Researchers have designed a system with which they managed to automate cell counting of bone marrow samples, which will help diagnose hematological diseases such as blood cancers more quickly and accurately.
The progress is the work of specialists from the SpotLab company, in collaboration with researchers from the Polytechnic University of Madrid (UPM), the Networked Biomedical Research Center for Bioengineering, Biomaterials and Nanomedicine (CIBERBBN), the Biomedical Research Center in cancer network (CIBERONC), the Complutense University of Madrid (UCM), and a group of hospitals including the 12 de Octubre University Hospital, the Vall d’Hebron Hospital and the Alcorcón Foundation University Hospital, in Spain, all these entities.
Differential cell counting of bone marrow aspirates is a technique that is currently performed manually in most health centers. This is a time-consuming task and the result may vary depending on the experience of the observer.
For this reason, this procedure proved to be a suitable candidate to be automated thanks to artificial intelligence, the objective of this multidisciplinary project with the clinical direction of Dr. Joaquin Martinez of the 12 de Octubre Hospital and CIBERONC.
The researchers specifically designed an AI algorithm based on deep learning (a mode of artificial intelligence) that can automatically differentiate and count different cell types in images of bone marrow samples, as detailed by David Bermejo-Peláez, researcher by SpotLab. To digitize images, the system does not require complex and expensive scanners or devices, but does so using smart mobile phones (smartphones), making it a large-scale adoptable system that can be implemented in any healthcare service of any type hospital in the world.
Capturing images of a bone marrow sample with a cell phone. (Photo: SpotLab)
According to María Jesús Ledesma, researcher at UPM and CIBERBBN: “The results obtained have demonstrated that this technology significantly reduces the analysis time of bone marrow samples, as well as the variability between observers during the analysis.” “The developed system increases the efficiency and precision in the diagnosis of haematological diseases such as leukemia or multiple myeloma,” says María Linares, researcher at UCM.
The study is titled “AI-augmented digital microscopy to interpret bone marrow samples for hematological diseases.” And it was published in the academic journal Microscopy and Microanalysis.
This work represents a step forward towards the integration of innovative artificial intelligence technologies into clinical routine to fight cancer. Currently, this line of work continues with the use of artificial intelligence to improve the accuracy of diagnosis, selection of effective treatments, and prognosis of patients with hematological diseases. (Source: UPM)
Interview Between Time.news Editor and AI Expert David Bermejo-Peláez
Time.news Editor (TNE): Welcome, David! We’re excited to discuss your groundbreaking work in automating cell counting for bone marrow samples. Can you start by explaining the significance of this innovation in the context of diagnosing hematological diseases?
David Bermejo-Peláez (DBP): Thank you for having me! The significance of our work lies in the speed and accuracy it brings to diagnosing diseases such as blood cancers. Traditionally, cell counting is a manual process that is both time-consuming and prone to human error. By automating this process with our AI algorithm, we aim to provide a reliable and efficient method for clinicians, leading to quicker diagnoses and better patient outcomes.
TNE: That’s impressive! You mentioned that traditional methods can vary based on the observer’s experience. How does your AI contribute to overcoming this variability?
DBP: Absolutely! One of the key advantages of our AI system is its ability to maintain consistent accuracy regardless of external factors like the operator’s experience. The deep learning algorithm we’ve developed can identify and differentiate various cell types based on image data autonomously, which reduces subjective interpretation and enhances diagnostic reliability.
TNE: Could you tell us more about how the technology works? What kind of data do you use for training the algorithm?
DBP: Certainly! We trained our algorithm using a large dataset of digitized bone marrow images, which included examples of different cell types. The AI learns through a process called supervised learning, where it analyzes labeled data—meaning we provided it with images annotated with the correct cell classifications. As the algorithm processes more images, it improves its accuracy in recognizing patterns, leading to effective cell counting.
TNE: It sounds like a monumental task! Can you tell us about the collaboration between different institutions that contributed to this project?
DBP: Collaboration was vital to our success. We partnered with esteemed institutions like the Polytechnic University of Madrid, CIBERBBN, CIBERONC, and several hospitals, including the 12 de Octubre University Hospital. This multidisciplinary effort brought together expertise in biomedical research, engineering, and clinical practice, ensuring that our solution met real-world healthcare needs while advancing scientific boundaries.
TNE: As with any technological advancement, there are questions about implementation. How do you foresee this system being adopted in healthcare settings, and what challenges do you anticipate?
DBP: That’s an important question. We envision our system being integrated into existing laboratory workflows. However, challenges remain, such as ensuring that users are adequately trained to work with the technology and addressing regulatory standards for medical devices. We are actively engaging with medical professionals to tailor our solutions to their needs and to simplify the transition process.
TNE: It’s great to see such dedication to real-world application! As an expert in AI and biomedical research, what excites you the most about the future of this technology in healthcare?
DBP: The potential is enormous! I believe our successful automation of cell counting is just the beginning. In the future, AI can play a critical role in diagnosing various diseases, predicting patient outcomes, and personalizing treatment plans. The combination of AI and healthcare could fundamentally change how we approach medical practices, making them more efficient and precise.
TNE: Thank you, David, for sharing your insights on this exciting development! It’s clear that your work is paving the way for advancements in hematological disease diagnosis. We look forward to seeing the impact of your research in medical practice.
DBP: Thank you for the opportunity! I’m looking forward to seeing where this journey takes us in improving healthcare together.