Jingjing Gao1, Min Du1, Shaodong Li1, Kai Xu1, and Zhongshuai Zhang2
1Department of Radiology, Affiliated Hospital of Xuzhou Medical University, Xuzhou, Xuzhou, China, 2SIEMENS Healthcare, Shanghai,China, Shanghai, China
Synopsis
This study evaluated the diagnostic sensitivities and confidence for monitoring the progression of brain metastases using brain auto-positioning and manually positioning methods. We found that automatic slice positioning do help in finding lesions by comparing initial imaging side-by-side, but the technique does not sacrifice the diagnostic sensitivity and the diagnostic confidence when comparing with the conventional manually positioning method for brain tumor.
Synopsis
This
study evaluated the diagnostic sensitivities and confidence for monitoring the progression
of brain metastases using brain auto-positioning and manually positioning
methods.Objectives
Brain metastasis is the most
common intracranial tumors accounts for up to 40% of all adult
brain neoplasms. Patient diagnostic outcome depends on the size and the number of
metastatic brain tumors. MRI plays a vital role in lesion detection,
lesion delineation and the response to treatment. Automatic slice positioning
with artificial intelligence (AI) algorithm [1] may help to ensure
the positioning consistency between different MR scans, thus could make progression
monitoring of the metastatic tumor accurately. Therefore, the purpose of this study
is to evaluate the diagnostic accuracy for detecting the development of
metastatic brain tumors using automatic slice positioning technique, and compare
with the conventional
manually positioning manner.
Materials and Methods
This
study enrolled 32
consecutive
patients with known brain metastatic disease, who underwent contrast-enhanced
MRI examinations on two different MR scanners, e.g. one 1.5 T MR system (Siemens
Healthcare, Erlangen, Germany) with automatic slice positioning technique and
one 3T system (GE, USA) without auto-positioning technique. All the 32 patients
received MR exams more than two times on each scanner. Images acquired in the
scanner with auto-positioning are grouped as group A, and the other images were
selected as group B. The number of metastatic brain tumors and the lesion progression
was reviewed and recorded by two radiologists independently by comparing the
later scans with the initial imaging for each scanner, and the degree of
diagnostic confidence (from1 to 4) was also noted for the two groups. The
diagnostic sensitivity between the two groups was calculated as well. All the evaluations
were performed twice with an interval of 2 weeks.Result
In
this study, 32 patients (10 women and 22 men; age range, 35–76 years; mean age,
57.2 years) with metastatic brain tumors were enrolled. For group A, the
diagnostic sensitivity of the number of metastatic brain tumors was 87.4%
(153/175), if attentions were only paid on the latest scan. After
referring to initial images, the diagnostic sensitivity
of group A is raised to 98.3% (172/175), and the statistic difference
was significant (p < 0.05,
paired-Samples T test). For group B, the corresponding diagnostic
sensitivities were 88.9% (112/126) and 99.2%
(125/126), respectively; and the statistic difference was significant (p < 0.05, Paired-Samples T test). The
diagnostic sensitivities of group A and B show no significant difference (p >0.05, Chi-square test). The
average degree of diagnostic confidence between group A and B were 3.9 VS 3.8,
respectively (p>0.05, Paired-Samples
T test).Conclusions
According to this preliminary study, we found that referring to initial imaging can raise the diagnostic sensitivity for detecting the development of metastatic brain tumors. Automatic slice positioning do help in finding lesions by comparing initial imaging side-by-side, and such technique does not sacrifice the diagnostic sensitivity and the diagnostic confidence when comparing with the conventional manually positioning method for brain tumor.Acknowledgements
No acknowledgement found.References
[1] A. J.W. Van der Kouwe, T. Benner, B. Fischl, F. Schmitt, D. H. Salat, M. Harder, A. G. Sorensen, A. M. Dale, On-line automatic slice positioning for brain MR imaging, NeuroImagie 27 (2005) 222 – 230.