New AI-based tool improves diagnosis of breast cancer tumors
Researchers at Karolinska Institutet in Sweden have developed an AI-based tool that improves the diagnosis of breast cancer tumors and the ability to predict the risk of recurrence. The greater diagnostic accuracy can lead to more personalized treatment for the large group of breast cancer patients with intermediate risk tumors. The results are published in the scientific journal Annals of Oncology.
Each year, approximately two million women worldwide develop breast cancer. In the diagnostic procedure, tissue samples from the tumor are analyzed and graded by a pathologist and categorized by risk as low (grade 1), medium (grade 2), or high (grade 3). This helps the doctor to determine the most appropriate treatment for the patient.
About half of breast cancer patients have a grade 2 tumor, which unfortunately gives no clear indication of how the patient should be treated. Therefore, some patients are over-treated with chemotherapy while others are at risk of being under-treated. This is the problem we tried to solve. “
Yinxi Wang, first author of the study and doctoral student, Department of Medical Epidemiology and Biostatistics, Karolinska Institute
Hospitals have recently started to make limited use of molecular diagnostics to improve the accuracy of breast cancer risk assessment, but these methods are often expensive and time consuming. Researchers at Karolinska Institutet have now developed and evaluated an AI (artificial intelligence) -based method for tissue analysis. The study shows that the AI-based method can further divide patients with grade 2 tumors into two subgroups, one at high risk and the other at low risk, which clearly differ in terms of risk of recurrence.
“A big advantage of the method is that it is cost effective and fast, as it is based on microscopic images of dyed tissue samples, which is already part of the hospital procedure,” says co-last author Johan Hartman, professor of pathology in the department. of oncology-pathology, Karolinska Institutet, and pathologist at Karolinska University Hospital. “It allows us to offer this type of diagnosis to more people and improves our ability to give the right treatment to any patient.”
The AI model was trained to recognize the characteristics of high resolution microscopic images of patients classified with grade 1 and 3 tumors. The study is based on a large microscopic image bank of 2,800 tumors.
“It’s fantastic that deep learning can help us develop models that not only replicate what medical specialists are doing today, but also allow us to extract information beyond the reach of the lens. ‘human eye,’ says co-last author Mattias Rantalainen, associate professor and research group leader at the Department of Medical Epidemiology and Biostatistics, Karolinska Institutet.
The method is not yet ready for clinical application, but a regulatory approved product is being developed by a newly formed company, Stratipath AB, which is supported by KI Innovations. Researchers will now further evaluate the method with the aim of bringing a product to market by 2022.
Wang, Y., et al. (2021) Improving the histological classification of breast cancer using deep learning. Annals of Oncology. doi.org/10.1016/j.annonc.2021.09.007.