For the past decade, computer-aided detection (CAD) has been developed to improve the sensitivity of mammographic detection of breast cancer. The mammograms are digitised and evaluated with algorithms, and the final diagnosis is made by the radiologist using information supplied by the CAD-system. The computer-aided detection can perform as an assistive tool to mammographic interpretation by a radiologist which complements rather than replaces the human observer. Extensive survey on existing state-of-the-art in relevant areas reveals that present computer-aided technology development, whether based on academic research or commercially available systems, do not perform well in the process of breast cancer screening. Recent comparative study also reveals that the current CAD approaches do not play significant roles in assisting experienced mammographers. Therefore, dramatically different approaches and innovative ideas are needed to forward the existing research towards development of efficient and effective breast cancer screening system. To this end, we propose to research and innovate new technologies for computerised and automatic normal mammogram exclusion rather than abnormal detection. As a result, doctors only need to read those mammograms that fail to get through the automatic normal mammogram exclusion system, and thus significant reduction of their work load, and indeed the cost of early diagnosis programme, can be achieved.

Objectives and project outcome: To develop a computerised and automatic normal mammogram exlusion system towarded efficient and effective breast cancer screening and early diagnosis via approaches of: (i) medical image processing and analysis; (ii) machine learning such as SVM and neural networks etc. (iii) virtual physiological human modelling.

Commercial potential of such a system include: (i) the system can be applied to hospitals where all mammograms can be automatically processed by such a system first, and only those mammograms that failed to get through need to be read and examined by doctors; (ii) the system can be installed in a manner of photo-me like machines, where people can have an automatic pre-check and only selected number of people require further examinations by doctors. Yet such a pre-check can be operated 24 hours a day and 7 days a week.

The project is to be organised in 7 work packages, which include:

WP1: project management;

WP2: specifications and requirements;

WP3: computerised normal mammogram exclusion via microcalcification detection;

WP4: computerised normal mammogram exclusion via others (lesion, mass etc.);

WP5: System integration;

WP6: clinical trials;

WP7: dissemination and exploitation.

(i) medical expertise in mammogram reading, examination and breast cancer screening;

(ii) computerised digital mammogram analysis and research;

(iii) commercial development of medical systems, especially mammogram scanning and examining systems.

(i) conduct clical trial in hospitals with patients;

(ii) industrial exploitation and development of the normal mammogram exclusion system;

(iii) computerised digital mammography research;

(i) hospitals or breast cancer screening units;

(ii) industrial companies specialising in medical imaging, digital mammogram scanning, medical systems etc.
(iii) research partners specialising in lesion, mass or other abnormal detections.