Digital Mammography: 8th International Workshop, IWDM 2006, Manchester, UK, June 18-21, 2006, Proceedings

This volume of Springer’s Lecture Notes in Computer Science series records the proceedings of the 8th International Workshop on Digital Mammography (IWDM), which was held in Manchester, UK, June 18–21, 2006. The meetings bring together a diverse set of researchers (physicists, mathematicians, computer scientists, engineers), clinicians (radiologists, surgeons) and representatives of industry, who are jointly committed to developing technology, not just for its own sake, but to support clinicians in the early detection and subsequent patient management of breast cancer. The conference series was initiated at a 1993 meeting of the SPIE in San Jose, with subsequent meetings hosted every two years by researchers around the world. Previous meetings were held in York, Chicago, Nijmegen, Toronto, Bremen, and North Carolina. It is interesting to reflect on the changes that have occurred during the past 13 years. Then, the dominant technology was film-screen mammography; now it is full-field digital mammography. Then, there were few screening programmes world-wide; now there are many. Then, there was the hope that computer-aided detection (CAD) of early signs of cancer might be possible; now CAD is not only a reality but (more importantly) a commercially led clinical reality. Then, algorithms were almost entirely heuristic with little clinical support; now there is a requirement for substantial clinical support for any algorithm that is developed and published. However, upon reflection, could we have predicted with absolute certainty what would be the key questions to be addressed over the subsequent (say) six years? No! That is the nature, joy, and frustration of research. There are more blind alleys to explore than there are rich veins that bring gold (in all senses of that analogy!). What are the current preoccupations? What are currently the ideas that we believe will bear handsome fruit over the next 20 years? These are reflected in the programme, and in the choice of invited speakers. However, it is first important to realize that what have been identified as the major challenges over the past 13 years continue to be challenges: robust, reliable, efficient algorithms for CAD, segmentation, registration, and texture analysis await definitive solution, as they do in image analysis generally (and mammography poses additional challenges). Second, the challenges of delivering the technology effectively to end-users remain unmet: what are the optimal prompts? How do you deliver CAD in large rural areas? How do you deliver mammographic image analysis over the emerging Grid? How do you integrate film-screen mammography with full-field digital? How do you fuse mammography with other imaging modalities, such as MRI, ultrasound, and PET. . . These observations explain about half of the sessions, as they did at previous meetings (though we all believe we have made progress)! Like great music, however, for all the increasingly understood and recurrent themes there are some newer ones that press for attention! Among these we can clearly identify tomosynthesis—subject of an invited address by Prof. Dan Kopans at UNC in 2004 and now increasingly a commercial and clinical reality—and the estimation and analysis of breast density—again, the subject of an invited address by Prof. Norman Boyd at UNC in 2004. However, with the exquisite hindsight of reflection on the past we will—six years hence—be able to identify a number of other emergent themes, although not only are we not able to see them clearly but would probably reject them as marginal! These might be fusion of mammography with other modalities and x-ray imaging techniques that currently seem avant garde. A successful conference is a blend of inspired organization, financial support, scientific insight; but, ultimately, the quality of the papers that were submitted. Two of us (Sue, Mike) were charged by the IWDM Scientific Committee to organize a meeting in the UK. We invited four-page outline papers, as opposed to the paragraphs that had previously been submitted. We believe that this simultaneously increased the quality and decreased the number of submissions. Each four-page abstract was assessed independently by at least two, often three, members of the Scientific Committee, and the final eight-page submissions were assessed independently by at least two members. We believe that the final proceedings, which you have in your hand, constitute a state-of-the-art statement of mammographic image analysis, its underlying physics, and clinical pullthrough. The invited addresses by Julietta Patnick—director of the UK national breast screening programme—and Profs. Andrew Maidment—digital mammography and tomosynthesis—and Etta Pisano—author inter alia of the influential DMIST trial—were not included in the published proceedings; but their influence on the future of the research of the community, and its pull-through into practice, cannot be over-emphasized. Finally, in keeping with the multi-disciplinary nature of the meeting, the meeting was supported by sponsors and there was an excellent industrial exposition, pulled together by Reyer Zwiggelaar. The timely and efficient production of the reviews, final versions, arrangements, etc. depended fundamentally on Dr. Chris Rose and the remarkable CAWS website at Manchester.