ai_agent_IP

ai_agent

maintenance

instantTrialDurationMinutes

isBiopharma

enableNewMyJoveSectionDesign

educationFeatures

seoMetaApi

ai_copilot_institution_blacklist

ai_chat_institution_blacklist

freeAccessWindow

ssoRoleCapture

scheduledMaintenanceBanner

joveLabsFeatures

disable_autoplay_for_bots

extended_trial_duration

5 minutes

The study was supported by the Internal Research Grant 2024 of Moravian-Silesian Hospital Havirov, Havirov, Czech Republic, and the Internal Grant IGA_LF_2026_012 of Palacky University Olomouc, Olomouc, Czech Republic.

Our study, published in 20252, used archived material and was conducted in accordance with the updated principles of the Declaration of Helsinki. Patients provided written informed consent for laboratory examination, anonymous data collection and analysis, and research use, including the publication of anonymized image data. The study was approved by the Local Ethics Committee of Hospital Hav&#237;&#345;ov (approval IGS_2023_HEM_PAT).
NOTE: The ADM protocol for PB and BM, following automated staining, is reported.
1. Preparation and staining of peripheral blood and bone marrow smears using an automated instrument
<ol>
	<li>Before turning on the instrument, perform the following checks: tubing condition, network connection, waste container status, sufficient reagents, and an adequate supply of microscope slides. 
	NOTE: Reagent containers must not be empty, and all reagents must be within expiry date.</li>
	<li>Turn on the instrument.</li>
	<li>The main unit automatically performs a self-check.</li>
	<li>Log in to the software using a valid username and password.</li>
	<li>Operate the instrument using the current software versions. Follow the manufacturer&#8217;s recommendations and user manual (home screen&#8212;Figure 1).</li>
	<li>Use the Menu to select operations, the Worklist to manage requests for slide preparation and smear preparation, the Browser to review the data in requests, and the Status tab to check the status of the smear being prepared.</li>
	<li>Download a new slide preparation request from the host computer or create a new request. 
	NOTE: Use slides with beveled and rounded edges measuring 24.7&#8211;26.3 mm by 74.7&#8211;76.3 mm, with a width of 0.9&#8211;1.2 mm and a matte edge of 15&#8211;20 mm.</li>
	<li>Edit or delete slide preparation requests as needed.</li>
	<li>Prepare the smear:
	<ol>
		<li>Always use latex or nitrile gloves when handling PB and BM samples and smears.</li>
		<li>Select the smear preparation mode.</li>
		<li>Place the sample rack into the sampler. 
		NOTE: Use only sample tubes that have been specifically approved by the stainer manufacturer.</li>
		<li>Check that the spreader-glass is intact, completely clean, and securely placed in the holder.</li>
		<li>Start smear preparation and staining using the dialog window.</li>
		<li>Check the status of the smear being prepared.</li>
	</ol>
	</li>
	<li>Review the details of the slide preparation requests.</li>
	<li>Remove the sample rack.</li>
	<li>Remove the prepared and stained slides. 
	Checkpoint: Does smear quality at visual inspection (post-output) meet the defined acceptance criteria? A minimum monolayer zone of adequate length and cell density must be present.</li>
	<li>Perform instrument inspection and cleaning before shutdown.</li>
	<li>Turn off the instrument. 
	NOTE: 
	Staining protocol for peripheral blood- 
	Stain 1: May&#8211;Gr&#252;nwald stain; Stain 2: Giemsa&#8211;Romanowski stain. Undiluted stain 1 (73 mL) 2:48 min; diluted stain 1 (73 mL, diluent pH 7.0 phosphate buffer, dilution ratio 1:10) 3:36 min. Diluted stain 2 (26 mL, diluent pH 7.0 phosphate buffer, dilution ratio 1:10) 4:00 min. Rinse count (439 mL, automated process) 
	1. Drying (automated process) 3:36 min. 
	 
	Staining protocol for bone marrow- 
	Stain 1: May&#8211;Gr&#252;nwald stain; Stain 2: Giemsa&#8211;Romanowski stain. Undiluted stain 1 (73 mL) 3:36 min; diluted stain 1 (73 mL, diluent pH 7.0 phosphate buffer, dilution ratio 1:10) 2:48 min. Diluted stain 2 (26 mL, diluent pH 7.0 phosphate buffer, dilution ratio 1:10) 10:00 min. Rinse count (439 mL, automated process) 
	1. Drying (automated process) 3:36 min. 
	 
