A new artificial intelligence tool can better analyze blood and bone marrow tests to make a diagnosis of cancer.
- Scientists have developed an artificial intelligence tool to analyze blood and bone marrow tests faster and more accurately.
- Once the data has been processed, it proposes a diagnosis to the doctor who must confirm or invalidate it.
Between 9,000 and 10,000 people are affected by leukemia each year in France, according to the Foundation for Medical Research. When an individual is healthy, they have a constant number of blood cells and lymphocytes (a type of white blood cell) circulating in the blood. But, in some cases, white blood cells accumulate abnormally and uncontrollably, which can cause two types of cancer depending on the location. If it is localized in the blood or bone marrow, it is leukemia, that is to say a cancer of the blood. If it’s in the lymph nodes or the lymphatic system – a component of the immune system – the cancer is called lymphoma. The diagnosis of these cancers is usually made after analysis of the results of a blood test. If the doctor is in doubt, he may also perform a bone marrow examination.
Flow cytometry to collect data from each particle
According to a study published in the journal patterns, a new artificial intelligence technique could improve the quality and speed of the analyzes of these two samples. Generally, these examinations are treated via a method called flow cytometry. This makes it possible to measure the individual characteristics of each particle – cells, bacteria, parasites, etc. – in the blood or bone marrow, independently of other particles. Thus, laboratory workers collect data such as the size, shape, biological activity and complexity of each particle. Essential information for making a diagnosis of cancer. In detail, with this flow cytometry, laboratories use antibodies that attach themselves to the surface of the particles to be analyzed. These antibodies are coupled with fluorescent dyes, which are called markers because, thus colored, they are recognizable. These can also be used to detect small differences between cancerous and healthy cells. With flow cytometry, the signals emitted by each cell are recorded and are then collected and transcribed in the form of graphs or statistics for example. It is with these results that doctors make a diagnosis.
Artificial intelligence to process data faster and more accurately
The more precise the characterization of cancer cells, the better and faster the diagnosis will be. However, in normal times, it is very difficult to analyze all the characteristics of the data resulting from flow cytometry because there are so many of them. The authors of this study have therefore developed an artificial intelligence tool capable of automatically and autonomously analyzing all this information. To create their algorithm, the team “trained” it with more than 30,000 data from lymphoma patients. “Artificial intelligence takes full advantage of data and increases the speed and objectivity of diagnostics“, explains Nanditha Mallesh, the main author of this study. Once the data has been processed, the artificial intelligence suggests a diagnosis to the doctor. “The purpose of using artificial intelligence is not to replace doctors, but to get the most benefit from the information contained in the data”, explains Peter Krawitz, one of the authors. The practitioner must then confirm or invalidate this diagnostic proposal. Time saving and more finely analyzed data.
The advantage of this artificial intelligence is that it can be used by all laboratories, especially small ones which would not have been able to develop such technology themselves. Another advantage: it could also be used for other pathologies, such as the diagnosis of rheumatism-related diseases whose examinations are often treated with the flow cytometry method.
.
