« Back to Research Home


The iPAD Project

We have created a tool called the image Physician Annotation Device (iPad; http://bimm.stanford.edu/ main/ipad), which enables radiologists and oncologists to collect and assess quantitative and qualitative aspects of tumor burden (location, measurements, and observations about the measurable disease) while performing their routine work in the clinical and research workflow. The goals of iPad are: (1) to enable structured and standards-based semantic annotation of radiology images, (2) to enable structured reporting of imaging results to be directly linked to images and even regions within images, and (3) to create a prototype for next generation image viewing workstations to support quantitative imaging. The iPad technology can enhance the efficiency of communicating results, and it provides standards for recording and sharing quantitative imaging data. We are currently applying iPAD to tackle the problems of inconsistent and incomplete collection of quantitative image features due to the uncoordinated workflow between radiologists and oncologists. We are also using it to enable extraction of semantic features from image for Content Based Image Retrieval (CBIR).
iPad is a plugin to OsiriX, an open-source image viewing platform for multimodality and multidimensional radiological image display and visualization. We chose OsiriX because it is open source, it contains a rich set of image visualization features, and it has a plug-in-architecture, so its functionality can be extended through applications such as iPad. OsiriX provides image access, display, and viewing workflow similar to commercial Picture Archiving and Communication Systems (PACS) workstations, an important factor for introducing iPad into the clinical workflow. iPad captures quantitative and qualitative image information in a structured manner from users as they view and annotate images to indicate the measurable lesions making up the tumor burden. As the user identifies lesions, measures them, describes them, and applies quantitative algorithms to them, the information is recorded in machine-readable format compliant with the Annotation and Image Markup (AIM) standard, a caBIG project in which we participate. AIM is a caBIG Silver-compliant information model for storing image information, such as lesion identification, location, size measurements, method of measurement, and other qualitative and quantitative features.
AIM is crucial for interoperability across information systems that collect and store quantitative image data; storing objective image-based measurements in AIM format will permit centers to share quantitative imaging data and catalyze discovery throughout the community. The iPad application comprises three components: (1) an information model for image annotations compliant with caBIG standards, (2) a user interface for collecting the annotations from users, and (3) a storage back-end to save annotations as XML and/or transmit them to repositories of quantitative image metadata. As the user records image information, the iPad application provides feedback to the user if annotations are invalid or incomplete. Text entered into iPad is matched to controlled terminologies such as RadLex to prevent spelling errors and to ensure that legal terms are recorded in AIM documents for caBIG compliance. iPad also checks to ensure that radiological descriptions are complete; specifically, (1) completeness: if an anatomic entity is mentioned, iPad checks to ensure that one or more observations about that entity are recorded (e.g., a mass in the liver); (2) dangling modifiers: if a modifier of an observation is mentioned, iPad checks to ensure that an observation is also mentioned; and (3) lesion name: the user is prompted to name each measurable lesion with a unique label (e.g., "Lesion 1"); this is critical information that applications can use for lesion tracking. Ultimately, we anticipate enabling researchers using iPAD to more easily and robustly exploit quantitative imaging in clinical trials and to use this information to evaluate treatment response.