	CAUTION: When handling chemicals in the laboratory, follow the instructions in their official safety data sheets, which include composition, hazard symbols, first aid instructions, emergency response procedures for accidental leaks and fires, handling and storage guidelines, personal protective equipment requirements, toxicological and environmental information, and instructions for transport and disposal. Dispose of waste liquids, reagents, and consumables as medical infectious waste, and/or industrial waste in accordance with local laws and regulations. If possible, use a central waste disposal system.</li>
</ol>
2. Automated digital cytomorphology of PB
<ol>
	<li>Use the application for the differential count of white blood cells (WBCs), assessment of red blood cell (RBC) morphology, and estimation of platelet (PLT) count. Verify the proposed classification of each cell type. 
	NOTE: The operator is an experienced certified laboratory technician with at least five years of practice or a certified laboratory scientist.</li>
	<li>Start the analyzer.</li>
	<li>Log in to the software using a valid username and password.</li>
	<li>Operate the instrument using the current software versions. Follow the manufacturer&#8217;s recommendations and user manual.</li>
	<li>Use the System Control View tab for quality control (QC) procedure and for processing orders.</li>
	<li>Perform daily internal QC using a control slide (routine PB smear with a WBC count above 10 &#215; 109/L). Run the cell detection test at least once a day, which recognizes 100 cells. 
	NOTE: Cell detection test validates the slide preparation process and verifies the analyzer&#8217;s ability to detect cells. The slide must be of sufficient quality for the analyzer to detect the cells required for analysis. For QC acceptance, 97% of cells must be successfully detected.</li>
	<li>Use a PB smear prepared with the slide-making and staining instrument.</li>
	<li>Place the stained smear into the analyzer&#8217;s loading drawer.</li>
	<li>Enter the test identification manually or via barcode scanner; the request from the laboratory information system is loaded automatically.</li>
	<li>Close the analyzer input door.</li>
	<li>Process the smear. 
	NOTE: The analyzer automatically moves the smear under the microscope, scans it, and captures images of WBCs, as well as a monolayer image of RBCs. The analyzer performs a preliminary classification of all detected WBCs and evaluates RBC morphology.</li>
	<li>Review the preliminary results and reclassify where necessary, either directly on the analyzer or using remote review software. Consider the proposed three classification options, including the probability percentages. 
	NOTE: The automatic preliminary classification takes only a few seconds. In cases of ambiguous or borderline cellular classifications, double reading is necessary. For cells with differing classifications, inter-observer consensus must be reached. The reclassified cell is labeled with a checkmark in the upper-right corner of the image.</li>
	<li>In WBC mode, review the preliminary classification of nucleated cells. 
	NOTE: Quality flags requiring digital morphology and mandatory expert review include blasts, immature granulocytes, atypical lymphocytes, reactive lymphocytes, neoplastic lymphocytes, monocytosis above 1.0 &#215; 109/L, lymphocytosis above 4.0 &#215; 109/L, and unreliable differential count. Any manual reclassification by the operator or pathologist must be logged with the user ID and timestamp.</li>
	<li>If incorrect classification is identified, reassign the cell to the correct category by dragging its image with the mouse. 
	NOTE: For examples of PB cell misclassifications, see Figure 2.</li>
	<li>In RBC mode, review the assessment of RBC morphological changes and correct any inaccurate evaluations or add findings by selecting additional pathological features (e.g., target cells, schistocytes, ovalocytes).</li>
	<li>In PLT mode, review the PLT evaluation.</li>
	<li>Add comments manually or select from predefined text options. 
	NOTE: Auto-verification rules must be formally validated, documented, and reviewed at least annually.</li>
	<li>Confirm the final result, which is then automatically sent to the laboratory information system. 
	NOTE: The final result is exported as a numerical differential count along with additional comments, if any. For samples with low cellularity (WBC below 2.0 &#215; 109/L), two smears are automatically prepared, scanned, and evaluated individually.</li>
	<li>Log out and shut down the analyzer. 
	NOTE: A 100-cell count is applied for each PB sample, following ICSH recommendations17. WBC classification categories (classes) include blast, promyelocyte, myelocyte, metamyelocyte, band neutrophil, segmented neutrophil, eosinophil, basophil, monocyte, lymphocyte, variant lymphocyte, and plasma cell. Non-WBC classification categories (classes) include smudge cell, artifact, giant platelet, platelet clump, and erythroblast. RBC morphology classification categories (classes) include polychromatic RBCs, hypochromic RBCs, anisocytosis, macrocytosis, microcytosis, and poikilocytosis.Za&#269;&#225;tek formul&#225;&#345;e 
	 
	CAUTION: When handling chemicals in the laboratory, follow the instructions in their official safety data sheets, which include composition, hazard symbols, first aid instructions, emergency response procedures for accidental leaks and fires, handling and storage guidelines, personal protective equipment requirements, toxicological and environmental information, and instructions for transport and disposal. Dispose of waste liquids, reagents, and consumables as medical infectious waste, and/or industrial waste in accordance with local laws and regulations. If possible, use a central waste disposal system.</li>
</ol>
Konec formul&#225;&#345;e
3. Automated digital cytomorphology of BM
<ol>
	<li>Use the system for the differential count of PB and BM cells. Verify the proposed classification of each cell type. 
	NOTE: The use of BM analysis is described. The operator is an experienced certified laboratory technician with at least five years of practice, or a certified laboratory scientist, or a certified physician.</li>
	<li>Start the analyzer.</li>
	<li>Log in to the software using a valid username and password.</li>
	<li>Operate the instrument using the current software versions. Follow the manufacturer&#8217;s recommendations and user manual.</li>
	<li>Use Project for processing orders and slide management, Edit for configuration of classification rules and pre-classification strategy, Report for report format settings and report generation, and the Manage tab for account settings and signature rules.</li>
	<li>Use a BM smear stained with the staining instrument.</li>
	<li>Enter the test identification manually or via barcode (QR code) scanner; the request from the laboratory information system is loaded automatically.</li>
	<li>Place the stained smears into the analyzer&#8217;s loading drawer.</li>
	<li>Close the analyzer input door.</li>
	<li>Process the smear. 
	NOTE: The analyzer automatically moves the smear under the microscope, scans it, and captures images of nucleated cells, including megakaryocytes. It performs a preliminary classification of all detected nucleated cells, including megakaryocytes, in only a few seconds.</li>
	<li>Review the preliminary results and reclassify where necessary, either directly on the analyzer or using remote review software. 
	NOTE: The system provides a clear view of the entire slide at 400&#215; magnification (whole-slide imaging) and displays all nucleated cells and megakaryocytes; nucleated cells (and possibly megakaryocytes) are captured at 1000&#215; magnification using an immersion objective. The average time for a complete scan of a normocellular smear is &#60;10 min per slide. For hypocellular smears, the scanning time is &#62;20 min per slide.</li>
	<li>In the event of incorrect classification, reassign the cell to the correct category by dragging its image with the mouse. 
	NOTE: For an example of BM cell misclassifications, see Figure 3. In cases of ambiguous or borderline cellular classifications, double reading is necessary. For cells with differing classifications, inter-observer consensus must be reached. Any manual reclassification by the operator or pathologist must be logged with the user ID and timestamp.</li>
	<li>Add comments and cytomorphological descriptions manually or by selecting from predefined text options.</li>
	<li>Complete the preliminary report.</li>
	<li>Confirm the final result, which is then automatically sent to the laboratory information system. 
	NOTE: The final result is exported as a numerical differential count (myelogram) along with additional comments and cytomorphological description.</li>
	<li>Log out and shut down the analyzer. 
	NOTE: A 500-cell count is applied for each BM sample, following ICSH recommendations17. 
	Classification categories (classes) include proerythroblast, early erythroblast (including megaloblastic), intermediate erythroblast (including megaloblastic), late erythroblast (including megaloblastic), myeloblast, promyelocyte, neutrophilic myelocyte, neutrophilic metamyelocyte, band neutrophil, segmented neutrophil, eosinophilic myelocyte, eosinophilic metamyelocyte, band eosinophil, segmented eosinophil, basophil, monoblast, promonocyte, monocyte, lymphoblast, prolymphocyte, mature lymphocyte (including atypical and reactive), plasmablast, immature plasma cell, plasma cell, and others (smudge cell, histiocyte, phagocyte, mast cell, and mitosis). 
	 
