Improvement of the artificial intelligence solution to support the diagnosis of mammography exams with the addition of the breast cancer risk rating, and an implementation of the methodology used for the breast tomosynthesis modality

• Grant number: 22/02264-4
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Duration: September 01, 2022 - August 31, 2024
• Field of knowledge: Engineering - Biomedical Engineering - Medical Engineering
• Principal Investigator: Catarina Cardoso Reis
• Grantee: Catarina Cardoso Reis
• Host Company: Harpia Health Solutions Ltda
• CNAE: Atividades de serviços de complementação diagnóstica e terapêutica
• City: São José dos Campos
• Associated scholarship(s): 23/03416-5 - Improvement of the Artificial Intelligence Solution to Support the Diagnosis of Mammography Exams with the Addition of the Breast Cancer Risk Rating, and an implementation of the methodology used for the Breast Tomosynthesis modality., BP.TT; 22/10601-0 - Improvement of the Artificial Intelligence Solution to Support the Diagnosis of Mammography Exams with the Addition of the Breast Cancer Risk Rating, and an implementation of the methodology used for the Breast Tomosynthesis modality., BP.TT; 22/10611-6 - Improvement of the Artificial Intelligence Solution to Support the Diagnosis of Mammography Exams with the Addition of the Breast Cancer Risk Rating, and an implementation of the methodology used for the Breast Tomosynthesis modality., BP.TT; 22/10602-7 - Improvement of the Artificial Intelligence Solution to Support the Diagnosis of Mammography Exams with the Addition of the Breast Cancer Risk Rating, and an implementation of the methodology used for the Breast Tomosynthesis modality., BP.TT; 22/10569-0 - Improvement of the artificial intelligence solution to support the diagnosis of mammography exams with the addition of the Breast Cancer Risk rating, and an implementation of the methodology used for the breast tomosynthesis modality, BP.PIPE

Abstract

In Brazil, approximately 65,000 new cases of breast cancer are diagnosed annually, and 20% of these patients die in less than 12 months after their discovery due to late diagnosis. This is a global problem since in 2018 there were an estimated 2.1 million new cases of the disease worldwide, and it caused 627 thousand deaths. Mammography, associated with the increasing use of breast tomosynthesis, are the most important medical imaging exam modalities for the diagnosis of the disease, providing information on the internal structures for the detection, characterization and monitoring of breast cancer. However, research shows that inconsistencies in the diagnosis of these tests are still very high. This is due to the exhaustive workday of radiologists and the heterogeneity of the quality of the equipment used for these exams. These inconsistencies affect patient well-being, increase the mortality rate, and add to the additional costs of diagnosis and treatment for service providers. To circumvent this situation, computational solutions based on artificial intelligence, computer vision and image processing have been developed and applied to highlight obscure information in medical images. As they provide additional, accurate and objective information, they are increasingly present in the radiologist's routine, enhancing success in diagnosis and subsequent treatment. Now on sale, Delfos Mamografia, a computational solution developed by Harpia Health Solutions, supports radiologists in carrying out the report in hospitals, clinics, health operators and teleradiology companies. Specifically, the solution extracts and delivers the main clinical elements of the mammography study, providing sets of information for the radiologist's decision directly in the reporting system. However, the extraction of clinical elements from tomosynthesis images is not yet available in the tool; as well as the detection of architectural distortion, and the risk grading of the evidenced elements. Thus, the general objective of this project is to improve the solution by grading the malignancy of suspicious findings and classifying the studies according to the BI-RADS standard and implementing and adjusting the methodologies adopted for breast tomosynthesis exams. To this end, combinations of state-of-the-art computational methods of artificial intelligence and image processing will be researched, developed, implemented and improved by Harpia's team and infrastructure. As a result of this project, we will expand the added value potential of the tool, promoting practicality, speed and precision in the diagnostic journey, bringing quality of life to doctors and patients, and consequently bringing savings to service providers. (AU)