The HAMAM project (‘Highly Accurate Breast Cancer Diagnosis through Integration of Biological Knowledge, Novel Imaging Modalities, and Modelling’) was launched in September 2008 under the 7th Framework Programme for Research, and was successfully finalised in April 2012. The HAMAM objective was to improve the early detection and accurate diagnosis of breast cancer by integrating the available multi-modal images and patient information on a single clinical workstation. The consortium of nine project partners from Europe and the United States aimed to facilitate an earlier detection of breast cancer and a reduction of unnecessary biopsies.
Among the key outcomes of the project are a number of tools designed to automatically correlate and jointly interpret information from different sources. With conventional imaging workstations, extensive training is necessary before readers are able to reliably identify correspondences of suspicious structures in 2D projection images, for example mammography, and 3D modalities, such as ABUS. A major result of the HAMAM project was a set of new techniques to automatically map spatially corresponding anatomical structures in each modality. The images can then be presented such that sizes and positions match between modalities, thereby instantly orienting the human reader and facilitating more efficient and accurate combined assessment of findings.
In addition, a novel system was developed to classify lesions as probably benign or malignant using image descriptors from mammography jointly with kinetic and morphological descriptors from MRI. Another computer-aided diagnosis (CAD) system assists radiologists in characterising suspicious lesions in ABUS; a promising technology for screening women with dense breasts. In a reader performance study this new CAD tool significantly improved the performance of radiology residents compared to conventional ABUS reading.
The majority of HAMAM’s scientific results have become part of a patient-centric workstation that enables the reader to quickly access all available patient-related imaging studies plus non-imaging information and make fully informed, computer-assisted decisions about diagnosis and treatment, offering the potential to dramatically improve the efficiency of breast cancer care.