![](bg6.png)
– it takes significantly longer, in fact nearly twice as
long, to perform two CT abdomen exams on two
patients (due to various changeover and
documentation times) than to perform two scans on
one patient. As a result, from an operations point of
view, it may not be accurate to say that one
technologist who has performed two billable exams
on the same patient has been as efficient as another
technologist who has performed two exams on two
different patients (assuming everything else is
comparable).
Distinguishing between billable and performed
exams may or may not have a significant impact
depending on the study mix performed at a given
institute. For instance, in our dataset, for a certain day,
there were 891 total billable exams based on HL7
messages whereas there were only 829 exams based
on DICOM. In general, the difference was between 5-
10%.
It should be noted that the ability to use the
accession count from DICOM to determine
operational volume depends on the particular
hospital’s workflow. Some hospitals, including the
one in our study, typically scan all images under a
single accession number, push them to the PACS, and
then either split, or link the images to the accession
numbers associated with the different orders.
Alternatively, the splitting can happen at the modality
workstation itself, in which case two accession
numbers (in the CT abdomen-pelvis example) will be
seen in DICOM. In this case, the reporting engine will
need to perform some logic, such as ‘same patient,
same acquisition times for different accession
numbers’ to determine which studies should be
merged for operational reporting purposes.
3.1.2 Exams with Multiple Modalities
Studies where multiple modalities are involved are
identified using the same accession number. A few
examples of such studies are PET-CT, PET-MR and
interventional radiology exams (which may often
involve XR and/or ultrasound and/or CT). In each
instance, the complete exam will often be billed under
a single accession number, although from an
operations point of view, two (or more) resources
were utilized to perform the exam. Images acquired
from different modalities can be determined using
DICOM Source AE Title tag. These exams need to be
correctly accounted for when determining relevant
metrics (such as operational volume, technologist
productivity and machine utilization).
3.1.3 Shared Resources
It is common practices for different departments
within radiology to share resources. For instance, a
PET/CT machine may be used mainly for PET scans,
but due to low PET volumes, the CT department may
often make use of this resource to perform certain CT
exams during busy periods. If PET and CT are
different cost centers, PET and CT volumes will be
shown separately for each departments, but for
machine utilization, both volumes need to be
accounted for.
3.1.4 Manual vs Automated Timestamps
Care must be taken when calculating various
turnaround times using timestamps. For instance, per
Figure 3, scan duration is calculated using times from
the DICOM header. These times will often be reliable
since these are machine generated timestamps. On the
other hand, depending on the clinical system, exam
start and end HL7 messages may be trigged manually.
This flexibility is provided often for valid practical
reasons, for instance, after acquiring all images for a
CT exam, a technologist may have time to ‘end exam’
in the system only after scanning a new emergency
patient (i.e., back-time the value for the previous
exam). Similarly, ‘start exam’ time may be entered
manually and may depend on the individual
technologist – some technologists may consider the
start of exam to be when they call the patient from the
waiting room, some may consider the start to be when
the patient walks into the scanning room, while others
may consider start of the exam when the patient is on
the scanner itself. As such, it is important to
standardize the terminology associated with granular
workflow steps. If the workflow can be standardized
so that all technologists start the exam when they go
to get the patient from the waiting room, then the time
difference between ‘patient arrived’ and ‘exam start’
HL7 messages will accurately reflect patient wait
time while the difference between ‘exam start’ HL7
message and ‘first DICOM image’ timestamp will
show the overhead associated with getting the patient
on the scanner (which could be significant for obese
and/or immobile patients) and adjusting the scanner
settings prior to image acquisition.
3.1.5 Same Information in HL7 and
DICOM
Some data can be available in both HL7 and DICOM.
Either source can be used if the value in both sources
is the same (such as the accession number), but there
could be instances where same data is entered slightly
HEALTHINF 2017 - 10th International Conference on Health Informatics
62