Usefulness of Artificial Intelligence for Surgical Support in Robot-assisted Distal Pancreatectomy: A Preliminary Case Report.

BACKGROUND/AIM: The incidence of postoperative complications in minimally-invasive surgery for pancreatic disease remains a concern. The application of artificial intelligence (AI) in surgery has been reported to improve the precision of anatomical identification, potentially improving the rate of postoperative complications. In this report, the utility of organ recognition support provided by surgical AI during robot-assisted distal pancreatectomy for a patient with pancreatic neuroendocrine tumor (NET) is described.

CASE REPORT: A 33-year-old woman with hereditary breast and ovarian cancer syndrome was followed-up after undergoing left breast cancer surgery and hormone therapy for bone metastasis. Endoscopic ultrasound revealed a 14 mm hypoechoic mass in the pancreatic tail. Following fine-needle aspiration, NET or solid pseudopapillary neoplasm was suspected. Give that the bone metastasis was well-controlled and no additional distant metastases were detected, surgical resection was chosen with the dual intent of therapeutic intervention and definitive diagnosis. Robot-assisted distal pancreatectomy was performed using the Da Vinci Xi Surgical System. The patient was discharged on postoperative day 25. Histopathological testing revealed that the tumor was a pancreatic NET (G1). A retrospective analysis using the video footage from the surgery was performed with aid of the Eureka AI system. The support system could recognize pancreatic parenchyma and loose connective tissue. Although the patient had a high body mass index, fat tissue was clearly distinguished from pancreatic tissue. During the surgery, the dissection line on the lower margin of the pancreas was falsely recognized and pancreatic tissue was damaged. The correct dissection line, however, was seen in the AI analysis image.

CONCLUSION

Organ recognition assistance using surgical AI during robot-assisted distal pancreatectomy enables enhanced color-coded visualization of anatomical structures, potentially preventing intraoperative errors.