Current state of digital morphology analysers
Automated digital analysers of blood smears has been on the rise due to the ever-progressing field of digital image analysis (by means of machine learning). Manual analysis of blood smears using light microscopy has been (and still is) the standard approach by practitioners, however, this manual method has its drawbacks such as:
- labour intensiveness.
- Require continuous training of staff.
- Susceptible to interobserver variability.
- Requires physical storage of slides.
- Ease of access to archived slides.
- Obtaining secondary consultations.
Morphological analysis of WBC, RBC and platelets are needed to identify haematological diseases. Automated digital analysers help automate some of the morphological examinations of blood smears and in theory should reduce the workload on the hematologists/morphologists. Some current digital morphology analyzers in the market are:
- CellaVision DM96, DM1200, and DM9600
- Sysmex DI-60 (uses the CellaVision DM1200 platform)
- Vision Hema (West Medica)
- EasyCell (Medica Corporation)
- Nextslide (Nextslide imaging LLC)
- Cobas m511 (Roche Diagnostics)
- HamaCam (Fraunhofer institute for integrated circuits IIS)
The CellaVision system has been around the longest and is considered to be a market leader in Automated digital analysis. The system captures images on a blood smear and pre-classify them using image analysis software then displays the pre-classified/unidentified cells on a computer screen. It is able to pe-classify WBC to give a differential count. The image analysis is usually done using artificial neural network, however, the ANN is unable to fully and reliability identity all cells, for example, it is able to identify neutrophils, lymphocytes and monocytes with better accuracy than basophils, lymphoblasts or immature granulocytes (identification of immature granulocytes and their classification into early or later maturity stages can also be difficult when done manually since its dependence on the skills of the morphologist).
In the context of RBC CellaVision has CellaVision ARBCA (Advanced Red Blood Cell Application), its is able to (based on morphological features) to classify RBC into 21 morphological categories. It is found that ARBCA is a useful tool to screen and quantify morphological changes that are linked to different forms of hereditary hemolytic anemia. The classification of schistocytes, however, isn’t fully reliable since the main challenge of the analysis of schistocytes remains the lack of a concrete morphological definition. This is also true for malaria screening where the CellaVision ARBCA system has too low of a sensitivity to rely on fully in the screening for the parasite.
The DI-60 Integrated slide processing system
It’s the first fully integrated system cell image analyzer that can perform CBC, perform smear preparation as well as determining cell identification and location, this can be done by only providing a blood sample to this automated machine. It uses the same software as the CellaVision DM-1200. WBC classification on this device was found to be acceptable, however RBC grading and malaria screening were not and thus it still required manual slide review.
No peer-reviewed paper is available for this system but according to promotional content the automatic classification of blood cells. It can analyse white and red blood cells as well as platelets and reticulocytes. Different versions of the systems are available for consumers ranging from The Vision System Ultimate analyzer that can load up to 200 slides and other versions that can load 4 and 8 slides at once. Other adds-ons are available that can be used in the identification and classification of cells in bone marrow samples, body fluids and cervical cytology.
Easy Cell Assistant
No peer-reviewed literature is available or this system but according to promotional material the system can locate and classify 100 or 200 leukocytes. The system can also pre-classify smudge cells, NRBCs and lymphs.
This system abilities include image processing, management, and online review. It can scan the image of cells at up to 100x magnifications and has a throughput of 26 slides per hour. When compared to manual microscopy it produced agreeable results, the results also correlated well with the CellaVision system in terms of neutrophils, lymphocytes, monocytes, eosinophils, blasts and immature granulocytes. This system did not perform as well as the CellaVision in terms of basophils, atypical lymphocytes, and bands.
This system can derive a 5 part differential on 600 nucleated cells on a slide. Using its high magnification module (50X lens) it can classify WBC, evaluate RBC and platelets – the system can provide quantitative and morphological information. It can also produce high-quality images without the use of oil immersion. The Cobas results correlate well with the Sysmex system, it also compares well with manual microscopy.
This system is certified by the Medical Devices Act in the European Union. It has a motorised microscope that can be controlled by a computer system. It can provide information on WBC, RBC and platelets. Further applications currently under development are malaria screening and detection of bone marrow cells for leukemia.
Education and training
Digital morphology helps with
- Ability to easily consult colleagues.
- Reference abnormal cells.
- Archival of cell images for education and training
- Development of competency assessments and quality assurance.
- Scientific research on archived images alongside additional patient data.
- Screening for atypical chronic Lymphocytic leukemia and classifications of lymphoid B cells.
- Development of competency software to help with the assessment of laboratory performance and further education.
When to use manual microscopic review over Digital Analyzers:
- When working with samples from newborns or patients with leukemia.
- Suspicion of presence of pathological cell types, including blasts, plasma cells and immature granulocytes.
- Suspected dysplastic cells.
- Suspected schistocytes.
- Screening for intracellular parasites e.g. Malaria and Babesia
- RBC agglutination
- Platelet clumps due to their location in feather or lateral edges.
As more automated digital analyzers enter the market the need for standardization becomes more apparent. Standardization such as of shareability of images and information( such as baseline colours and display parameters) between different machines will be needed. Mechanisms to evaluate image quality as well as mechanisms to correct minor errors in slide preparations is also needed. Differences of operation between laboratories in terms of tissue processing, staining as well as the hardware and software used to produce the cell type results are obstacles that are hindering the progress of standardization.
More studies are needed to determine the effect and the performances of the different automated digital analyzers, in many cases human expertise and intervention are needed alongside the analyzers. More work is needed to improve the accuracy and consistency of the analysis, but it is clear that the general direction of is eventually full autonomy. As more machines from competing manufacturers enter the market, innovation will be pushed continuously. Standardization is needed to exchange and access to information between machines of different manufacturers. Since different preparation techniques can produce different end results we need standardization in terms of tissue/slide processing/staining to make at least make comparisons between machines easier and more reliable.
This is a summary of the original review