	CAUTION: Dispose of waste liquids, reagents, and consumables as medical infectious waste, and/or industrial waste in accordance with local laws and regulations. If possible, use a central waste disposal system.</li>
</ol>

<ol>
	<li>Walter W, et al. Artificial intelligence in hematological diagnostics: Game changer or gadget? Blood Rev. 2023;58:101019.</li>
	<li>Starostka D, et al. The utility of automated artificial intelligence-assisted digital cytomorphology for bone marrow analysis in diagnostic haemato-oncology. Clin Transl Med. 2025;15(7):e70364.</li>
	<li>Zini G, Barbagallo O, Scavone F, B&#233;n&#233; MC. Digital morphology in hematology diagnosis and education: The experience of the European LeukemiaNet WP10. Int J Lab Hematol. 2022;44(Suppl 1):37-44.</li>
	<li>Saft L, et al. A deep-learning algorithm (AIFORIA) for classification of hematopoietic cells in bone marrow aspirate smears based on nine cell classes &#8211; a feasible approach for routine screening? J Hematop. 2025;18(1):12.</li>
	<li>Lewandowski K, Kurpierz K, Sledzinska A. External bone marrow cytological examination quality assurance (EQAhem) &#8211; summary after 6 years in Poland. Ann Hematol. 2015;94(10):1741-1748.</li>
	<li>Sasada K, et al. Inter-observer variance and the need for standardization in the morphological classification of myelodysplastic syndrome. Leuk Res. 2018;69:54-59.</li>
	<li>Tohyama K. Present status and perspective of laboratory hematology in Japan: On the standardization of blood cell morphology including myelodysplasia: On behalf of the Japanese Society for Laboratory Hematology. Int J Lab Hematol. 2018;40(Suppl 1):120-125.</li>
	<li>Font P, et al. Interobserver variance in myelodysplastic syndromes with less than 5% bone marrow blasts: unilineage vs. multilineage dysplasia and reproducibility of the threshold of 2% blasts. Ann Hematol. 2015;94(4):565-573.</li>
	<li>DeLima M, et al. Comparison of referring and tertiary cancer center physician&#39;s diagnoses in patients with leukemia. Am J Med. 1998;104(3):246-251.</li>
	<li>Hanna MG, Parwani A, Sirintrapun SJ. Whole slide imaging: Technology and applications. Adv Anat Pathol. 2020;27(4):251-259.</li>
	<li>Matek C, et al. Highly accurate differentiation of bone marrow cell morphologies using deep neural networks on a large image data set. Blood. 2021;138(20):1917-1927.</li>
	<li>Lewis JE, Pozdnyakova O. Digital assessment of peripheral blood and bone marrow aspirate smears. Int J Lab Hematol. 2023;45(Suppl 2):50-58.</li>
	<li>Zini G, et al. Digital morphology compared to the optical microscope: A validation study on reporting bone marrow aspirates. Int J Lab Hematol. 2024;46(3):474-480.</li>
	<li>Starostka D, et al. Automated digital cytomorphology increases the reliability of bone marrow diagnostics. Transfuze Hematol Dnes. 2025;31(2):66-80.</li>
	<li>Briggs C, et al. Can automated blood film analysis replace the manual differential? An evaluation of the CellaVision DM96 automated image analysis system. Int J Lab Hematol. 2009;31(1):48-60.</li>
	<li>Ceelie H, Dinkelaar RB, van Gelder W. Examination of peripheral blood films using automated microscopy; evaluation of Diffmaster Octavia and Cellavision DM96. J Clin Pathol. 2007;60(1):72-79.</li>
	<li>Lee SH, et al. ICSH guidelines for the standardization of bone marrow specimens and reports. Int J Lab Hematol. 2008;30(5):349-364.</li>
	<li>Chicco D, Jurman G. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genomics. 2020;21(1):6.</li>
	<li>La Gioia A, et al. A prolonged microscopic observation improves detection of underpopulated cells in peripheral blood smears. Ann Hematol. 2017;96(10):1749-1754.</li>
	<li>Fu X, et al. Morphogo: An automatic bone marrow cell classification system on digital images analyzed by artificial intelligence. Acta Cytol. 2020;64(6):588-596.</li>
	<li>Lv Z, et al. High-accuracy morphological identification of bone marrow cells using deep learning-based Morphogo system. Sci Rep. 2023;13(1):13364.</li>
	<li>Katz BZ, et al. Evaluation of Scopio Labs X100 Full Field PBS: The first high-resolution full field viewing of peripheral blood specimens combined with artificial intelligence-based morphological analysis. Int J Lab Hematol. 2021;00:1-9.</li>
	<li>Bagg A, et al. Performance evaluation of a novel artificial intelligence-assisted digital microscopy system for the routine analysis of bone marrow aspirates. Mod Pathol. 2024;37(9):100542.</li>
	<li>Sirinukunwattana K, et al. Artificial intelligence-based morphological fingerprinting of megakaryocytes: A new tool for assessing disease in MPN patients. Blood Adv. 2020;4(14):3284-3294.</li>
	<li>Ryou H, et al. Quantitative interpretation of bone marrow biopsies in MPN &#8211; What&#39;s the point in a molecular age? Br J Haematol. 2023;203(4):523-535.</li>
	<li>Wang X, et al. The application of Morphogo in the detection of megakaryocytes from bone marrow digital images with convolutional neural networks. Technol Cancer Res Treat. 2023;22:15330338221150069.</li>
	<li>Kimura K, et al. A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA. Sci Rep. 2019;9(1):13385.</li>
	<li>Lin Y, Chen Q, Chen T. Recent advancements in machine learning for bone marrow cell morphology analysis. Front Med (Lausanne). 2024;11:1402768.</li>
	<li>Nazha A, et al. Artificial intelligence in hematology. Blood. 2025;146(19):2283-2292.</li>
	<li>Nakamura I, et al. Evaluation of two semi-supervised learning methods and their combination for automatic classification of bone marrow cells. Sci Rep. 2022;12:16736.</li>
	<li>Gl&#252;ge S, Balabanov S, Koelzer VH, Ott T. Evaluation of deep learning training strategies for the classification of bone marrow cell images. Comput Methods Programs Biomed. 2024;243:107924.</li>
	<li>Eckardt JN, et al. Synthetic bone marrow images enhance real samples in the development of microscopy classification models for acute myeloid leukemia. npj Digit Med. 2025;8,173.</li>
	<li>Deltadahl S, et al. Deep generative classification of blood cell morphology. Nat Mach Intell. 2025;7:1791-1803.</li>
</ol>

The authors declare no conflict of interest.

