deep learning algorithm

深度学习算法
  • 文章类型: Case Reports
    我们已经证明了SYNAPSEVINCENT®(6.6版;富士胶片医疗公司,Ltd.,东京,Japan),三维图像分析系统,在胰周血管的半自动模拟中,胰管,胰腺实质,和胰腺周围器官使用深度学习算法开发的人工智能(AI)引擎。此外,我们调查了这种AI引擎对胰腺癌患者的有用性。这里,我们介绍了1例腹腔镜远端胰腺切除术的情况,并通过AI引擎使用手术模拟和导航进行了扩展的外科手术.一名80岁的妇女出现腹痛。增强的腹部计算机断层扫描(CT)显示主胰管扩张,最大直径为40mm。此外,胰头和胰体之间有一个17毫米的囊性病变,胰尾有一个14毫米的壁结节。因此,该病变术前诊断为胰尾导管内乳头状癌(IPMC),并根据第8版国际癌症控制联盟指南分类为T1N0M0IA期.本患者接受了腹腔镜远端胰腺切除术和区域淋巴结切除术。特别是,因为有必要包括胰腺颈部的囊性病变,胰腺切除术在门静脉的右边缘进行,比平常更靠近胰头。我们通常采用三维计算机图形学(3DCG)手术模拟和导航,这让我们认识到手术解剖结构,包括胰腺切除的位置.除了显示手术解剖的详细3DCG,这项技术使外科手术人员可以分享情况,据报道,这种方法提高了手术的安全性。此外,残余胰腺体积(47.6%),胰腺切除表面积(161mm2),使用3DCG成像研究切除位置的胰腺实质(12mm)的厚度。术中冰冻活检证实切缘阴性。组织学上,在胰尾观察到导管内乳头状黏液性肿瘤伴低度发育不良.没有恶性发现,包括那些与切除边缘有关的,在标本中观察到。在术后12个月的随访检查中,病人的情况并不显著。我们得出的结论是,SYNAPSEVINCENT®AI引擎是提取周围血管的有用手术支持,周围的器官,和胰腺实质,包括胰腺切除的位置,即使在延长的外科手术的情况下。
    We have demonstrated the utility of SYNAPSE VINCENT® (version 6.6; Fujifilm Medical Co., Ltd., Tokyo, Japan), a 3D image analysis system, in semi-automated simulations of the peripancreatic vessels, pancreatic ducts, pancreatic parenchyma, and peripancreatic organs using an artificial intelligence (AI) engine developed with deep learning algorithms. Furthermore, we investigated the usefulness of this AI engine for patients with pancreatic cancer. Here, we present a case of laparoscopic distal pancreatectomy with an extended surgical procedure performed using surgical simulation and navigation via an AI engine. An 80-year-old woman presented with abdominal pain. Enhanced abdominal computed tomography (CT) revealed main pancreatic duct dilatation with a maximum diameter of 40 mm. Furthermore, there was a 17 mm cystic lesion between the pancreatic head and the pancreatic body and a 14 mm mural nodule in the pancreatic tail. Thus, the lesion was preoperatively diagnosed as an intraductal papillary carcinoma (IPMC) of the pancreatic tail and classified as T1N0M0 stage IA according to the 8th edition of the Union for International Cancer Control guidelines. The present patient had laparoscopic distal pancreatectomy and regional lymphadenectomy. In particular, since it was necessary to include the cystic lesion in the pancreatic neck, pancreatic resection was performed at the right edge of the portal vein, which is closer to the head of the pancreas than usual. We routinely employed three-dimensional computer graphics (3DCG) surgical simulation and navigation, which allowed us to recognize the surgical anatomy, including the location of pancreatic resection. In addition to displaying the detailed 3DCG of the surgical anatomy, this technology allowed surgical staff to share the situation, and it has been reported that this approach improves the safety of surgery. Furthermore, the remnant pancreatic volume (47.6%), pancreatic resection surface area (161 mm2), and thickness of the pancreatic parenchyma (12 mm) at the resection location were investigated using 3DCG imaging. Intraoperative frozen biopsy confirmed that the resection margin was negative. Histologically, an intraductal papillary mucinous neoplasm with low-grade dysplasia was observed in the pancreatic tail. No malignant findings, including those related to the resection margin, were observed in the specimen. At the 12-month postoperative follow-up examination, the patient\'s condition was unremarkable. We conclude that the SYNAPSE VINCENT® AI engine is a useful surgical support for the extraction of the surrounding vessels, surrounding organs, and pancreatic parenchyma including the location of the pancreatic resection even in the case of extended surgical procedures.
    导出

    更多引用

    收藏

    翻译标题摘要

    我要上传

       PDF(Pubmed)

公众号