In clinical laboratories, the gel card test is a used method for ABO and Rh typing. ABO and Rh are blood type systems for blood classification based on the existence or lack of certain red blood cell (RBC) antigen and serum antibodies. In this test, small quantities of RBCs are added to microtubes present in the test card, and later the card is centrifuged according to test specifications. After centrifugation, the test result is usually determined by a laboratory employee.
The gel test method, originally described by Lapierre et al. [57], uses microtubes filled with a mixture of gel, buffer, and reagents that are centrifuged with a suspension of RBCs. After centrifugation, in negative reagent reactions, the RBCs pass through the gel and collect at the bottom of the microtube, whereas in positive reagent reactions they are retained at the top of the microtube. In these positive reactions, the RBCs antigens bind the antibodies present in the tube, producing aggregates of RBCs. Depending on the strength of the reaction, large aggregates are produced which are blocked from rolling down the microtube column.
The objective of this project is to build an automated system for ABO typing using machine learning and computer vision techniques. To do so a prototype should be developed that is capable of automatically reading and determining a gel card test, displaying the result to the laboratory employee. The final objective is to deploy a prototype in a clinical laboratory, connecting it with the laboratory’s digital record system.