Blood: Essential body fluid & essential diagnostic tool

Blood is an essential body fluid consisting of many cells that perform critical functions to combat disease. The key to disease detection and disease severity lies in the identification of blood cells, as well as counting, analyzing, and comparing of different types of blood cells.  ^{1 Islam} Typically, a person’s health status is determined by analyzing different shape size context and color features of blood cells (morphology) and their corresponding cell counts. ^{2 Desphande}

Accurate blood cell classification is essential for the diagnosis and follow up of many medical conditions including infectious diseases, blood cancers (hematological malignancies), inflammatory conditions, and nutritional deficiencies. ^{1 Islam} In fact, blood cell analysis is one of the most accurate diagnostic methods in medicine. ^{2 Deshpande}

Complexity and Demand for hematological expertise

The field of hematology has become increasingly more challenging and complex as a  comprehensive diagnosis and personalized treatment plan are needed for numerous hematological conditions including life-threatening diseases such as leukemia, lymphoma, myelodysplastic syndrome (MDS), multiple myeloma (MM), and myeloproliferative neoplasms.^{2 Deshpande , 3  Wang} Despite advances in new treatment options, which include targeted therapies, immunotherapy, and hematopoietic stem cell transplantation, patients may still experience poor outcomes, relapse, and further complications. ^{3 Wang} Blood morphological assessment, including white blood cell classification and analysis, is part of a structured and thorough diagnostic approach, placing significant demands on the hematopathologist’s expertise, methodology of disease detection, and wealth of knowledge. ^{3 Wang} Conventional methods used for morphological blood cell analysis and interpretation can be lengthy and susceptible to human error, therefore not providing the precision that is needed for early disease detection or ongoing disease management. ^{1 Islam}  

From manual cell counting and optical microscopy to AI and digital cell morphology

Overtime diagnostic morphological classification and assessment have evolved manual cell counting to the development of automated hematological systems. Automated complete blood counting came to support microscopic examination in diagnosing hematological disorders. Morphological analysis, which is an essential tool in predicting disease outcomes and guiding treatment decisions, was heavily reliant on optical microscopy. One of the most indispensable tools in the hematology laboratory remains the peripheral blood smear (PBS). ^{4 Fan} PBS requires detailed examination of blood cell morphology, providing key information on the shape, size, and number of blood cells to enable diagnostic accuracy of various hematological disorders. ^{3 Wang}

While PBS examinations under light microscopy remain the cornerstone of many laboratories, they are known for their limitations. The process is tedious and time-consuming, and accuracy is affected by high inter-and intra-observer variability and potential for diagnostic errors, all of which negatively influence the diagnostic value of morphology. ^{3 Wang}

The new era of digital morphology

The integration of digital microscopy combined with artificial intelligence has transformed hematology and introduced the concept of digital morphology in hematology. While digital microscopy has emerged as a powerful tool for rapid access and sharing of blood cell images, Artificial Intelligence (AI) enhances digital morphological analysis, improving diagnostic efficiency and accuracy. The process of optical and cerebral connection allowing for cell recognition now takes place on the digital screen.  ^{5 Zini}  In addition, fully digital morphological analysis can assist with completely remote digital access and collaboration.

Modern AI systems, particularly deep learning models are trained on large, accurately annotated datasets. The process of preparing a dataset involves scanning digital slides to acquire images, de-identifying images, review of image quality and resolution, storing images digitally and annotating or validating relevant cells and regions of interest. AI models then undergo training, validation, and testing allowing for advanced image analysis including segmentation, detection, and classification. Machine learning (ML) facilitates the recognition of distinct morphological characteristics of blood cells, improving diagnostic precision. ^{3 Wang} This includes to name just a few:  identifying dysplastic cell features, identifying blast cells, both in character and number, red blood cell morphological abnormalities and malaria detection.  ^{2 Deshpande, 5 Zini}  Thus modern AI hematology systems are able to provide laboratory professionals with a highly efficiency means of performing consistent WBC differentials, RBC morphology assessments and platelet estimations.

In 2019, the International Council for Standardization in Hematology (ICSH) recommended the application of digital imaging technology. highlighting the advantages of using AI algorithms to pre-classify cells. The combination of artificial intelligence (AI) and high throughput digital slide scanners has enabled laboratory medical technologists and hematologists to rapidly review large numbers of films, while maintaining a high diagnostic accuracy, to facilitate an effective clinical diagnostic pathway. ^{5 Zini, 6 Kratz} While AI has the potential to markedly enhance the precision and standardization of hematologic diagnosis, it is important to note that AI will not replace the skills and experience of the clinician and hematopathologist, particularly for complex cases, rather enhance them and allow them to focus on challenging diagnoses.

