Development of a cross-artificial intelligence system for identifying intraoperative anatomical landmarks and surgical phases during laparoscopic cholecystectomy

BACKGROUND

Attention to anatomical landmarks in the appropriate surgical phase is important to prevent bile duct injury (BDI) during laparoscopic cholecystectomy (LC). Therefore, we created a cross-AI system that works with two different AI algorithms simultaneously, landmark detection and phase recognition. We assessed whether landmark detection was activated in the appropriate phase by phase recognition during LC and the potential contribution of the cross-AI system in preventing BDI through a clinical feasibility study (J-SUMMIT-C-02).

METHODS

A prototype was designed to display landmarks during the preparation phase and Calot's triangle dissection. A prospective clinical feasibility study using the cross-AI system was performed in 20 LC cases. The primary endpoint of this study was the appropriateness of the detection timing of landmarks, which was assessed by an external evaluation committee (EEC). The secondary endpoint was the correctness of landmark detection and the contribution of cross-AI in preventing BDI, which were assessed based on the annotation and 4-point rubric questionnaire.

RESULTS

Cross-AI-detected landmarks in 92% of the phases where the EEC considered landmarks necessary. In the questionnaire, each landmark detected by AI had high accuracy, especially the landmarks of the common bile duct and cystic duct, which were assessed at 3.78 and 3.67, respectively. In addition, the contribution to preventing BDI was relatively high at 3.65.

CONCLUSIONS

The cross-AI system provided landmark detection at appropriate situations. The surgeons who previewed the model suggested that the landmark information provided by the cross-AI system may be effective in preventing BDI. Therefore, it is suggested that our system could help prevent BDI in practice. Trial registration University Hospital Medical Information Network Research Center Clinical Trial Registration System (UMIN000045731).

An intraoperative artificial intelligence system identifying anatomical landmarks for laparoscopic cholecystectomy: a prospective clinical feasibility trial (J-SUMMIT-C-01)

BACKGROUND

We have implemented Smart Endoscopic Surgery (SES), a surgical system that uses artificial intelligence (AI) to detect the anatomical landmarks that expert surgeons base on to perform certain surgical maneuvers. No report has verified the use of AI-based support systems for surgery in clinical practice, and no evaluation method has been established. To evaluate the detection performance of SES, we have developed and established a new evaluation method by conducting a clinical feasibility trial.

METHODS

A single-center prospective clinical feasibility trial was conducted on 10 cases of LC performed at Oita University hospital. Subsequently, an external evaluation committee (EEC) evaluated the AI detection accuracy for each landmark using five-grade rubric evaluation and DICE coefficient. We defined LM-CBD as the expert surgeon's "judge" of the cystic bile duct in endoscopic images.

RESULTS

The average detection accuracy on the rubric by the EEC was 4.2 ± 0.8 for the LM-CBD. The DICE coefficient between the AI detection area of the LM-CBD and the EEC members' evaluation was similar to the mean value of the DICE coefficient between the EEC members. The DICE coefficient was high score for the case that was highly evaluated by the EEC on a five-grade scale.

CONCLUSION

This is the first feasible clinical trial of an AI system designed for intraoperative use and to evaluate the AI system using an EEC. In the future, this concept of evaluation for the AI system would contribute to the development of new AI navigation systems for surgery.

An additional port in difficult laparoscopic cholecystectomy for surgical safety

BACKGROUND

Tokyo Guidelines 2018, clinical practice guidelines for acute cholangitis and cholecystitis, recommend bailout procedures to prevent bile duct injury (BDI) during laparoscopic cholecystectomy (LC) for difficult gallbladder. We first insert an additional port (AP) for difficult gallbladder that may require bailout procedures. Because the usefulness of an AP during LC is unclear, we therefore examined the efficacy of the AP during LC in this study.

METHODS

Data were collected from 115 patients who underwent LC for acute cholecystitis in our department. The indications for AP were excessive bleeding, scarring, and poor visual field around Calot's triangle. AP was inserted into the right middle abdomen so as not to interfere with other trocars and was used by the assistant. Surgical outcomes were evaluated based on AP use during LC.

