The Digital View

For decades, we’ve been counting and measuring blood cells by analyzing peripheral blood smears stained with May-Grünwald Giemsa or other appropriate stains using a microscope. It’s a labor-intensive and time-consuming procedure that requires intensive training and you need technical expertise to interpret what can be seen in the blood smear. It is also plagued with inter- and intra-observer inaccuracy. In the modern laboratory, fully automated hematological analyzers (or hemocytometers) take care of the largest part of the workload. Such apparatus enable quick and accurate blood counts together with classifying the normal and pathological blood cells, which overcomes most of the drawbacks of microscopic analysis (1).

Despite these considerable developments in blood analysis equipment, most hematological analyzers are not capable of accurate classification of all the normal and pathological cells that may be present in blood, so we still need to prepare a number of samples for optical analysis and manual interpretation.

But, it looks like our burdens will get lighter! Recent technological advances have made it possible to introduce automated image analysis systems that connect to hematological analyzers and other laboratory equipment. These automated instruments prepare blood films (wedging and staining the samples on glass slides) using customized criteria obtained from the complete blood count. The slides are scanned and digital blood smear images captured at high magnification. Images are then analyzed using artificial neural networks according to a pre-set database of blood elements. Importantly, the database is customizable and local users can update it, so it offers flexibility.

Another exciting feature is the ability for the operator to modify image size, magnify single parts, accept the actual categorization of blood cells or else shift some elements to other categories (2). This enables the categorization of white blood cells in normal elements or atypical leukocytes (immature cells, blasts, variant form lymphocytes). The system will also generate additional information about erythrocyte and platelet morphology, flagging samples for the possible presence of anysocytosis, sickle cells, schizocytosis, spherocytosis, acantocytosis, large platelets and platelet aggregates among others (3).

It is unlikely that automated microscopy will completely replace human eyes, but there are many clear benefits, as well as other less obvious advantages, emerging (Table 1). From my clinical perspective, automated image analysis systems allow a standardized approach to cell classification, so that you can compare the digitized blood-smear image with reference slides making the diagnosis consistent with the current morphological classification of hematological malignancies and associated disorders.

These systems also allow less skilled operators to send a digitized blood smear by email or via the web to expert hematologists to get their support and interpretation – they don’t need to be in the same location, which is a huge benefit for small or stat laboratories. Another important advantage is they enable digital storage of large numbers of images for each patient, which allows a more accurate longitudinal comparison of data in follow-up and therapeutic monitoring. Such images have a secondary – yet important – benefit because you can project them onto a large screen for training students and laboratory professionals, making it much easier to share knowledge and teach across the group (2,3).

There are also important cost savings with automated microscopy systems. For instance, you can optimize the software to identify suggestive abnormalities without the direct intervention of an operator, which obviates the need for optical scrutiny. And, because the whole process of selecting, preparing, staining and capturing the blood film is automated, you reduce the turnaround time and save on human resources.

Another valuable benefit: it can alleviate preanalytical problems, which are the main source of laboratory errors and diagnostic delay in clinical chemistry and hemostasis testing. Among these, hemolyzed specimens are the leading cause of sample rejection and test suppression (4). Although there have been many attempts to identify hemolyzed specimens, it remains a major challenge for laboratory hematology. Recent evidence does suggest that automated image analysis systems may help with detecting a number of abnormalities in the blood film that are frequently associated with red blood cell injury (for example, the appearance of cellular debris, anisocytosis, increased size and heterogeneous shapes of platelets), and this could be used for screening sample quality (5).

So, despite being the gold standard, microscopic analysis of blood smear carries a number of technical and practical drawbacks that can be at least in part overcome with automated microscopy. We just have to embrace it to begin to realize the benefits!