Improving the prediction of treatment response in breast cancer leading to more targeted treatments

Improving the prediction of treatment response in breast cancer leading to more targeted treatments

28 October, 2019

Breast cancer is the leading malignancy in women in Europe with over 562,500 cases in 2018, of which 20% are in women under 50 years old. Breast cancer is the second leading cause of death for women, and one in eight women in Europe will develop this kind of cancer before the age of 85.

This disease is also associated with a great economic cost. In fact, in 2010 this cost was estimated in more than 15 billion euro in Europe. About 46 % of the breast cancer cost is spent on treatments and drugs.

The BigMedilytics project is developing a radiomics pilot that aims to predict treatment response in breast cancer leading to treatments that are more targeted. IBM, Institut Curie, and VTT are working together within this pilot.

The pilot examines chemotherapy treatment given before surgery (neoadjuvant treatment). The pilot strives to improve the prediction by using artificial intelligence algorithms to analyze mammograms, ultrasound and magnetic resonance imaging (MRI) along with structured clinical data and predict patient response to neoadjuvant treatment.

Through the use of computer vision, deep neural networks and other machine learning techniques, finer stratification of the patient cohorts will be achieved. This will allow evaluating the ability to make personalised treatment decisions.

The pilot builds a radiomics system with a model-to-data paradigm where all the data never leaves Curie premises. Anonymized imaging and clinical data are being exported to the pilot servers that reside in Curie. Open data of similar characteristics is added to enrich the training dataset. Then, the multi-modal algorithms analyze the data and produce results that are embedded in objects according to the HL7 FHIR standard. The results are also visualized in a viewer.

Access the poster of the Breast cancer pilot presented during the event “Big Data: Fueling the transformation of Europe’s Healthcare sector” that took place on the 4th and 5th of September in Valencia (Spain), and learn more in detail about this pilot and its architecture.