The conundrum: Expanding diagnostic demand and shrinking skilled laboratory workforce

A significant challenge in the United States healthcare system is the expanded demand for diagnostics coupled with a paradoxically shrinking laboratory professional workforce. A cross-sectional study published in 2019, comparing a decade of data showed a worrying decline in the US pathologist workforce coupled with an increased diagnostic workload. ^{7 Metter}

Between 2007 and 2017: ^{7 Metter}

• The number of active pathologists in the United States decreased from 15 568 to 12 839 (−17.53%)
• The number of pathologists per 100 000 population showed a decline from 5.16 to 3.94
• As a percentage of total US physicians, pathologists have decreased from 2.03%in 2007 to 1.43%in 2017.
• When adjusted by new cancer cases per year, the diagnostic workload per US pathologist has risen by 41.73%; during the same period

The laboratory workforce operates behind the scenes and its workload imbalance often goes unnoticed by the public until diagnostic errors or delays occur. Unlike other medical professionals, laboratory personnel do not have a capped patient load; instead, they handle all clinical specimens sent for evaluation by their clinical colleagues. ^{7 Metter}

Furthermore, a 2013 pathologist workforce study projected a decline in full-time equivalent [FTE} pathologists from 17,570 in 2010 to 14,063 by 2030 due to a predicted supply shortage of pathologists due to an aging US population and anticipated retirements from the pathology workforce. A follow-up computational model forecasted a deficit of 5000 pathologist FTEs by the year 2030. ^{7 Metter}

AI-enabled digital morphological systems- to the rescue

Considering this anticipated crisis, the increased clinical demand and medical complexity, sets the foundation for digital morphological analysis to assist the healthcare laboratory workforce. AI algorithms help analyze and interpret vast clinical data, improving efficiency, objectivity, and accuracy of diagnosis and guiding clinical decision-making, particularly for inexperienced laboratory professionals.  ^{3 Wang}

Several AI-based digital systems are already in use in laboratories, assisting in cell pre-classifying, clinical decision support and standardization while reducing reliance on manual microscopy or completely replacing manual optical microscopy. AI- powered digital morphology analysis further complements traditional laboratory PBS and BMA analysis optimizing diagnostic throughput and accuracy. ^{4 Fan} However, some current digital cell imaging systems are still limited in scope, can analyze only limited areas of these smears, and require manual intervention. ^{5 Zini}

The Scopio Labs X100 and X100HT, Digital Cell Morphology platform is an AI-based digital microscope imaging system that uses a full-field approach to localize and classify blood cells in peripheral blood smears, marking a significant achievement by integrating a digital system with a microscope. ^{5 Zini}

The Future Democratization of AI in Hematology

AI-driven digital morphology platforms are transforming hematology by providing remote access and review through a secure hospital network, easy retrieval of cell images, reduced observer fatigue, faster turnaround times and improved intra- and inter-laboratory morphology standardization. These innovations in digital AI morphology platforms are rapidly becoming the standard of current and future of laboratory hematological process and practice. Enhanced and standardized morphological blood cell analysis competence is thus emerging from the eyes of the few, to the eyes of the many ensuring more accurate and efficient diagnoses for patients worldwide. ^{5 Zini}

DMS-48360 Rev. A

References.

1. Islam O, Assaduzzaman M, Hasan MZ. An explainable AI-based blood cell classification using optimized convolutional neural network. J Pathol Inform. 2024;15:100389. doi:10.1016/j.jpi.2024.100389. PMID: 39161471; PMCID: PMC11332798.
2. Deshpande NM, Gite S, Aluvalu R. A review of microscopic analysis of blood cells for disease detection with AI perspective. PeerJ Comput Sci. 2021;7:e460. doi:10.7717/peerj-cs.
3. Wang S-X, Huang Z-F, Li J, et al. Optimization of diagnosis and treatment of hematological diseases via artificial intelligence. Front Med. 2024;11:1487234. doi:10.3389/fmed.2024.1487234.
4. Fan BE, Yong BSJ, Li R, et al. From microscope to micropixels: A rapid review of artificial intelligence for the peripheral blood film. Blood Rev. 2024;64:101144. doi:10.1016/j.blre.2023.101144.
5. Zini G. Hematological cytomorphology: Where we are. Int J Lab Hematol. 2024;46:789-794. doi:10.1111/ijlh.14330.
6. Kratz A, Lee SH, Zini G, et al; International Council for Standardization in Haematology. Digital morphology analyzers in hematology: ICSH review and recommendations. Int J Lab Hematol. 2019;41(4):437-447. doi:10.1111/ijlh.13042. PMID: 31046197.
7. Metter DM, Colgan TJ, Leung ST, et al. Trends in the US and Canadian pathologist workforces from 2007 to 2017. JAMA Netw Open. 2019;2(5):e194337. doi:10.1001/jamanetworkopen.2019.4337. PMID: 31150073; PMCID: PMC6547243.