Page 53 - Read Online

Kościuszko et al. Hepatoma Res 2021;7:51 https://dx.doi.org/10.20517/2394-5079.2021.17 Page 11 of 16

[50]
by Souzaki et al. , utilised an augmented reality navigation system in laparoscopic surgery for two cases
(Wilms’ tumour and bronchogenic cyst). In the hepatoblastoma case, the augmented reality navigation
system was used for open surgery. The authors’ main hindrance was the increasing difference between the
AR image and reality with the operated organs’ displacement and compression. The system could not follow
organ deformations. Marescaux et al. reported the first real-time AR laparoscopic adrenalectomy. That
[68]
[69]
system used manually assisted deformable registration. Haouchine et al. proposed another solution for
rigid registration and published a paper the same year as Souzaki et al. . Haouchine et al. suggested a
[51]
[69]
real-time, physics-based system able to automatically register deformable organs using stereo endoscopy. It
was shown in 2009 that stereoscopic endoscopes can take accurate anatomical measurements .
[69]
Because excellent knowledge of the liver’s anatomy is crucial for successful resection (or transplantation)
[65]
and protection of the vascular inflow and outflow to the remaining liver , it is believed that 3D
visualisation may help in the surgical treatment of liver tumours.

All of the reviewed papers speak favourably of computer-aided surgery. However, Warmann et al.
[55]
claimed that, owing to 3D reconstructions, 12 out of 14 hepatoblastoma patients could avoid
transplantation. Dong et al. also stated that 3D surgical planning allowed more comfortable and safer
[48]
dissection. The authors declared that this technique should be applied especially in large or complex liver
tumours.

The critical, post-resection volume of the liver is called the future liver remnant (FLR). This is an essential
concept because it was proven that, in the case of FLR < 20%, more than half of patients develop severe
postoperative complications, while, for FLRs > 20%, the risk of liver failure is 13% . Vauthey et al. stated
[71]
[70]
that the minimal safe FLR volume is 25% for cases of extended right-sided hemihepatectomy. Shoup et al.
[72]
also found that an FLR volume that is less than 25% is a significant risk for hepatic dysfunction. There is no
consensus regarding the minimal FLR that is safe for liver surgery in children.
Most of the studies used medical imaging datasets obtained from CT studies. Only Souzaki et al. ,
[50]
Warmann et al. and Wang et al. reported the use of data from MRI studies. This is because MRI images
[55]
[60]
are less accurate, have a higher signal-to-noise ratio than CT and have lower image contrast differences
[49]
between tissues. Fuchs et al. noted that the spatial resolution of CT allows for more detailed
reconstructions when compared to MRI. In addition, the thickness of the slices is essential for 3D
reconstructions. The problem also lies in inhomogeneous signal intensities. Those obstacles can be avoided
through the proper design of an MRI protocol . However, good communication between the radiologist,
[73]
MRI team and the person who wants to use the MRI dataset for 3D reconstruction is necessary.

Liver volumetry is one of the essential computer-aided surgical tools. IT can be performed by surgeons
alone. Van der Vorst presented accurate volumetric measurements using Osirix® (Pixmeo, Geneva,
[74]
Switzerland). Dello et al. compared volumetry done in Osirix® and ImageJ by non-radiologists on
[75]
personal computers and compared the results with radiological software iNtuition® (National Institute of
Health, USA). The results were accurate. In another paper, Dello et al. presented how to perform liver
[76]
volumetry in ImageJ. Lodewick et al. compared volumetry in OsiriX® and iNtuition® and also found no
[77]
differences in measured volumes, but they also stressed that auto-segmentation with iNtuition® is three
times faster than manual volumetry in OsiriX®. In the reviewed articles, volumetry was performed by
radiologists or software engineers.