| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Editorial |
1 Department of Radiology, University of Aberdeen, Aberdeen, UK and 2 Technical University Berlin, Berlin, Germany
Computer-aided detection (CAD) has been developed over the last 20 years following advances in computer technology and software programs as well as improved image capture techniques with higher resolution. Using technology initially designed for military defence programmes and observatories, applications in medical imaging have evolved. The aim is often to improve reader sensitivity and specificity in a particular application, such as mammography, but other benefits are becoming apparent. A meeting was held at the British Institute of Radiology to bring together the scientific, industrial and medical communities to discuss the current position. This special edition of the British Journal of Radiology publishes the papers presented at the meeting.
Professor Kunio Doi presents a comprehensive review of CAD in medical imaging. He draws from his extensive experience at the University of Chicago and concentrates on pulmonary nodule detection on radiographs, low dose CT and high resolution CT, diagnosis of these nodules, quantitative analysis of diffuse lung disease and detection of intracranial aneurysms on magnetic resonance angiography. In particular, CAD techniques such as artificial neural networks, difference imaging, linear discriminant analysis, and morphological and three-dimensional (3D) selective enhancement filters are described with regard to their applicability. An extensive reference list gives the reader the opportunity to further explore the topics covered in this paper.
Dr Sue Astley describes CAD algorithms and the nature of prompting, how prompts are placed on images, and how researchers should assess whether a prompt has been correctly placed. She describes the principles that should be applied when evaluating algorithm performance and in the evaluation of CAD systems together with reader performance; for example, the impact of training is emphasised in order to achieve stable performances in trials.
Dr Paul Taylor discusses three related studies on CAD in mammography using R2 technology. His initial Health Technology Assessment (HTA)-funded study showed no effect of CAD when assessed with 50 readers and 60 cancer cases. In the second part, 35 readers read 40 more subtle cases that had all been prompted by CAD and, although there was some improvement in reader performance, it was not statistically significant. Using an improved R2 algorithm, a prospective study was undertaken double reading screening mammograms with CAD. CAD resulted in a greater number of cases going for arbitration and a small improvement in cancer detection. He concludes that his group found no evidence that CAD was effective but acknowledges that it may be in a different setting.
The article from Eugenio Alberdi and colleagues examines the effects of CAD prompts on performance, comparing the negative effect of no prompt on a cancer case with prompts on a normal case. They show that no prompt on a cancer case can have a detrimental effect on reader sensitivity and that the reader performs worse than if the reader was not using CAD. This became particularly apparent when difficult cases were being read. They suggest that the readers were using CAD as a decision-making tool instead of a prompting aid. They conclude that "incorrect CAD can have a detrimental effect on human decisions". This is clearly an issue of which radiologists need to be made aware when using CAD in clinical situations. In cases where a cancer is suspected and there is no CAD prompt, it may be that a second human reader should arbitrate so that these more subtle cancers are not overlooked.
The manufacturer's perspective is given by Jimmy Roehrig from R2 Technology, Inc. Improving the specificity of the CAD algorithm, training the readers, presenting more information to the reader, improving actual performance from potential performance, and decision support tools are all discussed. Like other authors, he also recognises that training radiologists in the use of CAD is of critical importance. Roehrig emphasises that more work is required to understand why radiologists ignore correct prompts on cancers.
Rafael Weimker, from Philips Research Laboratories Hamburg, presents an excellent summary of algorithms developed and tested on lung nodules, emphasising potential applications in early cancer detection and diagnosis of a nodule based on morphology and sequential volume changes. He points out that technical improvements in spatial and temporal resolution in CT thoracic image acquisition provide a good prerequisite for improving algorithmic performance in computer-assisted pulmonary nodule detection. This should provide a good basis for early detection of cancer and also for reducing the rate of biopsies. Detection of lung nodules combined with 3D volumetrics is outlined in detail. A brief description is given to algorithms to automate image registration between previous and follow-up CT scans, enabling not only support for diagnosis but also monitoring the response to oncological therapy.
CAD in CT colonography is discussed by Luca Bogoni from Siemens Medical Systems and colleagues and suggests that 90% of medium and large polyps can be detected using this Siemens software. This is similar to other published series. He points out, however, that a direct comparison of different CAD systems is difficult given that validation methods, patient population as well as technical and medical steps taken during the procedure may vary in CT colonography. The case for virtual colonoscopy in screening for colon cancer is currently being tested in the UK and elsewhere, so it is timely that CAD is being developed in this area.
CAD is becoming a clinically useful tool in a variety of areas and modalities. More evaluation is required and an awareness of ongoing developments is important. The imaging community should support the investigation of this important technology as it very likely to become part of our daily practice in the near future.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
