142 Chapter 8 With this thesis focused on rectal cancer we set out to develop a practical web-based tool to enable large scale testing and validation of (new) visual diagnostic classification and staging methods while at the same offering online feedback, training and teaching to a large platform of radiologists in the Netherlands as well as worldwide. Our iScore webplatform has made it possible to conduct the international diagnostic validation studies described in Chapters 2 and 5-7, and allowed us to make hands-on teaching more accessible through online resources as described in Chapter 3. In total 74 radiologists from >15 different countries participated in the projects described in this thesis. Via our studies we have thus reached a global audience that can disseminate the study results and implement these in their own daily practice. The pearls and pitfalls in rectal cancer staging and response evaluation that were identified can help to further optimize radiological staging guides, promote effective further clinical implementation, and ultimately improve diagnostic staging and reporting quality. iScore webplatform and research infrastructure In recent years, several solutions have been presented to facilitate the sharing of image data. Image biobanks and data repositories like The Cancer Imaging Archive (TICA) and Quantitative Imaging Biomarkers Alliance (QIBA) provide platforms for sharing research data but are primarily focused on quantitative image analysis and developing AI tools. Data sharing tools and platforms like Extensible Neuroimaging Archive Toolkit (XNAT) and Health Research Infrastructure‘s Translational Research IT platform (Health-RI’s TraIT-platform), focus primarily on controlling the import, archiving, processing, and safe dissemination of imaging data [1, 2]. Other web-based systems, such as Lesiontracker, ProstateCancer.ai, and Crowds Cure Cancer, contain javascript-based DICOM-viewers, such as the Open Health Imaging Foundation (OHIF) viewer. Though these viewers allow visual assessment of image data, they are fairly basic and primarily focused on allowing tools for simple AI-annotation. Due to the limited functionality devoted to visual diagnostic imaging assessment, these systems are not well suited to support clinical-diagnostic investigations. Despite recent advancements in quantitative imaging and artificial intelligence (AI), radiologists to date still rely primarily on their visual assessment of images to help guide treatment decisions. Moreover, human interpretation still plays a key role on the training and validation of AI-models. With this in mind we developed the iScore webplatform and research infrastructure that formed the basis for the studies in this thesis. iScore combines a DICOM imaging archive (to organize patient and case databases) with clinical DICOM viewing and case annotation tools that can be fully customized using integrated electronic case report forms (eCRFs). Users have their own iScore environment and cannot see the environment or annotations of other users. From a technical point of view, iScore was fully build on javascript using a MERN (MongoDB, Express, React and Node.js) stack approach, which
RkJQdWJsaXNoZXIy MjY0ODMw