DEVELOPMENT AND VALIDATION OF TUMOR VOLUMETRIC TOOLS
TUMOR PERFUSION ASSESSMENT METHOD AND TOOLS
TOOLS FOR MULTI-MODALITY SPATIO-TEMPORAL IMAGE ANALYSIS
ACCREDITATION LAB FOR CLINICAL TRIALS IMAGING (ALTI)
IMAGE VIEWING AND ANALYSIS PLATFORM
IMAGE SHARING NETWORK/ARCHIVE AND REFERENCE IMAGE BANK
DEVELOPMENT AND VALIDATION OF TUMOR VOLUMETRIC TOOLS
Project Leaders: Drs. A. Fenster and G. Parraga (RRI)
A single method to determine tumor response is not universally accepted; however, a new definition for solid tumor responses based on a single linear summation of a small number of target lesions, termed Response Evaluation Criteria in Solid Tumors (RECIST), has been adopted for clinical protocols. This linear summation is both rapid and reproducible which facilitates its use in clinical trials. RECIST represents a step forward from previous response criteria because it takes into account differences in scan thickness, minimum tumor sizes, and frequency of evaluations. Nevertheless, despite broad adoption of this technique, there is a growing appreciation that response measured by RECIST criteria does not necessarily correlate with clinical benefit, particularly with newer molecular targeted agents where biologic effects may not be manifest as tumor shrinkage. There is a need to develop radiological response phenotypes that are more closely linked to clinical outcomes that can evaluate response to treatment more reliably than current imaging methodologies. The capacity to use these new phenotypes along with established clinical end points is now critical for imaging in clinical trials and will continue to gain importance as newer targeted molecular therapies augment and perhaps replace traditional cytotoxic treatments. The extension of these same phenotypes to preclinical model systems will provide a unique method to translate validated preclinical imaging biomarkers directly into clinical trials.1 Many outstanding questions remain related to the best methods used to measure tumour response to therapy. What happens to tumours when they respond to anti-angiogenesis therapies? What differentiates those tumours that respond over longer periods of time and those that do not? How do such tumours change in shape, size and texture as the disease changes in response to changes in the vascular networks that serve the tumour? For these reasons improved methods and biomarkers to measure morphologic tumour response are clearly required. Such techniques can then be incorporated into a reporting tool capable of RECIST, WHO, volumetric and feature reporting.
TUMOR PERFUSION ASSESSMENT METHOD AND TOOLS
Project Leaders: Dr. M. Haider and Dr. I. Yeung (UHN)
A critical and early process in cancer progression is angiogenesis resulting in new blood vessel formation. Without recruitment/generation of new vessels, cancers are unable to progress beyond the size limits for diffusion of oxygen and glucose. As tumors grow and develop hypoxic regions, pro-angiogenic growth factors (i.e., VEGF) are secreted resulting in the development of new vasculature, permitting further local tumor growth and also enabling local invasion/tumor migration as well as extravasation and metastasis. In addition to facilitating tumor growth, angiogenesis is a marker of both tumor biology (i.e., tumor aggressiveness) and environmental stress (i.e., hypoxia) that may influence response of the tumor to treatment. The availability of small molecule inhibitors directed at angiogenic pathways provides a method to interrupt the angiogenic process and is improving outcomes in tumors such as renal and colorectal cancers. While static, contrast enhanced imaging (CT or MRI) can provide some information regarding tumor blood supply, tumor perfusion imaging with dynamic contrast enhanced CT or dynamic contrast enhanced MRI (DCE-CT/MRI) can provide a richer assessment of tumor vascularity and blood flow. In addition, DCE-CT/MRI, through direct and quantitative assessment of tumor blood flow, is potentially a valuable biomarker of response to anti-angiogenic agents as well astraditional therapies. We have prior experience in studying the prostate, cervix and brain.8,9,10,11,12 While analysis of DCE-CT/MRI imaging with derivation of metrics of blood flow is well described in the literature, these analyses are typically time consuming, requiring post-imaging processing with specialized software and mathematical techniques. As a result, DCT-CT/MRI is a potentially important imaging biomarker, which requires development of clinically useful tools for quantitative assessment.
