Over the past decade the performance of major clinical imaging platforms has greatly improved. Now, much more rich quantitative data are routinely acquired in many important clinical studies, but very little clinical management is based upon quantitative endpoints. Quantitative imaging in a clinical setting involves much more than observing an object on an x-ray. Truly quantitative imaging requires monitoring many aspects of the data acquisition process: How was the image acquired; how was that process quality controlled; was the instrument calibrated; what was the process of doing the quantitative measurement; what tool was used and with what controls? All of these parameters influence the variance in a quantitative measurement of a clinical image. In theory, it could be mandated as to how to address these parameters with a government agency tasked with monitoring to ensure full quality compliance. Another strategy is to develop protocol and practice consensus by working with the clinical and scientific communities through the peer-review process associated with the publication of clinical results and to use this as the change agent to improve the conduct and reporting of quantitative clinical imaging. The hypothesis is that if the full DICOM files of all clinical images associated with a manuscript submission are provided along with the imaging research manuscript, then the quality of the submitted images can be independently reviewed and verified. Knowing that their imaging data will be reviewed along with the manuscript provides a powerful incentive for potential authors to ensure that all of the submitted DICOM cases are of appropriate quality. This conceptual strategy has received broad support from the editors of a number of high-quality medical publications. However, despite this support, previously there had been no practical way to implement such an image submission strategy.
Enter Interactive Scientific Publishing. This is an online software and database infrastructure that allows quantitative image files, including DICOM files, to be accessible to the scientific community for inspection and analysis. The Optical Society (OSA) in partnership with the U.S. National Library of Medicine retained the imaging software and visualization expertise of Kitware, Inc., to develop this infrastructure to host quantitative imaging data for public access. This OSA-hosted environment is called Interactive Scientific Publishing (ISP). This resource not only allows the image to be inspected, but an array of quantitative analysis tools, including an open-source lesion sizing tool, are also available in the hosted environment, allowing readers to reanalyze and independently verify conclusions derived from the submitted image data. This extends the peer-review process and will be a powerful resource in advancing progress in quantitative imaging.
In addition to enriching scientific dialogue, ISP also enables scientists to rapidly explore contributed data sets and rapidly consider new hypotheses and ideas. Allowing a reader to easily download a difficult-to-acquire and state-of-the-art data collection contributed by an ISP author permits the reader to skip potentially years of data collection effort and expense, thereby accelerating and making more efficient scientific research.
The current supplemental issue of Optic Express is a dedicated issue that explores the utility of this new Interactive Scientific Publishing resource. The use-case for this effort is the vibrant field of lung cancer drug discovery. Lung cancer is the world’s most lethal cancer and the focus of intense pharmaceutical interest. It may not be obvious why this would be an attractive point of departure until one considers the normal function of the lungs. Specifically, they function to exchange gases. Thus, in contrast to virtually every other part of the body, early changes in the lung suggestive of cancer are surrounded by air and not by fluid or solid tissue. Therefore the signal-to-noise contrast in the lung is particularly favorable in allowing the fine anatomy of the lung to be clearly visualized at high resolution. This also allows for more effective separation (or segmentation) of the normal from the cancerous tissues, providing precise quantification of important characteristics of such structures.
Contributing to this supplement is an array of image processing and measurement scientists from academia, industry, and the government. Several submissions explore, essentially for the first time, aspects of foundational work that define the sources of variance in the use of computed tomography to quantify volume change under various scanning conditions. The approach to addressing these issues is to use objects of defined shape and density (termed “phantoms”), so that objective evaluation of imaging performance can be measured. While this is an artificial approach that may deviate from the actual clinical situation, it is a necessary starting point for tackling the nature and degree of measurement variance within CT imaging. Leading scientists from the U.S. Food and Drug Administration, the U.S. National Institutes of Health, and the U.S. National Institute of Standards and Technology have contributed in this regard. In addition, dimensions of this challenge have been evaluated by a consortium of imaging scientists organized by the Radiological Society of North America. Their consortium is called the Quantitative Imaging Biomarker Alliance, and this group has been deeply engaged in responsibly applying major imaging modalities as measurement tools. The members of the Quantitative Imaging Biomarker Alliance and the editors of this supplement agree that many principles for validating quantitative imaging in specific modalities such as CT imaging will be shared across the other imaging modalities, such as magnetic resonance imaging, ultrasound, and optical coherence tomography.
Much of the preliminary clinical quantitative imaging background data and some of the imaging resources reported in this issue of Optics Express were developed in conjunction with a series of Workshops on Image Processing and Lung Cancer Drug Discovery, sponsored by the Prevent Cancer Foundation over the past seven years. The editors of this supplement have been on the Organizing Committee for those Workshops and worked together to produce the Optical Society Monograph published in 2008 entitled “Quantitative Imaging Tools for Lung Cancer Drug Assessment.”
A major accomplishment since the previous Monograph has been the development by OSA and Kitware, Inc., of the ISP system. Therefore all of the images for this submission are available for digital evaluation using the capabilities of ISP. In addition to the peer-review aspect of this resource, which was already discussed, ISP supports the repurposing of DICOM imaging files. This ability to download and aggregate image files from published research papers may also allow additional new imaging research questions to be addressed.
OSA is providing an important new infrastructure for image research to the user community. The reports in this supplement contain very important validating information about the performance of quantitative imaging in this volume CT setting with access to a wealth of supporting data. We anticipate that progress will occur rapidly in this dynamic field, and we are confident that the capabilities provided by Interactive Scientific Publishing will be a great asset in accelerating the pace of progress in quantitative imaging. We hope that this volume is a useful introduction for many to this array of new resources for more timely and successful drug development research.