RESULTS

AP was inserted in 19 patients during LC (AP group). The indications for AP were excessive bleeding in nine patients, scarring around Calot's triangle in seven patients, and poor visual field around Calot's triangle in three patients. Open conversion was performed in two patients in the non-AP group. BDI occurred in one patient in the non-AP group. In patients with Difficulty Score 3, operation time was significantly shorter (P = .038) and Critical View of Safety (CVS) score was significantly higher in the AP group (P = .046).

CONCLUSION

AP is useful in patients with excessive bleeding to shorten operation time and increase the CVS score. AP may be one useful option for difficult gallbladder.

Artificial intelligence software available for medical devices: surgical phase recognition in laparoscopic cholecystectomy

Background

Surgical process modeling automatically identifies surgical phases, and further improvement in recognition accuracy is expected with deep learning. Surgical tool or time series information has been used to improve the recognition accuracy of a model. However, it is difficult to collect this information continuously intraoperatively. The present study aimed to develop a deep convolution neural network (CNN) model that correctly identifies the surgical phase during laparoscopic cholecystectomy (LC).

Methods

We divided LC into six surgical phases (P1–P6) and one redundant phase (P0). We prepared 115 LC videos and converted them to image frames at 3 fps. Three experienced doctors labeled the surgical phases in all image frames. Our deep CNN model was trained with 106 of the 115 annotation datasets and was evaluated with the remaining datasets. By depending on both the prediction probability and frequency for a certain period, we aimed for highly accurate surgical phase recognition in the operation room.

Results

Nine full LC videos were converted into image frames and were fed to our deep CNN model. The average accuracy, precision, and recall were 0.970, 0.855, and 0.863, respectively.

Conclusion

The deep learning CNN model in this study successfully identified both the six surgical phases and the redundant phase, P0, which may increase the versatility of the surgical process recognition model for clinical use. We believe that this model can be used in artificial intelligence for medical devices. The degree of recognition accuracy is expected to improve with developments in advanced deep learning algorithms.

Gastric Wall Thickness and Linear Staple Height in Sleeve Gastrectomy in Japanese Patients with Obesity

BACKGROUND

Laparoscopic sleeve gastrectomy (LSG) is a standard procedure due to its low complication rates and favorable outcomes. However, limited data are available regarding the optimal size of linear staplers in relation to gastric wall thickness (GWT).

METHODS

Between August 2016 and December 2020, we performed LSG in 70 patients with an average age, body weight, and body mass index of 42 years, 107 kg, and 40 kg/m², respectively. We measured the GWT at the antrum, body, and fundus using resected specimens. We used an endo-linear stapler, and the closed staple height (CSH) was 1.75 mm.

RESULTS

We found that the average GWT at the antrum was significantly thicker than the GWT at the body and fundus. There was a statistically significant relationship between body weight and the GWT at the antrum and body and obstructive sleep apnea and the GWT at the body. The average CSH/GWT ratios were 0.55, 0.62, and 0.90 at the antrum, body, and fundus, respectively. However, in 20 patients (29%), the CSH/GWT ratio at the fundus area was ≥ 1.0, and only preoperative body weight was a significant predictor for a CSH/GWT ratio of ≥ 1.0.

CONCLUSION

A light body weight may be related to a CSH/GWT ratio of ≥ 1.0 at the fundus.

