HAEMATOLOGY
are some of the current applications: n Data processing. One of the most time-consuming steps in traditional flow cytometry is manual gating. It involves defining cell populations based on marker expression by visually interpreting two-dimensional plots.23-26
Stained sample
Mirror filters
Sheath fluid
The process is subjective and
tends to vary between users, resulting in inconsistencies. AI algorithms, especially those using machine learning, can analyse large datasets quickly and objectively. For instance, Uniform Manifold Approximation and Projection (UMAP) is a machine learning algorithm used for dimensionality reduction to visualise high-parameter datasets in a two-dimensional space. That helps identifying patterns and detecting outliers with high precision, thereby minimising user bias and saving significant time.
n Detection of rare cells. Detection of rare cell populations, including minimal residual disease in leukaemia patients, is an important yet difficult task. Deep learning will perform very well in this area, especially by identifying low- frequency events that may not be noticed using traditional methods.20 This is particularly significant in oncology, where the detection of a few cancerous cells can help guide prognosis and treatment decisions, potentially leading to better patient outcomes.
n Improved imaging capabilities. AI- based solutions like Invitrogen Attune CytPix have the potential to combine cytometry data with high-resolution imaging. It is the AI that enables morphological data to be correlated with fluorescence signals and offers a richer and more holistic view of cell behaviour and characteristics.22,23 For example, such an integration can describe in finer detail the morphological features of apoptosis or cancer cell phenotype changes.
n Inter-laboratory standardisation. One
Hydrodynamic focusing
Detectors
Data acquisition and analysis Electronics system
Laser
Role of flow cytometry in the diagnosis of haematological malignancies.15 To waste
of the common issues arising from flow cytometry is variability in results due to differences in expertise of the operators, instrument calibration, and experimental set-up. Artificial intelligence is able to minimise this by automating crucial processes to ensure consistency of outputs, which is more valuable in the case of multicentre clinical trials or large research studies where reproducibility is vitally important.6-8
Future potential In the future, AI will have immense potential regarding flow cytometry. Among these possibilities include the following transformative examples: n Automated gating through deep learning. Deep learning algorithms could completely replace traditional gating. Advanced deep learning can learn from historical datasets to set exact gates for detecting cell populations, thereby significantly streamlining workflows and removing variability. It would mean that more time could be devoted by researchers and clinicians to result interpretation
rather than repetitive tasks.14,17
n Integrated multi-omics analysis. The integration of flow cytometry data with other omics datasets, such as genomics, transcriptomics, and proteomics, using AI could provide unparalleled insights into disease mechanisms. For example, the cellular phenotyping integrated with genetic mutations identified in sequencing data could provide a comprehensive understanding of cancer progression and resistance to treatment.
n Predictive analytics. Such AI-based predictive models could, through flow cytometry, forecast the course of the disease or predict therapeutic response.18
For instance, it can predict,
based on machine learning algorithms, a possible cellular signature and whether this would elicit a desired therapeutic response from a drug. That is, predictive capacity towards the individual in therapy; a major stride towards personalised medicine.
n Unsupervised learning: discovering new facts. AI may discover previously unknown cell subpopulations or
WWW.PATHOLOGYINPRACTICE.COM APRIL 2025
35
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56
