Page 54 - Read Online

Page 12 of 16 Kościuszko et al. Hepatoma Res 2021;7:51 https://dx.doi.org/10.20517/2394-5079.2021.17

[57]
The currently used systems rely on software assistants that use specific software packages such as
iNtuition®, Fraunhofer MeVis or Visible Patient Planning™ (Ircad, France; not mentioned in the reviewed
articles). Those software packages or services include medical image analysis with deep learning and
machine learning techniques: automated organ segmentation in CT and MR, automatic liver lesion
detection and segmentation, automatic vessel segmentation in contrast - enhanced CT, etc. Several open-
[78]
source neural networks exist that can perform automatic segmentation tasks, such as U-net . A complete
automated liver segmentation problem is not fully solved yet, although the situation has changed since
[79]
[78]
2009 thanks to deep neural networks [80,81] . The paper by Ibtehaz et al. is different from the other
reviewed articles, as it describes U-net’s use in automatic volumetry of liver graft.
Several well-designed review have been published recently on augmented and virtual reality in oncologic
[14]
[82]
[83]
liver surgery , 3D printing in liver surgery , 3D printing in general medical setting , precision
hepatectomy and liver segmentation techniques . However, we feel that this landscape is incomplete due
[84]
[85]
to the lack of a comprehensive review of recent advances in radiological assessment of the liver focused on
preoperative liver surgery planning in the paediatric population. We believe that this focused approach on
the topic is essential because it enables a thorough analysis of the achievements in the field. In other words,
it allows us to define “where we are”.
The reviewed studies have mostly evaluative characteristics or are simply case reports. Only Zhang et al. ,
[57]
Su et al. and Wang et al. compared the results of patients who had computer-aided surgical planning
[60]
[54]
performed with patients without additional planning. However, there is a high risk of bias in those studies,
mainly confirmation bias. Due to the low numbers of patients in the groups, the studies may be
underpowered and not show significant results. More prospective, comparative studies with well-defined
outcome measures should be performed. It is also not known whether the computer-aided preoperative
planning should be used in all cases or only in complex cases.

It is also not known how surgeons perceive different types of 3D preoperative planning. What is the
difference between 3D-printed object, 3D visualisation seen on a computer screen and using 3D glasses or
mixed reality glasses? Even subjective opinions of surgeons would be helpful to choose the best tool for
further evaluation. The use of these advanced methods during surgery may be problematic because, for
now, they need active input from the surgeon (to hold the 3D-printed model or ask someone to hold and
move the model for her/him).

Most authors used 3D reconstructions presented on monitors and 3D-printed objects. The application of
augmented reality utilising stereo head-mounted displays enabling holographic imaging has not been
described yet in paediatric liver tumour surgery. An example of such a device is Microsoft’s HoloLens™. The
3D models can be manipulated during surgery by the operating surgeon and tracked by hand gestures and
voice commands. The use of HoloLens™ in adult liver surgery has been described , and the results are
[86]
favourable.

On the other hand, a holographic image still needs to be set appropriately (zoomed-in or out, rotated, etc.).
When overlying the operated organ, the holographic image may partly cover the organ and distract the
surgeon. In addition, the 3D-printed models and 3D visualisations presented on the screens may be a source
of distraction. It has not been yet resolved whether computer-aided surgery is beneficial only before the
operation (preoperative planning) or during the surgery, despite all enthusiasm. This may also be a question
of a long learning curve. In addition, there is a possibility that experienced surgeons will be less likely to use
novel technologies. However, this technology may also work as a booster of a learning curve in