TOOLS FOR MULTI-MODALITY SPATIO-TEMPORAL IMAGE ANALYSIS
Project Leader: Dr. D. Jaffray
Multiple imaging modalities are used to characterize the anatomic and functional characteristics of the therapy target and normal tissues to be spared, and to assess target response to therapy. The current standard for response assessment in solid tumors is the RECIST criteria.20 As discussed in IPP-Trials Project 1.1, RECIST is a uni-dimensional morphometric measurement and consequently does not account for functional changes within the tumor that may signify response to therapy.21 Furthermore, response assessment based upon single-parameter image-based quantification of change from baseline (i.e., standard uptake value in PET) does not fully characterize the target and its temporal evolution in response to therapy. Recent investigations are employing voxel-to-voxel comparisons of image signal to evaluate the disease state. These comparisons are applied across modalities (e.g., T1-T2 on MR) and across time (time dependent FDG-PET,23 MR diffusion24). These methods exploit recent advances in hardware and software image registration by characterizing tissue on a voxel-by-voxel basis with a multi-dimensional vector representation of the registered imaging data. A major challenge to these approaches, when applied over a time-course or between modalities, is the uncertainty in geometric correspondence between temporal and/or cross-modality voxel pairs, arising from deformation, motion, and/or growth/shrinkage of the therapy target. Spatio-temporal image and voxel-based metrics are to be developed to provide a confidence measure of geometric correspondence.
ACCREDITATION LAB FOR CLINICAL TRIALS IMAGING (ALTI)
Project Leader: Dr. H. Keller (UHN)
Performance assessment is an important component of the IPP-Trials. Imaging for clinical trials is challenging as outcomes trials based on image-based surrogate endpoints will require the combined effort of a larger number of participating institutions in order to achieve the statistical power to accept or reject a hypothesis in a timely manner. This requires that strict procedures and protocols (i.e., Standardized Operating Procedures, SOPs) be in place that specify patient handling, acquisition and data analysis for a particular clinical trial. In addition, tools developed in Projects 1.1, 1.2, and 1.3 for image analyses need to be independently validated.
IMAGE VIEWING AND ANALYSIS PLATFORM
Project Leader: J. Adziovsky (JDMI) / ClearCanvas
To establish an image-viewing platform that will provide core functionality for clinical trial image management and be DICOM compliant, affordable, and compatible with clinical radiology environments. We will establish accreditation and commissioning procedures to enable the deployment of the imaging viewer and analysis tools across all partnering sites. The UHN Medical Imaging team and ClearCanvas have gathered the talents of a number of engineers from the medical imaging industry with over ten years of experience in product development. Iterations of image viewing development regularly seek user feedback to allow high priority change requests to be implemented in as little as 1 month. This team of developers is to be seen not as a product vendor, but rather an engineering consultancy that collaborates with the OICR Imaging investigators. The UHN Medical Imaging team and ClearCanvas are committed to reducing the cost of healthcare software and at the same time, giving users unprecedented freedoms in how they may use the software. The adoption of the Open Source license allows the products developed by the team to be distributed free of charge, without limits. The source code is available publicly, and may be modified freely to address users’ particular needs.
IMAGE SHARING NETWORK/ARCHIVE AND REFERENCE IMAGE BANK
Project Leader: J. Adziovsky (JDMI)
To establish a secure network for IPP-Trials Investigators to allow the sharing of images either for the purposes of image response assessment (in the case of clinical trials imaging) or imaging research and development. The Network activities will proceed in a staged fashion:
- Distribution of image-analysis tools that reflect the workflow of the SOP to multiple institutions to operate on images contained within their PACS environment.
- Establishment of a common server containing a database of images that were employed in the certification/accreditation of the SOP tools, including calibration phantoms.
- Development of image-analysis tool registry that records the use of the specific tool and the institution at which it was applied
- Development of a peer-to-peer scheme for accessing cases that were analyzed using these tools for subsequent data-mining studies and tool development
It is anticipated that the central reference image bank will be populated by the clinical images used to develop and validate the imaging tools developed in Goal 1. “Value added” imaging information obtained through the validation process (i.e., gold standard radiologist reads, volume segmentations, clinical data such as treatment and clinical response correlates) will be included with images. All subject data, images and image measurements will be recorded in a database on a secure, password-protected server that can be accessed by password. It will be the first such tumor image database in Canada that can be accessed by cancer researchers for image research.
More about IPP-Trials: