Search

In this paper, we aimed to evaluate clinical and imagistic features, and also to provide a diagnostic algorithm for patients presenting with gastrointestinal involvement from hepatocellular carcinoma (HCC). We conducted a systematic search on the PubMed, Scopus and Web of Science databases to identify and collect papers oncases of HCC with gastrointestinal involvement. This search was last updated on 29 April 2022. One hundred and twenty-three articles were included, corresponding to 197 patients. The majority of the patients were male (87.30%), with a mean age of 61.21 years old. The analysis showed large HCCs located mainly in the right hepatic lobe, and highly elevated alfa-fetoprotein (mean = 15,366.18 ng/mL). The most frequent etiological factor was hepatitis B virus (38.57%). Portal vein thrombosis was present in 27.91% of cases. HCC was previously treated in most cases by transarterial chemoembolization (32.99%) and surgical resection (28.93%). Gastrointestinal lesions, developed mainly through direct invasion and hematogenous routes, were predominantly detected in the stomach and duodenum in equal measure—27.91%. Gastrointestinal bleeding was the most common presentation (49.74%). The main diagnostic tools were esophagogastroduodenoscopy (EGD) and computed tomography. The mean survival time was 7.30 months. Gastrointestinal involvement in HCC should be included in the differential diagnosis of patients with underlying HCC and gastrointestinal manifestations or pathological findings in EGD. [ABSTRACT FROM AUTHOR]

We performed a meta-analysis of published data to investigate the diagnostic value of artificial intelligence for pancreatic cancer. Systematic research was conducted in the following databases: PubMed, Embase, and Web of Science to identify relevant studies up to October 2021. We extracted or calculated the number of true positives, false positives true negatives, and false negatives from the selected publications. In total, 10 studies, featuring 1871 patients, met our inclusion criteria. The risk of bias in the included studies was assessed using the QUADAS-2 tool. R and RevMan 5.4.1 software were used for calculations and statistical analysis. The studies included in the meta-analysis did not show an overall heterogeneity (I2 = 0%), and no significant differences were found from the subgroup analysis. The pooled diagnostic sensitivity and specificity were 0.92 (95% CI, 0.89–0.95) and 0.9 (95% CI, 0.83–0.94), respectively. The area under the summary receiver operating characteristics curve was 0.95, and the diagnostic odds ratio was 128.9 (95% CI, 71.2–233.8), indicating very good diagnostic accuracy for the detection of pancreatic cancer. Based on these promising preliminary results and further testing on a larger dataset, artificial intelligence-assisted endoscopic ultrasound could become an important tool for the computer-aided diagnosis of pancreatic cancer. [ABSTRACT FROM AUTHOR]