Toshimasa James Clark MD, Presenter: Nothing to Disclose

Suresh Maximin MD, Abstract Co-Author: Nothing to Disclose

Sooah Kim MD, Abstract Co-Author: Nothing to Disclose

Carolyn Lee Wang MD, Abstract Co-Author: Nothing to Disclose

Mariam Moshiri MD, Abstract Co-Author: Consultant, Reed Elsevier Author, Reed Elsevier

Puneet Bhargava MD, Abstract Co-Author: Editor, Reed Elsevier

Tao Li MD, PhD, Abstract Co-Author: Nothing to Disclose

Algorithm-based characterization of hepatic lesions is well defined by the LI-RADS criteria for a small subset of studies. LI-RADS is not applicable to patients without underlying liver disease or on MR performed with hepatobiliary contrast agents. We therefore created a web-based application to categorize hepatic observations along with semi-automated report generation of the differential diagnosis in descending order of probability. It will apply to livers with or without underlying disease, imaged with multiphase CT or MR, with any contrast agent.

We developed an algorithm encompassing all hepatic observation signal intensities and attenuations on all phases where applicable, building upon prior extant work. We then developed a free Javascript web application implementing both our algorithm and differential diagnosis generator. The decision tree is based on current literature, can be updated as consensus evolves, and will include a comprehensive set of benign and malignant hepatic entities. It will also allow input of other relevant data, e.g. prior studies or history of malignancy.

There exists literature supporting definitive characterization of various hepatic entities based on imaging characteristics on multiphase studies, yet we are unaware of any decision support applications that can be used in imaging of both non-cirrhotic and cirrhotic livers. This application’s evidence-based algorithm can be used for decision support, structured report generation, and as a learning tool.

We have developed a free web application that can be used to characterize lesions and auto-generate a differential diagnosis for CT or MR of both non-cirrhotic and cirrhotic livers, regardless of contrast agent choice. Given the increasing complexity of hepatic imaging, our goal is to provide decision support that will be useful as a learning tool as well as help radiologists become more consistent, efficient and accurate in the interpretation and reporting of hepatic lesions.

http://abstract.rsna.org/uploads/2014/14012420/14012420_8a4p.jpg

Clark, T, Maximin, S, Kim, S, Wang, C, Moshiri, M, Bhargava, P, Li, T, Universal Decision Support Application and Semi-automated differential Diagnosis Generation System for Liver CT and MR Imaging.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14012420.html