Technological advancements in high-resolution image processing, automated digital image classification, and AI-supported decision-making using convolutional neural networks have enabled substantial improvements in ADM. As a result, ADM complements standard optical microscopy. The correct identification of individual cytomorphological categories can be challenging, particularly among closely related classification groups that share morphological similarities2,11,12,14,15,19,20. As expert findings are always influenced by a degree of subjectivity, they may not be considered definitive or the only possible correct results&#8212;especially in cases of borderline or ambiguous cellular classification (e.g., dysplastic or highly atypical elements)2. These misclassification patterns have been reported in PB, showing stronger performance for common cell types but weaker classification consistency for some clinically important subsets, including blasts12,15,19,20. This may significantly affect PB and BM cytomorphological diagnosis.
Compared with PB, only a few studies have focused on the classification accuracy of ADM in BM analysis2,20,21,22,23. In a Chinese study on ADM classification, consistency reached 99%, with most BM cell types showing extremely high correlation with hematopathological expertise21. However, the proportion of hematological malignancies in the study cohort was less than 30%21. Other comparative studies evaluating performance between automated PB and BM smear systems and manual analysis have been conducted in multicenter trials22,23. In a recent comparison, 795 BM specimens from patients with neoplastic and other clinical conditions were analyzed, achieving an overall agreement of 91.1% with optical microscopy23.
A study of a real-world cohort of European ancestry patients, with 59% neoplastic conditions, compared AI-driven ADM with expert optical microscopy across six diagnostic groups2. Critical misclassifications of neoplastic cells with atypical morphology had misdiagnosis potential2. Therefore, this study identified candidate cell types for future training and testing of AI performance2. However, this study had several limitations. First, the study included a limited number of patients and archived material, and some rare BM disorders were absent, including pediatric samples. Second, the cytomorphology of certain neoplastic cells was very abnormal, with questionable consensus in classification even among experienced experts. Third, only samples of European ancestry were analyzed. Fourth, no external validation of relevant and irrelevant misclassifications was performed2.
Cellular misclassifications represent the major limitation of ADM in diagnostic hemato-oncology2. There are several reasons for misclassification. In addition to cytomorphological overlap, contributing factors likely include the rarity of certain conditions and variability in expert consensus regarding cell types. This may lead to inconsistent annotations and gaps in training data. The detection of dysplastic changes in hematopoiesis is unreliable, as is the detailed evaluation of megakaryocyte cytomorphology2. Although adverse outcomes are largely mitigated by mandatory expert supervision, limitations remain regarding the absolute reliability of ADM, particularly in leukemias, lymphomas, and multiple myeloma2. Correct expert annotation of cells for AI training is an essential prerequisite for ADM performance. Accurate data labeling is fundamental to building reliable ground-truth datasets. In cytomorphology, defining ground truth can be particularly difficult in ambiguous or borderline cases, as a degree of subjectivity persists even among experienced, widely recognized experts. When expert opinions differ, ground truth may be established through majority consensus or agreement between two or more independent reviewers of comparable expertise, particularly when diagnostic judgments are broken down into separate components2. Refinement of classification capabilities focused on specific candidate cell types is essential. It includes targeted retraining on difficult cell types, hierarchical classification strategies, or the incorporation of multimodal data to improve discrimination between morphologically similar cells. Extensive, diverse, and well-annotated training datasets, especially in hematolymphoid neoplasms, must include sufficient representation of atypical morphology, rare entities, borderline cases, and diagnostically ambiguous cells.
Regarding critical protocol steps and troubleshooting, PB smears must be made within two hours of collection; the BM aspirate must be spread immediately from the first pull. A validated, automated staining protocol kept strictly consistent between batches is required. A well-defined, artifact-free monolayer is mandatory. Prolonged scanning is required in leukopenic PB samples and hypocellular BM smears. Depleted immersion oil or thin smears trigger autofocus failure and abort analysis. Neither system is approved for autonomous reporting, and an individual cell image gallery inspection is mandatory. Local clinical validation is mandatory prior to deployment; full diagnostic and legal responsibility rests with the reporting morphologist.
Future rigorous multi-institutional validation across different clinical settings is essential for model generalizability and predictive performance. An AI-based approach to megakaryocyte cytomorphology may further refine the integrated diagnosis of myeloproliferative and myelodysplastic neoplasms24,25,26. A substantial gap also remains in the development of reliable AI-supported recognition of lineage-specific dysplastic changes2,27. For the model optimization, innovative techniques such as hyperparameter tuning, regularization, ensemble learning, and data augmentation may serve to improve robustness and classification reliability28,29,30. In particular, data augmentation strategies &#8211; ranging from geometric transformations30,31 to the generation of synthetic training images using generative AI32,33 &#8211; have recently emerged as promising approaches to expand limited datasets and address class imbalance in BM cell classification. The level of education and long-term experience of laboratory staff are essential for the evaluation and decision-making process in difficult cytomorphological cases. Training the next generation of cytomorphologists in the context of ongoing digitization presents an educational challenge. On the one hand, ADM offers tremendous educational potential; on the other hand, it carries the risk that less experienced operators may overestimate its capabilities.
Despite the promising diagnostic potential of ADM systems, several critical limitations and implementation challenges must be acknowledged. A fundamental concern relates to bias in training data: most commercially available ADM platforms, including those evaluated in the present study, were developed using datasets derived predominantly from specific geographic regions, staining protocols, and patient populations, which may limit their generalizability to more ethnically and clinically diverse cohorts or to laboratories employing different preparation techniques. Equally important is the issue of transparency in AI decision-making. Current ADM systems largely operate as black-box models, which hampers the ability of clinicians to critically evaluate individual cell assignments. This is directly linked to the question of clinical responsibility, as regulatory frameworks for AI-assisted diagnostics continue to evolve. The practical integration of ADM into standard diagnostic workflows presents substantial organizational and technical barriers, including the need for high-quality slide digitization infrastructure, staff training, validation against local reference standards, and the establishment of quality control protocols to monitor system performance over time. Thus, expert-driven refinement and exploration of ADM&#8217;s full potential in PB and BM assessment will continue to improve its utility as a supportive tool in routine laboratory practice.

