Researchers at the University of Veracruz (Universidad Veracruzana, UV) developed a method to analyze Western blot images to effectively detect breast cancer in shorter time. This clinical innovation, jointly developed by the Center for Research in Artificial Intelligence (Centro de Investigación en Inteligencia Artificial, CIIA) and the UV Institute of Biological Research (IBR), involves the use of computer vision algorithms for the classification of protein bands.
The origin of project was a joint research between the Institute of UV-IBR and the Institute of Biomedical Research of the National Autonomous University of Mexico (UNAM), which compared the protein bands in a Western blot obtained from women with cancer with those obtain from women with a benign breast tumor and without any mammary pathology.
Timely, fast and reliable diagnosis
Dr. Tania Romo González, UV-IBR researcher, explained that the objective of the project is to develop and provide doctors with an earlier detection method for breast cancer. The algorithm can be used to automatically analyze any Western blot images.
“The diagnosis for breast cancer nowadays —at least in Mexico— is through mammograms. However, it is not a good method, since the tumor mass must be at least one centimeter in diameter to be recognizable. When this happens, the disease is already advanced, causing a poor prognosis for the patient,” said Dr. Romo.
She added that, unlike mammography, identifying the presence of autoantibodies in the blood, which appear in the presence of tumor breast cells, makes it possible to diagnose the disease early stages.
“Using the Western blot technique or 1-dimensional immunoblot (1D immunoblot), and comparing sera from women with breast cancer, with breast pathology and without mammary pathology, we detected that there were (protein) bands that were present only in women with pathology.”
However, the comparison and analysis of the protein bands is visually complex for the experts; therefore, in collaboration with the CIIA the group of researchers are working to develop an automatic method for classification of images according to criteria based on their discoveries.
Artificial Intelligence
Dr. Héctor Gabriel Acosta Mesa, a researcher at the CIIA, participates in the technological development of the project. The project called ‘Using Computer Vision Algorithms for Breast Cancer Screening in Protein Bands’ is a semi-automatic method for studying any Western blot imaging.
“The advantage of analyzing images using computer vision algorithms is that they perform a quantitative analysis, that is, they directly measure the values of the bands and extract the patterns that they form. This eliminates the subjective aspect of the human eye,” said the specialist.
This method allows a more precise analysis since the algorithm has a mathematical model trained with examples, and is faster compared to the evaluation of an expert, which reduces the diagnosis from days to minutes. Finally, he indicated that the idea is to continue working to create a software tool that supports experts, making their work more precise, faster and eliminating subjectivity.
Source: Agencia Informativa CONACYT
