Breast cancer is the most common form of cancer among women, accounting for 25 % of all cancer cases worldwide. Much progress has been made in detecting breast tumours, and survival rates are relatively high compared to other forms of cancer.

The project has designed an intelligent computer-assisted system for detecting and diagnosing problematic breast lesions that will help save lives by cutting the number of missed or misinterpreted cases. Its researchers also hope to reduce the need for unnecessary biopsies.

“The project’s research enables the translation of novel and complex image-processing algorithms to quantify the features of suspicious lesions into prognostic markers of the progression of lethal, invasive cancers. This technology can save lives, reduce misdiagnosis and improve the quality of life for millions of women worldwide,” says project coordinator Anke Meyer-Baese of Florida State University and an affiliated professor at Maastricht University in the Netherlands.

Better software, better detection

A technique known as ‘Breast Imaging-Reporting and Data System (BI-RADS) descriptors’ is currently used to assess breast tumours in mammography. However, it is known to fail to correctly assess lesions that are difficult to diagnose, for example when the boundary between the tumour and background tissue is difficult to detect.

“These lesions exhibit heterogeneous behaviour and cannot be characterised solely based on their tumour shape or contrast-enhancing behaviour. While their shape mimics a benign tumour, their contrast-enhancement uptake is of malignant nature and vice versa,” says Meyer-Baese.

This particular type of tumour poses an enormous challenge for both radiologists and the current computer-assisted evaluation systems that have the potential to reduce human error in cancer diagnosis.

Throughout the project, MAMMA developed software used spatiotemporal descriptors to capture the shape and contrast-enhanced behaviour of diagnostically challenging lesions. The team also used a novel computational approach – called radiomics – to represent oncological tissues.

“Integrated in a radiomics approach, the new spatiotemporal descriptors showed superior capabilities for the detection and diagnosis of diagnostically challenging lesions compared to the standard BI-RADS descriptors,” says Meyer-Baese.

Personalised medicine

In line with the growing importance of personalised medicine, the technology has developed treatment strategies able to respond to the specific characteristics of each patient and each cancer type. Furthermore, the software will provide tailored cancer management strategies for patients diagnosed with early-stage breast cancer.

Moreover, novel computer-assisted diagnostics for diagnostically challenging lesions will help cut costs. The technology can distinguish lethal from non-lethal cancer, avoiding over-diagnosis, unnecessary treatment and associated costs, as well as needless patient anxiety.

The original of this article was published at the European Union’s site, ec.europa.eu/info/index_en. For more on Meyer-Baese, go to www.sc.fsu.edu.