The correct cytomorphological assessment of peripheral blood (PB) and bone marrow (BM) smears remains a cornerstone of multidisciplinary diagnostics in hemato-oncology, influencing critical clinical decision points1,2,3,4. There is evidence of substantial inter-expert and intra-expert variability in manual PB and BM smear analysis, with downstream effects on patient diagnosis2,5,6,7,8,9. Recent advances in image analysis, machine learning, and computational pathology have enabled automated assessment of PB and BM smears at a level that may complement expert cytomorphologic review1,2,3,4,10,11,12,13,14. Artificial intelligence (AI)-powered automated digital cytomorphology (ADM) of PB and BM can support labor-intensive and time-consuming optical microscopy and expert decision making1,2,3,4,11,12.
The core elements of ADM technology include standardized smear preparation and staining, a scanner with advanced imaging capabilities, AI, and an efficient, user-friendly interface. The precision mechanics of ADM operate with micrometer-level accuracy at high speed, while advanced imaging solutions ensure high image quality14. A user-friendly interface makes this sophisticated technology accessible in an intuitive and streamlined manner14.
ADM has long been used for PB assessments, where it is highly developed and has become an integral part of standard laboratory workflows. In BM assessments, however, ADM presents challenges due to complex and intricate cytomorphological patterns, a high number of distinct cell types, and the morphological similarity of precursor cells2. The AI classification capability is based on training with large, comprehensive image datasets containing up to several million cells. All cells included in these training datasets are reviewed and annotated by experts. This expert input, among other factors, has a direct impact on classification quality.
Nevertheless, certain PB and BM cells remain particularly challenging for ADM to classify, which can significantly affect cytomorphological diagnosis2,12,15. Critical cellular misclassifications remain the major limitation of ADM and may have serious clinical consequences, thereby undermining the reliability of the method2.
Our objective is to present a practical ADM workflow, including its specific steps and the expert review process. Detailed reporting of the ADM process using two analytical platforms as examples will enable users to fully exploit the capabilities and benefits of the method. The high classification accuracy of ADM for PB has been demonstrated in numerous studies12,15,16. Therefore, in our research, we focused exclusively on ADM for BM, where reliable validation data of this type are limited. The quantitative validation results are based on BM analysis. Furthermore, we emphasize a key limitation of the method, namely, critical misclassifications. We also propose a categorization of misclassifications and highlight the most common types of critical misclassifications.

<table><tbody><tr><td>Automated Hematology Slide Preparation Unit SP-50</td><td>Sysmex Corporation, Kobe, Japan</td><td>N/A</td><td>Automated slide maker and stainer</td></tr><tr><td>Barcode scanner for Sysmex SP-50</td><td>Datalogic Gryphon</td><td>S/N: G21M20996</td><td>Barcode scanner</td></tr><tr><td>CellaVision DC-1</td><td>CellaVision AB, Lund, Sweden</td><td>PM-10895-15</td><td>Automated digital morphology analyzer</td></tr><tr><td>CellaVision DM Software</td><td>CellaVision AB, Lund, Sweden</td><td>Version 7.1.1 build 8</td><td></td></tr><tr><td>Giemsa Solution for SP Automated Systems</td><td>RAL Diagnostics, Martillac, France</td><td>750305X1000</td><td>Component of panoptic staining for peripheral blood and bone marrow smears</td></tr><tr><td>Immersion Oil (CellaVision Immersion Oil 50 mL)</td><td>CellaVision AB, Lund, Sweden</td><td>XU-10319</td><td>Immersion oil</td></tr><tr><td>May-Gr&amp;uuml;nwald for SP Automated Systems</td><td>RAL Diagnostics, Martillac, France</td><td>750105X2500</td><td>The component of panoptic staining for peripheral blood and bone marrow smears</td></tr><tr><td>Morphogo Bone Marrow Analyzer</td><td>Hangzhou ZhiWei Information Technology Co. Ltd., Hangzhou, China</td><td>N/A</td><td>Automated digital bone marrow morphology analyzer</td></tr><tr><td>Morphogo Review Software</td><td>Hangzhou ZhiWei Information Technology Co. Ltd., Hangzhou, China</td><td>Version 1.0.6</td><td>Bone marrow review software</td></tr><tr><td>pH 7.0 Buffer Solution for SP Automated Systems</td><td>RAL Diagnostics, Martillac, France</td><td>750505X5000</td><td>Buffer solution</td></tr><tr><td>QR scanner for CellaVision DC-1</td><td>Datalogic Gryphon</td><td>GD 4500</td><td>QR scanner</td></tr><tr><td>RAL Cleaning solution</td><td>RAL Diagnostics, Martillac, France</td><td>750725X5000</td><td>Cleaning solution</td></tr><tr><td>S-Monovette K3 EDTA tube</td><td>Sarstedt</td><td>05.1167.001</td><td>Tube for SP-50</td></tr><tr><td>SP-Rinse Solution</td><td>RAL Diagnostics, Martillac, France</td><td>37000305</td><td>Rinse solution</td></tr><tr><td>SP Slides</td><td>Sysmex Corporation, Kobe, Japan</td><td>ZE 001906</td><td>Slides&amp;nbsp; for peripheral blood and bone marrow smears</td></tr><tr><td>SP Slides</td><td>Sysmex Corporation, Kobe, Japan</td><td>ZE 001906</td><td>Slides&amp;nbsp; for QC</td></tr><tr><td>SP Spreader glass</td><td>Sysmex Corporation, Kobe, Japan</td><td>95102918</td><td>Glass for spreading the smear&amp;nbsp;</td></tr><tr><td>Tube rack for Sysmex SP-50</td><td>Sysmex Corporation, Kobe, Japan</td><td>AU330213</td><td>Tube rack</td></tr></tbody></table>

workflow for automated digital peripheral blood and bone marrow cytomorphology, with consideration of critical cellular misclassifications

Workflow for Automated Digital Peripheral Blood and Bone Marrow Cytomorphology, With Consideration of Critical Cellular Misclassifications

Growing evidence supports the ability of AI-powered automated digital cytomorphology (ADM) of peripheral blood (PB) and bone marrow (BM) to assist expert decision-making. The ADM workflow for PB and BM smears is described. The core components of ADM technology include precision mechanics, advanced imaging systems, artificial intelligence (AI), and a user-friendly interface. The process begins with standardized smear preparation and staining using protocols adapted for automation and digitization. An automated microscope equipped with an immersion objective and digital camera captures high-resolution images of the specimen at multiple magnifications. AI-powered software then locates, records, and preliminarily classifies nucleated cells, including megakaryocytes when applicable. This is followed by mandatory expert review and, as needed, reclassification. Digital images and reports are subsequently stored on a high-capacity local server. Key benefits of ADM include increased speed, efficient data management, educational support, remote access for evaluation and consultation, digital archiving, and multimodal data integration. The most significant limitation of ADM is the risk of critical cellular misclassification, which can have a major clinical impact. Irrelevant misclassifications are tolerable and diagnostically neutral, whereas relevant misclassifications are unacceptable and may lead to serious diagnostic and clinical consequences. Critical misclassification of neoplastic lymphocytes, small lymphoblasts/myeloblasts, granular monoblasts, dysplastic promonocytes and monocytes, plasmablasts, and immature plasma cells&#8212;all of which may present with atypical morphology&#8212;carries a risk of misdiagnosis. Further advancements are required to achieve reliable diagnostic performance. Extensive, high-quality training data, combined with regular validation, are essential to ensure that AI functions as an effective supporting tool for diagnostic judgment.

