Skip to main content
. 2020 May 28;33(4):1041–1046. doi: 10.1007/s10278-020-00348-8

Table 1.

DICOM/GPU server mean time delays for breast classification model

Category Delay time (s) N
Xception model load time 19.40 100
Single image analysis time 0.91 559
PACS upload time 0.97 100
Study analysis times
  Batch size = 1 study 86.27 10
  Batch size = 10 studies 32.45 100
  Batch size = 100 studies 26.52 100
Study stability times
  Batch size = 1 study 60.83 10
  Batch size = 10 studies 6.34 100
  Batch size = 100 studies 0.62 100

Time delays greater than 0.50 s in DICOM/GPU server when processing studies transmitted from PACS. Single image analysis time is the time required to preprocess and performance inference on a single DICOM image and does not include model load time. PACS upload time is the time required to send the output DICOM file from the DICOM/GPU server to the hospital PACS and is a measure of network delay. Study analysis time is the total time for the pipeline to process one study, starting from when a study is received on the DICOM/GPU server and ending when the model output has been successfully uploaded to PACS. Study stability time is the time required to ensure no more DICOMs will be loaded into Orthanc as part of the study