SURGIOMICS – IMPROVING PREOPERATIVE DECISION-MAKING IN PRIMARY BREAST CANCER BY RISK MODELS AND ARTIFICIAL NEURAL NETWORKS

Breast cancer is today diagnosed at an early stage and only 20-25% of the patients are diagnosed with nodal metastasis.

However, clinically node-negative patients are subject to invasive axillary staging with documented side effects and health care costs.

The project aims to develop a preoperative decision-making tool by integrating preoperatively available information on patient´s characteristics, tumour- and radiological features into an artificial neural networks (ANN) model.

The purpose is to identify breast cancer patients with a low risk of axillary nodal metastasis before surgery whom can be candidates for omission of surgical staging allowing personalised surgery.The research group represents multidisciplinary competences including bioinformatics and clinical specialists with experience of developing risk models for clinical decision making.

An initial ANN model with an area under the curve (AUC) of 0.74) was recently presented and the project aims to improve its performance by integrating features from axillary ultrasound and mammograms.

The research program outlines well-defined deliveries and the novel version of the ANN model is expected to be developed within four years.

The ANN algorithm will be implemented in a web interface and carefully evaluated according to guidelines for diagnostic tests.

In parallel, we are deciphering molecular characteristics associated with nodal metastasis for improved understanding of lymphatic spread.