Validation study &#8211; methods 
In this validation study, we reproduce the results of our previous work published in Clinical and Translational Medicine2. Conventional optical microscopy and ADM were performed on 328 BM smears. Data for analysis included image information for all analyzed cells and classification results, including rare and atypical morphological cases. The cases were divided into six diagnostic groups: myelodysplastic neoplasms (MDN: 15%), multiple myeloma (MM: 14%), mature B/T-cell neoplasms (B/T-lymphoma: 13%), acute leukemia and chronic myelomonocytic leukemia (AL+CMML: 9%), myeloproliferative neoplasms (MPN: 8%), and reactive hematopoiesis (reactive: 41%). ADM cell recognition (classification) capabilities were evaluated by assessing its ability to correctly classify cells into one of 25 categories, compared to independent consensual expert annotation with double reading2.
Each case was reviewed by two experts, with matching classification (annotation) in &#62;99% of cells. The experts were very experienced, regularly participating in successful external quality evaluations. For cells with differing classifications (&#60;1%), inter-observer consensus was reached. The principle of the consistency analysis was a comparison of the automatic pre-classification of each cell with the final expert classification of each cell; the consensual expert classification was considered the true classification reference (ground truth)2.
Cellular and clinical classification consistency was assessed, and critical misclassifications were identified. Cellular classification consistency was determined from the confusion matrix, based on classification agreement for individual cell types. Standard statistical methods and visual data mining were applied. The Matthews correlation coefficient (MCC) was calculated for each cell type and as an average value, and was interpreted in a standard manner. An MCC value of 0.400 and above was considered satisfactory2,18.
For each patient, clinical classification consistency was determined as the percentage of correctly pre-classified elements among all cells evaluated in the case, after elimination of unclassifiable cells. The expert classification was considered the true reference (ground truth)2.
Misclassifications are considered irrelevant when they are diagnostically neutral and do not affect the final diagnosis. These include reciprocal misclassifications between lymphoblast, myeloblast, and monoblast; neutrophilic, eosinophilic, and basophilic promyelocyte/myelocyte; myelocyte/metamyelocyte; metamyelocyte/band; band/segmented neutrophil; proerythroblast/early erythroblast; early/intermediate and intermediate/late erythroblast; promonocyte/monocyte; prolymphocyte/lymphocyte; plasmablast/immature plasma cell; and immature plasma cell/mature plasma cell. All other misclassifications are considered relevant, as they are diagnostically unacceptable and may lead to serious clinical consequences (Figure 4)2.
Overall clinical classification consistency was calculated as the median of individual case values, considering only relevant cell misclassifications. Critical misclassification was defined as a case in which the individual clinical classification consistency value was below 80%. This arbitrary limit was set by expert consensus, considering minimal requirements of external quality assessment in cytomorphology rounds2.
Key platform-specific differences include a two-step BM workflow (40&#215; and 100&#215; objective), bringing additional complexity absent from the PB protocol, distinct cell count targets (100 cells in PB, 500 cells in BM), and varying classification categories. The single most critical shared determinant of inter-platform and inter-sample-type comparability is the smear preparation procedure.
The quantitative performance metrics reported are clearly linked to the validation cohort described in the Methods (n = 328 BM smears). The percentage of correctly classified relevant cells out of all classified cells (cellular classification consistency) was 95.4%. Satisfactory correlation values, as indicated by the MCC, were &#8805;0.400 for 22 of 25 (88.0%) cell types, while unsatisfactory MCC values (&#60;0.400) were observed for 3 of 25 (12.0%) cell types (lymphoblasts, prolymphocytes, and promonocytes). The overall relevant clinical consistency was 97.1% (median), with 94.5% of cases showing consistency rates between 80% and 100%. In 5.5% of patients, critical misclassification was observed, with relevant clinical consistency values ranging from 36% to 79%, due to failure to recognize neoplastic cells2.
Cellular misclassification compromises the reliability of ADM in BM analysis and represents a major limitation of the method in diagnostic hemato-oncology (Table 1). In cases involving relevant misclassification, the morphology of correctly and incorrectly classified cells was examined. Correctly classified cells generally display typical cytomorphological features. In contrast, misclassifications most often occur when distinguishing between morphologically similar or atypical cells, particularly atypical lymphocytes and blasts; myeloblasts and lymphocytes; lymphoblasts and lymphocytes; monoblasts or promonocytes and promyelocytes; and immature plasma cells and blasts. Misclassified cells frequently exhibit atypical cytomorphology. In lymphocytic neoplasms, misclassified neoplastic lymphocytes, such as those seen in marginal zone lymphoma and hairy cell leukemia, tend to be medium-sized, with abundant basophilic or pale cytoplasm, cytoplasmic projections, and finer chromatin with a nucleolus. These features differ from those of correctly classified neoplastic cells. In mantle cell lymphoma, atypical lymphocytes may show a blastoid appearance, increasing the likelihood of misclassification. Similarly, in the pleomorphic variant of chronic lymphocytic leukemia, neoplastic lymphocytes may be misidentified as blasts. We demonstrate qualitative representative examples in Figure 4 and Figure 5.
In acute leukemia and chronic myelomonocytic leukemia (CMML), misclassified lymphoblasts and myeloblasts are small, with a high nuclear-to-cytoplasmic ratio, coarser chromatin, and nucleoli. Numerous granular monoblasts may be misclassified as promyelocytes. In CMML, misclassification is frequently associated with dysplastic monocytic cells. In multiple myeloma, misclassifications occur within the plasma cell lineage, particularly involving plasmablasts and immature plasma cells. We demonstrate qualitative representative examples in Figure 5.
Critical misclassifications of neoplastic lymphocytes, small lymphoblasts/myeloblasts, granular monoblasts, dysplastic promonocytes and monocytes, plasmablasts, and immature plasma cells&#8212;all of which may show atypical morphology&#8212;pose a risk of misdiagnosis. Mistaking lymphoma for acute leukemia and vice versa holds potential for serious or even fatal consequences for patient management. False-negative results, such as failure to recognize acute myeloid leukemia (particularly acute promyelocytic leukemia) or misdiagnosis of CMML or multiple myeloma, may potentially lead to inappropriate treatment decisions or delays in therapy.
<img alt="Figure 1" class="xfigimg" src="/files/ftp_upload/71690/71690fig1.jpg" /> 
