Keywords:
Lung, Pulmonary vessels, Artificial Intelligence, CT-Angiography, CAD, CT, Computer Applications-Detection, diagnosis, Contrast agent-intravenous
Authors:
N. Hendriks1, J. J. Zigterman2, M. Roelofs1, J. Nijboer-Oosterveld1, E. De Boer1, M. F. Boomsma1; 1Zwolle/NL, 2Zwolle, Overijsel/NL
DOI:
10.26044/ecr2019/C-2541
Results
We included the available 100 CTPA-patient scans (51% female,
average age 62,3 years).
According to the reference standard 17% was positive for pulmonary embolism.
Interobserver agreement for the reference standard was K=0,97.
Interobserver agreement for evaluating the CAD-detections on a per-CAD base on HIR (n=416) and MBIR (n=475) was respectively K=0,66 and K=0,59.
A 100% sensitivity was perceived in both reconstruction methods.
HIR reconstruction shows a higher outcome with respect to the MBIR reconstruction concerning specificity and PPV.
A significant difference was not evinced.
The results of the 2x2 table are found in table 1.
The average SNR within the HIR- and MBIR scans were respectively 18,3 and 27,6 (p<0,05) (figure 3).
The SNR in MBIR scans increased 52% with use of MBIR compared to HIR.
The average CNR within the HIR- and MBIR scans were respectively 17.0 and 25.8 (p<0,05) (figure 4).
The CNR in MBIR scans also increased with 52% by use of MBIR.