Breast Cancer

In Europe, 495.000 women are newly diagnosed each year with breast cancer. The number of breast cancer survivors in Europe is increasing, at a rate of 352.000 patients per year, which calls for advancing prognostic models for cancer follow-up. The treatment and survival rate of breast cancer patients is primarily correlated to the stage of disease upon diagnosis, where patients diagnosed with stages 0, 1 and 2, on average experience 5-year survival rates of approximately 90%.

The early diagnostic is critical precondition for increase of survival rate, thus an innovative, yet reliable low-cost technological solution would facilitate affordable periodic screening examinations. Additional challenges are associated with monitoring and predicting cancer evolution, spread and relapse which currently relies on Health Care Professionals’ experience in integration of multivariate and multimodal patient history data and presents a fertile ground for development and application of machine learning techniques.

The early diagnostic is critical precondition for increase of survival rate, thus an innovative, yet reliable low-cost technological solution would facilitate affordable periodic screening examinations.

Mammography alone or in combination with ultrasound represents the common initial method to diagnose breast cancer. These methods are usually sensitive and specific enough to establish the diagnosis of a suspicious lesion and to refer the patient to percutaneous biopsy. In some cases, additional magnetic resonance mammography is performed. PET/CT scans to identify the cancer spread at diagnostic stage may be included in specific cases where there is evidence for metastasis.

Depending on the cancer type and size, as determined by biopsy and biochemical reports, the treatment may start first, to be followed by another set of imaging (mammography and ultrasound) prior to surgical procedures. Otherwise, surgical procedures are followed with treatment and imaging as a follow up. Depending on these results additional imaging (e.g., PET/CT) for identification of metastasis and treatment is performed.

The challenges addressed in this pilot of INCISIVE are the following:

1. Support to early diagnostic via improvement of detection and classification of tissue changes through AI enabled fusion of multiple images obtained through mammography and ultrasound. Also via improving the performance of lower-cost imaging modalities to facilitate wider population recruitment and more regular screening using affordable methodology;


2. Enhancement of services oriented to Health Care Professionals (HCP) to ensure standardized examination pipelines for improved image and data, management, interpretable analytics and valuable recommendations to HCP;


3. Improving the specificity of magnetic resonance using AI tools driven by the INCISIVE pan-European image repository;


4. Enhancing digital pathology tools to enable wider implementation of automated classification tools in histopathology examination to complement and support subjective and experience-dependent expert manual tissue examination;


5. Improving image registration tools for efficient visualization of treatment effects and cancer evolution


6. Automated recommendations and prognostic modelling based on the fused patient data as a support to HCPs’ manual aggregated data analysis for therapeutic decision-making and effective prognostic outcome prediction. The labelled image, clinical and biological data repositories accumulate large pools of knowledge to mine useful information with respect to treatment planning and cancer evolution modelling as they fuse experience and multiple views on the problem otherwise beyond single expert data inspection potentials.