Recurrence of gastric gastrointestinal stromal tumor 12 years after repeat hepatectomies for liver metastases: report of a case

No consensus has been reached on the postoperative treatment and follow-up duration for high-risk malignant gastrointestinal stromal tumor (GIST). We herein report a case of recurrent liver metastasis from gastric GIST in a patient who had been receiving adjuvant chemotherapy with imatinib mesylate for 10 years. A 78-year-old woman underwent local gastrectomy for a 20-cm gastric GIST with a mitotic index of 25/50 high-power fields (HPF) 17 years before. Partial hepatectomy for recurrent liver metastases was repeatedly performed 14 and 12 years before. After the second hepatectomy, postoperative adjuvant chemotherapy with imatinib mesylate was given for 10 years, during which no recurrence was observed. Two years after the completion of adjuvant chemotherapy, computed tomography revealed a 2-cm hepatic tumor; thus, laparoscopic partial hepatectomy was performed. Histopathological findings revealed a liver metastasis of gastric GIST with a mitotic count of 20/50 HPF and MIB-1 labeling index of 20%. Mutation analysis of the KIT gene revealed an exon 11 mutation. The patient is currently undergoing postoperative adjuvant chemotherapy with imatinib mesylate. The combination of surgery and long-term adjuvant chemotherapy for high-risk malignant GIST and liver metastases may be effective to achieve a good prognosis.

Development of Endoscopic Surgery Navigated by Artificial Intelligence

Endoscopic surgery, which was first introduced in the late 1980s, has rapidly become widespread. However, despite its popularity, the occurrence of intraoperative organ damage has not necessarily decreased. To avoid intraoperative bile duct injury in laparoscopic cholecystectomy, which is one of the most popular procedures in endoscopic surgery, we are developing a laparoscopic surgical system that uses Artificial Intelligence (AI) to identify four anatomical landmarks (cystic duct of the gallbladder, common bile duct, lower surface of hepatic S4, and Rouviere's sulcus, related to "Calot's triangle") in real time during surgery. The development process consists of 5 steps: 1) identification of anatomical landmarks, 2) collection and creation of teaching data, 3) annotation and deep learning, 4) validation of development model, and 5) actual clinical performance evaluation. At present, anatomical landmarks can be identified with high accuracy in an actual clinical performance test in laparoscopic cholecystectomy, whereas issues for practical clinical use, such as a need to recognize the scene of surgical steps and surgical difficulties related to inflammation of the gallbladder, have also been clarified. The development of an AI-navigation system for endoscopic surgery, which could identify anatomical landmarks in real time during surgery, could be expected to support surgeons' decisions, reduce surgical complications, and contribute to improving the quality of surgical treatments.

Surgical and oncological outcomes of laparoscopic versus open radical antegrade modular pancreatosplenectomy for pancreatic ductal adenocarcinoma

PURPOSE

To compare the outcomes of laparoscopic radical antegrade modular pancreatosplenectomy (L-RAMPS) with those of open RAMPS (O-RAMPS) in patients with pancreatic ductal adenocarcinoma (PDAC).

METHODS

We reviewed, retrospectively, the medical records of 50 patients who underwent RAMPS for PDAC without resection of major vessels and adjacent organs between 2007 and 2019, and analyzed the relationship between the operative method and surgical and oncological outcomes.

RESULTS

Nineteen of the 50 patients underwent L-RAMPS and 31 patients underwent O-RAMPS. L-RAMPS was associated with significantly less blood loss (P = 0.034) but a longer operative time (P = 0.001) than O-RAMPS. There were no significant differences in patient characteristics, tumor factors, or postoperative course; or in the rates of recurrence-free survival (P = 0.084) or overall survival (P = 0.402) between the L-RAMPS and O-RAMPS groups.

CONCLUSION

L-RAMPS for PDAC resulted in less blood loss but a longer operative time than O-RAMPS. Although L-RAMPS may be feasible, the operative time needs to be reduced by standardizing the procedure.

Predictors of early withdrawal from follow-up visits after laparoscopic sleeve gastrectomy in a Japanese institution

PURPOSE

Postoperative weight loss is related to postoperative adherence to follow-up after bariatric/metabolic surgery, but many patients stop attending follow-up visits early. The aim of this study was to clarify predictors of early withdrawal from follow-up after laparoscopic sleeve gastrectomy (LSG) in a Japanese institution.