Figure 1. Automated stainer, home screen icons. Courtesy of Sysmex; manufacturer&#8217;s information materials (basic operation manual). <a href="https://www.jove.com/files/ftp_upload/71690/71690fig1large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 2" class="xfigimg" src="/files/ftp_upload/71690/71690fig2.jpg" /> 
Figure 2. Misclassifications of PB white blood cells. In screenshots of the software, misclassified cells are highlighted in red boxes. In EBV infection, reactive lymphocytes were misclassified as monocytes (A, red box). In chronic lymphocytic leukemia, some lymphocytes were misclassified as blasts (B, red box). In CMML, dysplastic monocytes were misclassified as segmented neutrophils (C, red box), and a monoblast was misclassified as a dysplastic myelocyte (D, red box). PB, CellaVision DC-1, Review Software, automated staining May&#8211;Gr&#252;nwald and Giemsa&#8211;Romanowski with the Sysmex SP-50 stainer, magnification 1000&#215;, expert: certified technician. <a href="https://www.jove.com/files/ftp_upload/71690/71690fig2large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 3" class="xfigimg" src="/files/ftp_upload/71690/71690fig3.jpg" /> 
Figure 3. Misclassification of neoplastic lymphocytes in BM. Screenshot of the software. Neoplastic lymphocytes in high-grade B-cell lymphoma in the BM were misclassified as myeloblasts. The proposed alternative classification with the five most likely options (the &#8220;top five list&#8221;), including percentage probabilities, is marked with a yellow arrow. For the cell in the yellow box, the label shows percentage probabilities for myeloblast, monoblast, lymphoblast, proerythroblast, and early erythroblast, with no option for lymphocyte. BM, Morphogo, Review Software, automated staining May&#8211;Gr&#252;nwald and Giemsa&#8211;Romanowski with the Sysmex SP-50 stainer, magnification 1000&#215;, expert: certified scientist, certified physician. <a href="https://www.jove.com/files/ftp_upload/71690/71690fig3large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 4" class="xfigimg" src="/files/ftp_upload/71690/71690fig4.jpg" /> 
Figure 4. Irrelevant and relevant cell misclassifications. (A) Examples of irrelevant misclassifications in myeloid and erythroid lineages are shown in a comparison between expert classification and ADM pre-classification. (B) Examples of relevant misclassifications in lymphoid, plasma cell, and monocytic lineages are shown in a comparison between expert classification and ADM pre-classification. The categorization of some elements was difficult even for an expert. Cell images were extracted from the software. BM, Morphogo, Review Software, automated staining May&#8211;Gr&#252;nwald and Giemsa&#8211;Romanowski with the Sysmex SP-50 stainer, magnification 1000&#215;, expert: certified scientist, certified physician. A scale bar was added to each cell. Scale bars represent 10 &#181;m. Calibration based on ToupCam camera (1.85 &#181;m pixel size) with 100&#215; Olympus Plan N objective. 
Modified from: Supplementary Material, Starostka D, Dolezilek R, Kvasnicka HM, Kudelka M, Miczkova P, Kriegova E, et al. The utility of automated artificial intelligence-assisted digital cytomorphology for bone marrow analysis in diagnostic haemato-oncology. Clin Transl Med. 2025; 15:e70364. https://doi.org/10.1002/ctm2.70364.&#160;<a href="https://www.jove.com/files/ftp_upload/71690/71690fig4large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 5" class="xfigimg" src="/files/ftp_upload/71690/71690fig5.jpg" /> 
Figure 5. Representative examples of the most frequent misclassifications of BM cells in patients with hematological neoplasms. Misclassified neoplastic lymphocytes, myeloblasts, lymphoblasts, monoblasts, and plasma cells are shown in selected cases of marginal zone lymphoma (MZL), mantle cell lymphoma (MCL), hairy cell leukemia (HCL), chronic lymphocytic leukemia (CLL), acute myeloid leukemia (AML), acute T-lymphoblastic leukemia (T-ALL), and multiple myeloma (MM). The misclassification of atypical lymphocyte/blast was the most frequent error. The main cytomorphological features of misclassified neoplastic lymphocytes included medium size, finer chromatin (red arrows), nucleoli (blue arrows), abundant basophilic or pale cytoplasm with projections (green arrows), and blastoid appearance (orange arrow). Misclassified lymphoblasts and myeloblasts were small, with a high nuclear-to-cytoplasmic ratio and coarser chromatin with nucleoli (grey arrows). Plasmablasts and immature plasma cells were misclassified as lymphocytes or were not recognized as belonging to the plasma cell lineage (classified as &#8220;blast, unspecified&#8221;). Cell images were extracted from the software. BM Morphogo Review Software, automated staining May&#8211;Gr&#252;nwald and Giemsa&#8211;Romanowski with the Sysmex SP-50 stainer, magnification 1000&#215;, expert: certified scientist, certified physician. A scale bar was added to each cell. Scale bars represent 10 &#181;m. Calibration based on ToupCam camera (1.85 &#181;m pixel size) with 100&#215; Olympus Plan N objective. 
The image was modified from: Supplementary Material, Starostka D, Dolezilek R, Kvasnicka HM, Kudelka M, Miczkova P, Kriegova E, et al. The utility of automated artificial intelligence-assisted digital cytomorphology for bone marrow analysis in diagnostic haemato-oncology. Clin Transl Med. 2025; 15:e70364. https://doi.org/10.1002/ctm2.70364.&#160;<a href="https://www.jove.com/files/ftp_upload/71690/71690fig5large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always">
	<tbody>
		<tr>
			<td>Diagnosis</td>
			<td>Number of cases</td>
			<td colspan="2">Misclassified cells</td>
		</tr>
		<tr>
			<td></td>
			<td></td>
			<td>Expert classification</td>
			<td>ADM classification</td>
		</tr>
		<tr>
			<td>Mature B-cell neoplasms</td>
			<td>9/43</td>
			<td></td>
			<td></td>
		</tr>
		<tr>
			<td>Marginal zone lymphoma</td>
			<td>4/8</td>
			<td>Atypical lymphocyte</td>
			<td>Blast, NOC</td>
		</tr>
		<tr>
			<td>Mantle cell lymphoma</td>
			<td>2/5</td>
			<td>Atypical lymphocyte</td>
			<td>Blast, NOC</td>
		</tr>
		<tr>
			<td>CLL</td>
			<td>1/14</td>
			<td>Atypical lymphocyte</td>
			<td>Blast, NOC</td>
		</tr>
		<tr>
			<td>Hairy cell leukaemia</td>
			<td>2/2</td>
			<td>Atypical lymphocyte</td>
			<td>Blast, NOC / Monocyte</td>
		</tr>
		<tr>
			<td>Multiple myeloma</td>
			<td>5/46</td>
			<td>Plasma cell</td>
			<td>Blast, NOC</td>
		</tr>
		<tr>
			<td>AML</td>
			<td>3/16</td>
			<td>Myeloblast / Monoblast</td>
			<td>Lymphocyte / Promyelocyte</td>
		</tr>
		<tr>
			<td>T-ALL</td>
			<td>1/1</td>
			<td>Lymphoblast</td>
			<td>Lymphocyte</td>
		</tr>
	</tbody>
</table>
Table 1:&#160;Critical misclassifications by ADM. In 18/328 (5.5%) of patients, critical misclassification was observed. The clinical group of critical misclassifications by ADM included nine out of 43 patients with mature B-cell neoplasms, five out of 46 patients with MM, three out of 16 patients with AML, and one (out of one) patient with T-ALL. The misclassification of atypical lymphocytes/blasts was the most frequent. NOC: no otherwise classified. 
Reproduced from: Supplementary Material, Starostka D, Dolezilek R, Kvasnicka HM, Kudelka M, Miczkova P, Kriegova E, et al. The utility of automated artificial intelligence-assisted digital cytomorphology for bone marrow analysis in diagnostic haemato-oncology. Clin Transl Med. 2025; 15:e70364. https://doi.org/10.1002/ctm2.70364.

