disease progression, and impair quality of life. Deferasirox (DFX) and deferoxamine (DFO) are among the most common iron chelation agents available in the clinical setting. The aim of this study was to investigate if the combination therapy with DFX and DFO is superior in hematopoietic recovery and iron chelation.

Methods

Briefly, we developed a composite mouse model with AA and iron overload that was consequently treated with DFX, DFO, or with a combination of both agents. The changes in peripheral hemogram, marrow apoptosis, and its related protein expressions were compared during the process of iron chelation, while the iron depositions in liver and bone marrow and its regulator were also detected.

Results

The obtained results showed that compared to DFX, DFO has a better effect in protecting the bone marrow from apoptosis-induced failure. The combination of DFO and DFX accelerated the chelation of iron, while their efficiency on further hemogram improvement appeared limited.

Conclusion

To sum up, our data suggest that single treatment with DFO may be a better choice for improving the hematopoiesis during the gradual chelation treatment irrespective of the convenience of oral DFX, while the combination treatment should be considered for urgent reduction of the iron burden.

A composite mouse model of aplastic anemia complicated with iron overload

Iron overload is commonly encountered during the course of aplastic anemia (AA), but no composite animal model has been developed yet, which hinders drug research. In the present study, the optimal dosage and duration of intraperitoneal iron dextran injection for the development of an iron overload model in mice were explored. A composite model of AA was successfully established on the principle of immune-mediated bone marrow failure. Liver volume, peripheral hemogram, bone marrow pathology, serum iron, serum ferritin, pathological iron deposition in multiple organs (liver, bone marrow, spleen), liver hepcidin, and bone morphogenetic protein 6 (BMP6), SMAD family member 4 (SMAD4) and transferrin receptor 2 (TfR2) mRNA expression levels were compared among the normal control, AA, iron overload and composite model groups to validate the composite model, and explore the pathogenesis and features of iron overload in this model. The results indicated marked increases in iron deposits, with significantly increased liver/body weight ratios as well as serum iron and ferritin in the iron overload and composite model groups as compared with the normal control and AA groups (P<0.05). There were marked abnormalities in iron regulation gene expression between the AA and composite model groups, as seen by the significant decrease of hepcidin expression in the liver (P<0.01) that paralleled the changes in BMP6, SMAD4, and TfR2. In summary, a composite mouse model with iron overload and AA was successfully established, and AA was indicated to possibly have a critical role in abnormal iron metabolism, which promoted the development of iron deposits.

A new thresholding technique based on fuzzy set as an application to leukocyte nucleus segmentation

BACKGROUND AND OBJECTIVES

The main aim of this paper is to segment leukocytes in blood smear images using interval-valued intuitionistic fuzzy sets (IVIFSs). Generally, uncertainties occur in terms of vagueness through brightness levels of image. Processing of such uncertain images can be efficiently handled by using fuzzy sets, particularly IVIFSs.

METHODS

Logarithmic membership function is utilized for computing membership values corresponding to intensities of the pixel. Non-membership function of IVIFS is constructed by using Yager generating function. By varying parameters, 256 IVIFSs are generated. An IVIFS is selected from 256 IVIFSs having maximizing ultrafuzziness along with varying threshold. Threshold is determined by finding an IVIFS with maximum similarity between ideal segmented and segmented results obtained from the proposed method.

RESULTS

Quantitatively, the segmented images are evaluated using precision-recall, receiver operator characteristic curves, Jaccard coefficient and measure for structural similarity index along with the time taken for segmenting nucleus, and their results are compared with results of existing methods. Performance measures reveal that the proposed method seems to segment leukocytes better than other comparable methods.

CONCLUSIONS

Segmentation of leukocytes using the proposed method helps the analyst in differentiating various types of leukocytes and in the determination of leukocyte count, and the counting is essential in finding out diseases related to reduction or surplus quantity of these cells.

An automated image analysis system can be beneficial in preclassification of leucocytes in children with hematological disease

This study was aimed to evaluate the analytical performance of an automated image analysis system (a pilot model of Diff Master(™) Octavia) for the preclassification of leucocytes in children with hematological disease. Manual microscopy performed by pediatric hematologists was used as the reference method. Five mature cell class and blasts were evaluated. Diff Master Octavia correctly preclassified 87.4% of all leucocytes with a high reproducibility. The overall accuracy was found to be 93.0%. Clinical sensitivity was 97.7% and specificity was 76.0%. The average time per slide for Diff Master(™) Octavia was 2.3  min lower than that of manual method. Our results indicated that the Diff Master(™) Octavia can detect and preclassify leucocytes accurately; therefore, it can be used as an efficient and fast method in pediatric hematology routine.

Image resizing using saliency strength map and seam carving for white blood cell analysis

Background

A new image-resizing method using seam carving and a Saliency Strength Map (SSM) is proposed to preserve important contents, such as white blood cells included in blood cell images.