METHODS

One hundred and fifty-three patients who underwent LSG were retrospectively included in this study. Multivariate analysis was performed to evaluate independent predictors of withdrawal from follow-up visits within 12 months after LSG among significant or nearly significant factors in the univariate analyses. The discrimination power of significant factors was estimated using area under the receiver operating characteristic curve (AUC).

RESULTS

Within 12 months after LSG, 25 of the 153 patients withdrew from follow-up visits. The multivariate analysis showed that age was the only significant predictor of withdrawal. The AUC for age was 0.685, and the cut-off value was < 40 years. The younger patients (< 40 years old) had a significantly higher rate of withdrawal compared with the older patients (≥ 40 years) (27.0% vs. 8.9%).

CONCLUSION

Older Japanese patients (≥ 40 years old) may be better candidates for LSG. We consider it significant to continue to emphasize the importance of follow-up visits in younger patients after LSG.

[Long‒Term Survival after Lung Resection for Lung Metastasis of Pancreatic Cancer-A Report of Two Cases]

Some reports have shown that the prognosis of recurrent lung metastases after resection of pancreatic cancer is better than that for other organs. We report on 2 cases of long‒term survival after lung resection for lung metastases from pancreatic cancer. Case 1: A 73‒year‒old man underwent distal pancreatectomy for pancreatic body cancer. Adjuvant chemotherapy of S‒1 was administered for 6 months. At 3 years after surgery, 2 small metastatic nodules were detected in the right lung, and the patient underwent thoracoscopic right upper lobectomy. At 5 years and 9 months after the initial surgery, he is alive without recurrence. Case 2: An 81‒year‒old woman underwent pylorus‒preserving pancreatoduodenectomy for pancreatic head cancer. Adjuvant chemotherapy of S‒1 was administered for 4 months. At 1 year and 9 months after surgery, a metastatic nodule was detected in the left upper lung, and the patient underwent thoracoscopic left upper lobectomy. At 4 years and 6 months after initial surgery, radiation therapy was performed for localized bone metastasis. At 5 years and 1 month after the initial surgery, she is alive without other recurrences. In conclusion, recurrent lung metastases of pancreatic cancer with a small number of metastases may result in long‒term survival following resection.

Clinical Significance of Splenic Vessels and Anatomical Features in Laparoscopic Splenectomy

Introduction: Laparoscopic splenectomy (LS) has become the standard treatment for benign hematological disorders and hypersplenism. However, serious complications such as pancreatic fistula and portal venous thrombosis (PVT) sometimes occur. We investigated the clinical significance of splenic vessels and anatomical features in LS. Methods: Patient data were collected from 32 patients who underwent LS. The indications for LS were hypersplenism due to liver cirrhosis, idiopathic thrombocytopenic purpura, hereditary spherocytosis, and others. Close contact of pancreatic tail with splenic hilum, spleen volume, and diameters of splenic vessels were evaluated on computed tomography images. Results: Close contact of pancreatic tail with splenic hilum was recognized in 15 of the patients. The close contact was significantly associated with operation time (P = .038), spleen volume (P = .021), and spleen volume/body surface area (BSA) ratio (P = .001). In multivariate analysis, spleen volume/BSA ratio was an independent factor for close contact (P = .022). PVT occurred in 3 cirrhosis patients, and the diameter of the splenic vein (SV) was significantly associated with PVT as a result of multivariate analysis (P = .027). Conclusion: Close contact of the pancreatic tail with the splenic hilum may cause a longer operation time at LS and be associated with spleen volume/BSA ratio. A larger SV diameter in cirrhosis patients may be related to PVT after LS.

A case of ruptured mucinous cystadenoma of the pancreas with recurrence-free survival for 8 years

BACKGROUND

Pancreatic mucinous cystic neoplasm (MCM) presenting with rupture is extremely rare, and very few studies have followed up patients over the long term after ruptured mucinous cystadenoma (MCA). We report a case of ruptured MCA of the pancreas with recurrence-free survival for 8 years.

