- [In German] Eckhoff JA et al. Anwendung von künstlicher Intelligenz in der onkologischen Chirurgie des oberen Gastrointestinaltrakts. Onkologie 2023; 29: 515–521.
Translated from the original German version.
In medical diagnostics, AI has already been used successfully for some time. For example, retrospective image analysis significantly improves sensitivity in mammography screening or melanoma staging. Artificial intelligence has also become an important tool in surgery. AI algorithms analyse pre- and postoperative information, derive statistical predictions from it and help with therapy decisions.
In the endoscopic diagnosis of early gastric cancer, AI algorithms now achieve 97% precision through image-based analysis of microvascularisation. Artificial intelligence can also support the choice of the right therapy. Through the preoperative evaluation of CT data, it has already been possible to determine the resectability of oesophageal carcinomas with 79% accuracy. The response to radiochemotherapy could be assessed with 77% accuracy. AI has also proven helpful in the assessment of postoperative complications. It can predict anastomotic insufficiency after oesophagectomy with 98% accuracy.
In the acute, time-critical situation of the operation itself, however, the use of artificial intelligence is incomparably more complex. Here, intraoperative information must be captured and processed in real time in order to provide the surgeon with relevant decision-making aids. To do this, AI must penetrate the surgical field and the individual surgical phases, recognise the spatial arrangement of anatomy and instruments, and comprehend the temporal sequences. In order to train artificial intelligence specifically for this purpose, comprehensive video and image data sets are first required. Current data protection regulations currently still stand in the way of this.
Therefore, the use of AI in the operating theatre has so far been limited mainly to standardised routine procedures such as laparoscopic cholecystectomy. In complex oncological operations of the upper gastrointestinal tract, artificial intelligence and its programmers are still learning. In combination with advancing autonomous robotics, entirely new applications for AI could emerge in the future.
One thing remains clear, however: artificial intelligence will not replace the surgeon. Rather, the objective and highly efficient analysis of the surgical field will provide the surgeon with additional information and thus help to avoid errors and intraoperative complications.
Surgery offers an optimal application area for artificial intelligence. So far, AI algorithms have mainly helped in the processing and prognostic classification of perioperative data. It remains exciting to see how they will be used in the future for operations in real time.