Methods

To apply seam carving to cell images, a SSM is initially generated using a visual attention model and the structural properties of white blood cells are then used to create an energy map for seam carving. As a result, the energy map maximizes the energies of the white blood cells, while minimizing the energies of the red blood cells and background. Thus, the use of a SSM allows the proposed method to reduce the image size efficiently, while preserving the important white blood cells.

Results

Experimental results using the PSNR (Peak Signal-to-Noise Ratio) and ROD (Ratio of Distortion) of blood cell images confirm that the proposed method is able to produce better resizing results than conventional methods, as the seam carving is performed based on an SSM and energy map.

Conclusions

For further improvement, a faster medical image resizing method is currently being investigated to reduce the computation time, while maintaining the same image quality.

Performance evaluation and relevance of the CellaVision DM96 system in routine analysis and in patients with malignant hematological diseases

The CellaVision™ DM96 is an automated image analysis system dedicated to locating and preclassifying the various types of white blood cells in peripheral blood smears. The system also partially characterizes of the red blood cell morphology and is able to perform platelet counts. We routinely analyzed the blood samples from 440 patients with quantitative and/or qualitative abnormalities detected by the XE-2100 Sysmex™. Only 2.6% of cells are not identified by DM96™. After classification of the unidentified cells very good correlation coefficients are observed between DM96™ and manual microscopy for most hematological parameters and accuracy is judged excellent up to 98%. For most common parameters, false positive and false negative ratios are also very good. Whatever the pathology and the number of blasts on smear, all patients were positive for blast detection on DM96™. The system is a useful tool for assisting in the diagnosis and classification of most acute or chronic leukemia. Automatic cell location and preclassification, along with unique cell views on the computer screen, could reduce the time spent performing differentials and make real-time collaboration between colleagues a natural part of the classification process. The workstation also provides an ergonomically correct and relaxed working environment. We suggest its use in routine analysis; the system could be very helpful for the accurate morphological diagnosis of samples from patients with malignant hematological disease.

Automation and validation of a rapid method to assess neutrophil and monocyte activation by routine fluorescence flow cytometry in vitro

The aim of the present study was to design an automated-gating hematology fluorescence flow cytometry methodology permitting the assessment of neutrophil and monocyte activation in EDTA-anticoagulated whole blood based on cell granularity, lipid membrane components, cell shape and volume, and total cell nucleic acid (NA) compounds. For particularly monitoring the proper functioning of patients' innate immune system as the first line defense against microbial invaders, the suitable test system should be rapid, simple, reliable by yielding reproducible results. It must be validated against established methods, and it must prove to work in selected clinical settings, e.g. in intensive care unit (ICU) environments. The adaptation of a routine hematology cell analyser utilizing fluorescence flow cytometry resulted in a potentially useful system for all requirements. It proved to detect in real-time and in a reliable and reproducible way the main cellular response reactions of neutrophils and monocytes during externally stimulated immune defense. Validation was successful when comparing it to established methods. The quantified activation effects were dose dependent from the applied activating agents. Cellular response kinetics could be measured and described and showed to be in line with the prevailing cell response models. Upon applying the test method to a healthy population of volunteers and a first cohort of ICU patients with and without evident immune depression, the test revealed excellent cellular responses to external activating cytotoxic stimuli (lipopolysaccharide; LPS) for the control group, slightly weaker response from ICU patients without immune depression and no response from patients with evident immune depression.We conclude that routine hematology fluorescence flow cytometry can accurately and reproducibly measure different activation steps of monocytes and polymorphonuclear neutrophilic granulocytes to defined external stimuli. This may potentially be applied as a STAT (Latin statim = immediately) and routine screening and surveillance method for inflammatory diseases.

Identification and quantification of high fluorescence-stained lymphocytes as antibody synthesizing/secreting cells using the automated routine hematology analyzer XE-2100

OBJECTIVES

The aim of this study was to classify and quantify the high fluorescence lymphocytes area (HFL-count) from the SYSMEX XE-2100 leucocyte differential channel as antibody-synthesizing or -secreting cells (ASC, plasma cells or lymphoplasmacytoid cells) in reactive diseases. To unequivocally identify the HFL cells, all possibly eligible cell populations have been investigated: activated B-lymphocytes, activated T-lymphocytes, large granular lymphocytes (LGL), activated monocytes, and immature granulocytes.

METHODS

In total, 85 patients were analyzed on the XE-2100 and compared with the automated image analysis system Cellavision Diffmaster 96 based on artificial neural network and immunophenotyping method with the BD FACSCalibur.