CASE PRESENTATION

A 28-year-old Japanese woman was admitted to the emergency department of a local hospital after experiencing acute abdominal pain. Abdominal computed tomography revealed massive ascites and the presence of a cystic tumor measuring 60 mm in diameter in the pancreatic tail. Conservative therapy with antibiotics and abdominal drainage were performed to treat peritonitis that occurred secondary to the ruptured pancreatic cystic tumor, after which the patient's symptoms improved. The patient was referred to our department for further examination and treatment. We diagnosed a ruptured MCN and performed laparoscopic spleen-preserving distal pancreatectomy. Histopathological findings revealed ovarian-type stroma, which tested positive for estrogen and progesterone receptors by immunohistochemistry. The histopathological diagnosis was MCA. The postoperative course was uneventful, and the patient remains alive without any evidence of recurrence at 8 years postoperatively.

CONCLUSION

A good prognosis is possible even in cases of ruptured MCA. Because of the risk of peritoneal dissemination after ruptured MCA, long-term follow-up is important.

Fucosyltransferase 8 plays a crucial role in the invasion and metastasis of pancreatic ductal adenocarcinoma

PURPOSE

Pancreatic ductal adenocarcinoma (PDAC) is the most common type of pancreatic cancer. It is an aggressive malignancy associated with poor prognosis because of recurrence, metastasis, and treatment resistance. Aberrant glycosylation of cancer cells triggers their migration and invasion and is considered one of the most important prognostic cancer biomarkers. The current study aimed to identify glycan alterations and their relationship with the malignant potential of PDAC.

METHODS

Using a lectin microarray, we evaluated glycan expression in 62 PDAC samples. Expression of fucosyltransferase 8 (FUT8), the only enzyme catalyzing core fucosylation, was investigated by immunohistochemistry. The role of FUT8 in PDAC invasion and metastasis was confirmed using an in vitro assay and a xenograft peritoneal metastasis mouse model.

RESULTS

The microarray data demonstrated that core fucose-binding lectins were significantly higher in carcinoma than in normal pancreatic duct tissues. Similarly, FUT8 protein expression was significantly higher in carcinoma than in normal pancreatic duct tissues. High FUT8 protein expression was significantly associated with lymph-node metastases and relapse-free survival. FUT8 knockdown significantly reduced the invasion in PDAC cell lines and impaired peritoneal metastasis in the xenograft model.

CONCLUSIONS

The findings of this study provide evidence that FUT8 plays a pivotal role in PDAC invasion and metastasis and might be a therapeutic target for this disease.

Pharmacokinetic and toxicodynamic evaluation of 5-fluorouracil administration after major hepatectomy in a rat model

BACKGROUND

Chemotherapy after hepatectomy for colorectal liver metastasis has not been established, due to the toxic side effects, which are likely related to impaired drug clearance during liver regeneration. We investigated the pharmacokinetic and toxicodynamic evaluation of 5-fluorouracil (5-FU) during liver regeneration after major hepatectomy in a rat model.

METHODS

Thirty-six male Wistar rats were divided into control (C), control with chemotherapy (CC), hepatectomy (H), and hepatectomy with chemotherapy (HC) groups. The CC and HC groups were administered 5-FU for 4 days. Plasma 5-FU, liver weight, and liver dihydropyrimidine dehydrogenase (DPD) were measured. The ileal villous height was measured to determine adverse effects.

RESULTS

The area under the curve and maximum plasma concentration of 5-FU increased by up to 51% and 32%, respectively, in the HC group compared to the CC group. The liver regeneration rate was significantly lower in the HC group than in the H group (67.3 ± 7.4 vs 33.0 ± 5.7%, p < 0.001). The HC group had a significantly lower liver DPD than the CC group (4.4 ± 1.1 mg vs 6.9 ± 1.1 mg, p < 0.01). The HC group had a significantly lower ileal villous height than the CC group (253 ± 40 μm vs. 318 ± 36 μm, p < 0.05).

