Model predicts fidelity of JPEG 2000-compressed abdominal CT images

A Korean research team will describe how a multiple logistic regression (MLR) model can use DICOM header information to predict the fidelity of JPEG 2000-compressed abdominal CT images.

The researchers had noticed that some of the image characteristics that affect compression artifacts are available in DICOM header information, and they hypothesized that this header information could be utilized to predict the fidelity of CT images. Results from an MLR model are promising, according to the team from Seoul National University Bundang Hospital.

Presenter Dr. Kyoung Ho Lee will share the details on how the model fared in a study involving 360 abdominal CT images.