RESULTS

Reproducibility tests for HFL demonstrated a mean coefficient of variation of 13.9% for very low results and 1.5% for high results. The linearity data showed a good correlation (R(2) = 0.99) between expected and measured HFL. The comparison with possibly eligible cell populations showed no significant correlation between activated monocytes and immature granulocytes, with most immature granulocytes (promyelocyte I or II), natural killer cells or LGLs, activated T-lymphocytes, and sub-T-lymphocytes populations. However, for activated B-lymphocytes an excellent significant correlation with the peripheral blood smear, and the immunophenotyping method has been found with R(2) = 0.900, P < 0.001 and R(2) = 0.897, P < 0.001, respectively. The slope of 1.1 and intercept of minus 5 cells/microL of the regression equation between HFL-count and ASC (smear) do indicate an excellent quantification of the HFL-count, as well.

CONCLUSION

The fully automated SYSMEX XE-2100 HFL-count identifies and quantifies the ASC cells (activated B-lymphocytes) with high precision and reliability in patients without hematology system diseases, thus providing a potential screening and monitoring tool for any patient with suspected infection. Additional studies are required to comprehend in more detail the full clinical utility of an HFL (ASC) count as a potential diagnostic indicator of inflammation, infection, or sepsis.

Examination of peripheral blood films using automated microscopy; evaluation of Diffmaster Octavia and Cellavision DM96

BACKGROUND

Differential counting of peripheral blood cells is an important diagnostic tool. Yet, this technique requires highly trained staff, is labour intensive and has limited statistical reliability. A recent development in this field was the introduction of automated peripheral blood differential counting systems. These computerised systems provide an automated morphological analysis of peripheral blood films, including a preclassification of both red and white cells (RBCs and WBCs, respectively).

AIMS

To investigate the ability of two automated microscopy systems to examine peripheral blood smears.

METHODS

Two automated microscopy systems, the Cellavision Diffmaster Octavia (Octavia) and Cellavision DM96 (DM96), were evaluated.

RESULTS

The overall preclassification accuracy values for the Octavia and the DM96 systems were 87% and 92%, respectively. Evaluation of accuracy (WBC analysis) showed good correlation for both automated systems when compared with manual differentiation. Total analysis time (including post classification) was 5.4 min/slide for the Octavia and 3.2 min/slide for the DM96 (100 WBC/slide) system. The DM96 required even less time than manual differentiation by an experienced biomedical scientist.

CONCLUSIONS

The Octavia and the DM96 are automated cell analysis systems capable of morphological classification of RBCs and WBCs in peripheral blood smears. Classification accuracy depends on the type of pathological changes in the blood sample. Both systems operate most effectively in the analysis of non-pathological blood samples.

Does the band cell survive the 21st century?

OBJECTIVES

The differentiation of white blood cells is a worldwide-accepted method to obtain medical information. The conventional microscopic differential, however, is a laborious and expensive test with a low statistical value. Especially for band cell identification there is a wide range of variance. In this report we describe the intervariability of band cell enumeration.

METHODS

From a septic patient, an EDTA anti-coagulated blood sample was obtained and a smear was made and stained (May-Grünwald Giemsa). A PowerPoint presentation was made twice of 100 random cells and sent to 157 different hospital laboratories in the Netherlands for a leukocyte differential. In the first survey neutrophils were differentiated in segmented and band neutrophils whereas in the second survey no discrimination was made between segmented and band neutrophils.

RESULTS

The first survey was responded by 68% of the laboratories (756 individuals) and the second survey by 73% of the laboratories (637 individuals). The laboratory mean values of the segmented neutrophils were 42.9% (SD: 7.8, range 22-64%) and 69.9% (SD: 1.4, range 62-72%) for the first and second survey respectively. For the individual technicians the values of the segmented neutrophils were 43.9% (SD: 11.2, range 15-72%) and 70.0% (SD: 2.0, range 59-77%) for the first and second survey respectively.

CONCLUSIONS

Because of the enormous variation of band cell counting we recommend to cease quantitative reporting of band cells, especially since the results only have a clinical relevance in a limited number of pathological circumstances.

Segmentation of complex cell clusters in microscopic images: Application to bone marrow samples

BACKGROUND

Morphologic examination of bone marrow and peripheral blood samples continues to be the cornerstone in diagnostic hematology. In recent years, interest in automatic leukocyte classification using image analysis has increased rapidly. Such systems collect a series of images in which each cell must be segmented accurately to be classified correctly. Although segmentation algorithms have been developed for sparse cells in peripheral blood, the problem of segmenting the complex cell clusters characterizing bone marrow images is harder and has not been addressed previously.

METHODS

We present a novel algorithm for segmenting clusters of any number of densely packed cells. The algorithm first oversegments the image into cell subparts. These parts are then assembled into complete cells by solving a combinatorial optimization problem in an efficient way.

RESULTS

Our experimental results show that the algorithm succeeds in correctly segmenting densely clustered leukocytes in bone marrow images.

CONCLUSIONS

The presented algorithm enables image analysis-based analysis of bone marrow samples for the first time and may also be adopted for other digital cytometric applications where separation of complex cell clusters is required.