Differences have implications for assessment, quality assurance, and training
Healthcare imaging technologies and options are continually evolving, and their applications in radiology can be puzzling. One area of confusion is the respective roles of Quantitative Image Analysis (QIA) and Computer-Aided Diagnosis (CAD). Recently we collaborated with several colleagues to compare and contrast the two imaging technologies and advance the work that has been accomplished by the Computer Aided Detection in Diagnostic Imaging Subcommittee (CADSC) of the American Association of Physicists in Medicine (AAPM).
CAD systems, which have been around for about 15 years, incorporate pattern recognition and data analysis capabilities. Computer-aided detection (CADe) systems are intended to mark regions of an image that might reveal specific abnormalities and alert the clinician to these regions during image interpretation. Computer-aided diagnosis (CADx) systems provide to the clinician an assessment of disease, disease type, severity, stage and progression.
QIA, on the other hand, is fairly new. Using computerized tools, it extracts quantitative imaging biomarkers from medical images. A quantitative imaging biomarker is an objectively measured characteristic derived from an in vivo image as an indicator of normal biological processes, pathogenic processes or a response to a therapeutic intervention. Quantitative imaging includes the development, standardization and optimization of anatomical, functional and molecular imaging acquisition protocols, data analyses, display methods and reporting structures.
Similarities and differences in QIA and CAD
The two technologies have some similarities. They both provide advanced image analysis techniques for clinicians. Both commonly use computer methods to extract features. And both emphasize appropriate image acquisition protocols, display methods, training and reporting. However, there are important differences that must be considered.
CAD essentially relies on the ability to make reproducible, quantitative measurements from medical images and combine them into a score or a marker to help clinicians provide a diagnosis. The emphasis is on how the CAD outputs aid the clinicians in decision-making, diagnosis, treatment planning, treatment response monitoring or outcome prediction.
In QIA the emphasis is on the extraction of biomarkers and on establishing a specific imaging biomarker’s association with a disease condition. QIA can go beyond the anatomical view and into the molecular level. Examples of applications of the two technologies can be seen in the presentation delivered at RSNA.
Through our research, we determined that the differences and similarities between CAD and QIA have considerable implications for assessment, quality assurance and training. These consequences are also documented in the presentation.
CAD and QIA share many common components and both leverage a richness in medical images that has yet to be fully tapped. There is a natural synergy between the two and we expect that methodologies developed in one field are likely to be applicable in the other. Also, both techniques can benefit from standardized assessment, QA and user training procedures. We hope this study increases the coalescence of new ideas and standardized approaches in these closely related fields. #RSNA #imaging #radiology
Zhimin Huo, PhD, is a researcher in Carestream Health’s Research & Innovation Laboratories
The following people also contributed to the study: Berkman Sahiner, PhD and Nicholas Petrick, PhD, US Food and Drug Administration; Samuel G. Armato III, PhD, University of Chicago; Heang-Ping Chan, PhD, University of Michigan; Ronald M. Summers, National Institutes of Health, Clinical Center