CONCLUSIONS

Reduction of the total liver DPD following major hepatectomy caused increased plasma 5-FU levels and 5-FU-associated toxicity.

[IPMC of the pancreas with enlarged para-aortic lymph nodes with a high accumulation of FDG-PET:a case report]

A 74-year-old man was admitted for evaluation of a pancreatic tumor. Abdominal CT revealed a multilocular cystic tumor with thickened septal walls and an enhanced mural nodule in the head of the pancreas, as well as enlarged para-aortic lymph nodes. FDG-PET revealed FDG accumulation in the nodule and in the para-aortic nodes. The tumor was diagnosed as intraductal papillary mucinous carcinoma (IPMC) with para-aortic lymph node metastases. After 2 courses of chemotherapy with gemcitabine and nab-paclitaxel, the tumor decreased slightly in size, but the lymph nodes did not change. Surgery was then performed. Intraoperative pathology examination confirmed that the para-aortic lymph nodes had only inflammatory swelling. Accordingly, pylorus-preserving pancreaticoduodenectomy was performed. Histopathology examination revealed atypical cells without invasion, and IPMC (TisN0M0;stage 0) was diagnosed. It is thus important for selection of the appropriate treatment approach to determine if enlarged para-aortic lymph nodes are benign or malignant.

Improvement of non-alcoholic fatty liver disease after laparoscopic sleeve gastrectomy in Japanese obese patients

PURPOSE

The purpose of this study was to evaluate changes in non-alcoholic fatty liver disease (NAFLD) after laparoscopic sleeve gastrectomy (LSG) using computed tomography (CT) images.

METHODS

We analyzed data from 57 patients who underwent LSG and had CTs performed before and after surgery. The patients included 34 women and 23 men (with an average age of 43 years); their mean preoperative weight and body mass index were 120 kg and 46 kg/m2, respectively. Obesity-related health disorders included type 2 diabetes mellitus (T2DM) in 33 patients, hypertension in 33 and dyslipidemia in 32. We diagnosed NAFLD in cases with liver to spleen ratios (L/S ratio) <0.9 on non-contrast CT images. We evaluated changes in body weights, BMIs, comorbidities, metabolic parameters, L/S ratios, and liver volumes after surgery.

RESULTS

The mean interval between CT scans before and after surgery was 26 months. The total weight loss and % excess weight loss were 35 kg and 72%, respectively. The remission rates for T2DM, hypertension, and dyslipidemia were 85%, 76% and 84%, respectively. After LSG, the L/S ratio increased in all the patients, while all except for one had L/S ratio >0.9. We diagnosed 33 out of 57 patients (58%) as having NAFLD before the operation. After the operation, the L/S ratios and liver volumes were not statistically different between the patients with previous NAFLD and those without it.

CONCLUSION

Laparoscopic sleeve gastrectomy is an effective treatment for obesity-related health disorders including NAFLD in Japanese obese patients.

Pylorus-preserving pancreatoduodenectomy preserving the right gastroepiploic vessels following proximal gastrectomy: report of two cases

BACKGROUND

Blood flow of the remnant stomach is supplied via the right gastric and right gastroepiploic vessels after proximal gastrectomy (PG). Whether the remnant stomach can be safely preserved in patients who undergo pylorus-preserving pancreatoduodenectomy (PPPD) after PG remains unclear. We herein report two cases in which the remnant stomach was safely preserved by performing PPPD.

CASE PRESENTATION

The first patient, a 76-year-old man, was diagnosed with cancer of the common bile duct and underwent PPPD 2 years after PG for gastric cancer. The remnant stomach and right gastroepiploic vessels were safely preserved. The second patient, a 56-year-old man with a history of PG for gastric cancer 20 years previously, was diagnosed with cancer of the common bile duct and underwent PPPD. We could safely preserve the remnant stomach and right gastroepiploic vessels.

CONCLUSION

The remnant stomach could be preserved in performing PPPD following PG by preserving the right gastroepiploic vessels.