This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Computer applications, Liver, Interventional vascular, MR-Functional imaging, Image manipulation / Reconstruction, Segmentation, Radioembolisation, Computer Applications-General, Cancer
Authors:
S. Thoduka1, P. Schilling1, G. Chlebus2, A. Schenk2, N. Abolmaali1; 1Dresden/DE, 2Bremen/DE
DOI:
10.1594/ecr2017/C-2917
Methods and materials
Hepatic Gd-EOB-DTPA-enhanced late phase MRI scans of 38 patients consecutively registered for SIRT with unresectable primary or secondary liver-tumours were obtained as part of the clinical routine pre-SIRT protocol.
Volumetry of liver and tumour tissues in both liver lobes performed routinely by an expert radiologist (with 20 years of experience of abdominal imaging) for SIRT (SIR-Spheres®; Sirtex Medical Limited,
Sydney,
Australia) dose calculation were compared to volumes segmented by a novice radiology resident.
Volumetries were done using the interactive contouring software EclipseTM(Varian Medical Systems).
The middle hepatic vein was used to divide the liver into the left and right lobes.
Both observers were blinded to the segmentations of the other observer.
Bland- Altman graphs plotting the mean against the difference of volume measurements were generated to analyse the correlation between the volumetry performed by the two observers [6].
The Dice similarity coefficient,
as a degree of spatial overlap between the segmentation lines of the expert and the novice,
were calculated on the EclipseTM contouring software [7].
The statistical analysis was performed using the SciPy and NumPy libraries in Python.
Table 1. Patients’ characteristics
Demographic
|
Total (n=38)
|
Number of patients with:
· Hepatocellular carcinoma
· Cholangiocarcinoma
· Metastatic colorectal cancer
· Metastatic breast cancer
· Metastatic melanoma
|
23
8
5
1
1
|
Median age (years)
|
65
|
Sex (M/F)
|
29/9
|
Median weight (kg)
|
79
|
Median height (cm)
|
170
|