The workflow of automated digital cytomorphology of peripheral blood and bone marrow is described. After standardized smear preparation and staining, a scanner captures specimen images. AI-powered software locates, records, and preliminarily classifies nucleated cells, followed by expert review. The focus is on critical cellular misclassifications with potential for misdiagnosis.

Growing evidence supports the ability of AI-powered automated digital cytomorphology (ADM) of peripheral blood (PB) and bone marrow (BM) to assist expert ...

workflow-for-automated-digital-peripheral-blood-bone-marrow

Research

JoVE Journal

Medicine

This video describes the technique of aspirating bone marrow from iliac spine and smear preparation for biopsy. In the example, we perform bone marrow aspiration from an acute myeloid leukemia (AML) patient and examine the cells to check for measurable residual disease (MRD), an important prognostic biomarker.

Bone Marrow Aspiration: A Method to Obtain Bone Marrow to Examine Cell Morphology

A strategy to quantitatively analyze histological data in the bone marrow is presented. Confocal microscopy of fluorescently labeled cells in tissue sections results in 2-dimensional images, which are automatically analyzed. Co-localization analyses of different cell types are compared to data from simulated images, giving quantitative information about cellular interactions.

Automated Quantification of Hematopoietic Cell &#8211; Stromal Cell Interactions in Histological Images of Undecalcified Bone

Detection of minimal or measurable residual disease (MRD) is an important prognostic biomarker for refining risk assessment and predicting relapse in acute myeloid leukemia (AML). These comprehensive guidelines and recommendations with best practices for consistent and accurate identification and detection of MRD, may aid in making effective AML treatment decisions.

Comprehensive Protocol to Sample and Process Bone Marrow for Measuring Measurable Residual Disease and Leukemic Stem Cells in Acute Myeloid Leukemia

The MDS diagnosis is difficult in the absence of morphological criteria or non-informative cytogenetics. MFC could help refine the MDS diagnostic process. To become useful for clinical practice, the MFC analysis must be based on parameters with sufficient specificity and sensitivity, and data should be reproducible between different operators.

Database-guided Flow-cytometry for Evaluation of Bone Marrow Myeloid Cell Maturation

We describe the use of hematopoietic stem cell transplantation (HSCT) to assess the malignant potential of genetically engineered hematopoietic cells. HSCT is useful for evaluating various malignant hematopoietic cells in vivo as well as generating a large cohort of mice with myelodysplastic syndromes (MDS) or leukemia to evaluate novel therapies.

Use of Hematopoietic Stem Cell Transplantation to Assess the Origin of Myelodysplastic Syndrome

Here, we present a protocol to process fresh bone marrow (BM) isolated from mouse or human for high-dimensional mass cytometry (Cytometry by Time-Of-Flight, CyTOF) analysis of neutrophil-lineage cells.

Preparation of Whole Bone Marrow for Mass Cytometry Analysis of Neutrophil-lineage Cells

Here, we describe a simple protocol for the isolation and staining of murine bone marrow cells to phenotype hemopoietic stem and progenitor cells along with the supporting niche endothelial and mesenchymal stem cells. A method to enrich cells located in endosteal and central bone marrow areas is also included.

Flow Cytometry Analysis of Murine Bone Marrow Hematopoietic Stem and Progenitor Cells and Stromal Niche Cells

The success of single-cell/single-nuclei transcriptomics and multi-omics largely depends on the quality of cells/nuclei. Therefore, isolating cells/nuclei from tissue and their purification must be highly standardized. This protocol describes the preparation of nuclei from the brain and bone marrow for downstream single-nuclei multiome assay.

High-Quality Brain and Bone Marrow Nuclei Preparation for Single Nuclei Multiome Assays

Here, a phosphoflow cytometry-based method is described to analyze signaling downstream of the mTORC1, JAK/STAT5, and MAPK pathways in acute human myeloid leukemia cells xenografted into mice and obtained from bone marrow aspirates. p-STAT5, p-4EBP1, p-RPS6, and p-ERK1/2 levels are simultaneously measured using a next-generation spectral flow cytometer with high sensitivity.

Intracellular Phosphoflow Cytometry of Acute Myeloid Leukemia Patient-Derived Xenotransplants

This protocol outlines a procedure for inducing differentiation of human hematopoietic stem and progenitor cells derived from patient bone marrow and peripheral blood mononuclear cells. The assay uses a semisolid culture medium supporting erythroid and myeloid lineage commitment and incorporates standardized metrics for reproducible colony classification for preclinical drug characterization.

Colony-Forming Unit Assay Adapted For Patient-Derived Bone Marrow and Peripheral Blood Mononuclear Cells

Moravian-Silesian Hospital Havirov.  The workflow of automated digital cytomorphology of peripheral blood and bone marrow is described. After standardized smear preparation and staining, a scanner captures specimen images. AI-powered software locates, records, and preliminarily classifies nucleated cells, followed by expert review. The focus is on critical cellular misclassifications with potential for misdiagnosis.

Workflow for Automated Digital Peripheral Blood and Bone Marrow Cytomorphology, With Consideration of Critical Cellular Misclassifications

Explore More Videos