Comparative performance investigation of DICOM C-STORE and DICOM HTTP-based requests

Increasingly, physicians have to access clinical images distributed over multiple healthcare organizations. To this end, two DICOM protocols may be used: a regular DICOM C-STORE transaction or an HTTP-based DICOM request such as WADO or STOW. A major problem of the DICOM C-STORE transaction is that it is inefficient to transfer DICOM data sets that consist of thousands of DICOM objects (such as functional MRI data set) because of the large number of negotiations involved in the transfer. We compare the performances of C-STORE transactions with the STOW HTTP-based protocol, and show that the STOW protocol can divide the transfer time by about 50 when compared to a DICOM C-STORE transaction for studies that consists of thousands of DICOM objects.