人工智能技术在胃肠道肿瘤中的应用及展望
Application and Prospect of Artificial Intelligence Technology in Gastrointestinal Tumors
DOI: 10.12677/ACM.2022.12111512, PDF, HTML, XML, 下载: 254  浏览: 451 
作者: 郭晨阳:延安大学,第一临床医学院,陕西 延安;白铁成*:延安大学附属医院,胃肠疝外科,陕西 延安
关键词: 人工智能胃癌结直肠癌综述文献Artificial Intelligence Gastric Cancer Colorectal Cancer Review Literature
摘要: 人工智能是一个广泛的跨学科领域,其根源在于逻辑,统计学,认知心理学,决策理论,神经科学,语言学,控制论和计算机工程,致力于构建能够执行通常需要人类水平智能的任务的智能机器。人工智能作为计算机科学的一个分支,其目的是模仿思维过程,学习能力和知识管理,在实验和临床医学中找到了越来越多的应用。人工智能在医疗诊断、风险预测和治疗技术支持领域的可能性正在迅速增长,AI的发展已经渗透到医学的各个领域,取得了巨大的成功。人工智能技术在诊断和治疗几种类型的癌症,特别是胃癌与结直肠癌方面的广泛使用引起了人们的广泛关注。
Abstract: Artificial intelligence is a broad interdisciplinary field with roots in logic, statistics, cognitive psy-chology, decision theory, neuroscience, linguistics, cybernetics, and computer engineering, dedi-cated to building intelligent machines capable of performing tasks that normally require hu-man-level intelligence. As a branch of computer science, artificial intelligence aims to mimic thought processes, learning abilities and knowledge management, and has found increasing appli-cations in experimental and clinical medicine. The possibilities of artificial intelligence in the fields of medical diagnosis, risk prediction and therapeutic technology support are growing rapidly. The development of AI has penetrated all areas of medicine with great success. The widespread use of artificial intelligence technology in the diagnosis and treatment of several types of cancer, especially gastric and colorectal cancer, has attracted extensive attention.
文章引用:郭晨阳, 白铁成. 人工智能技术在胃肠道肿瘤中的应用及展望[J]. 临床医学进展, 2022, 12(11): 10497-10502. https://doi.org/10.12677/ACM.2022.12111512

1. 引言

胃肠道肿瘤,即在胃及肠道发生的肿瘤性病变,是影响人们健康的恶性疾病之一。常见的胃肠道肿瘤类型包括胃癌、直肠癌、结肠癌等。胃癌约占全球癌症的6%,是第五大最常诊断的恶性肿瘤,也是癌症相关死亡的第三大原因,仅次于肺癌和结直肠癌。根据最新的GLOBOCAN癌症估计,胃癌在2018年导致超过1,000,000例新发癌症病例和783,000例死亡 [1]。在全球范围内,结直肠癌占女性所有癌症病例的9.4%,是女性诊断出的第二大常见癌症,男性第三大癌症,占所有癌症的10.6%。 [2] 总而言之,每年约有200万人被诊断患有结直肠癌 [2]。所以对胃肠道肿瘤开展相关研究具有非常重要的价值。

人工智能(Artificial Intelligence, AI)一词是由约翰·麦卡锡(John McCarthy)在1956年举行的一次会议上创造的,在那里它被广泛称为“思考机器”。简单来说,人工智能可以定义为机器从足够有代表性的例子中学习和识别模式和关系的能力,并有效地利用这些信息对看不见的数据做出决策。人工智能 (Artificial Intelligence, AI)正在迅速重塑癌症研究。高维数据集的可用性,加上高性能计算的进步以及创新的深度学习架构,导致了人工智能在肿瘤学研究各个方面的爆炸式增长 [3]。胃镜、胶囊内镜、肠镜、超声等设备的人工智能已逐渐应用于临床 [4] [5]。人工智能的快速发展要求胃肠外科医师了解人工智能的优势和缺陷。以下就胃肠道肿瘤中人工智能的应用进行叙述。

2. 人工智能在胃肠道肿瘤诊断中的应用研究

2.1. 在胃癌中的应用研究

人工智能技术在胃癌诊断方面的应用研究主要集中在胃癌早期诊断、判断肿瘤浸润深度、识别幽门螺杆菌感染三个方面。

早期发现胃癌和癌前病变是提高生存率的关键策略。内窥镜检查已广泛用于胃癌的筛查。然而,早期胃癌(EGC)的图像诊断是困难和主观的,既往研究报道,胃癌检测的假阴性率在4.6%~25.8%之间因为它容易受到认知和技术因素的影响。提高诊断准确性的一种显着方法是使用IEE,例如蓝色激光成像(Blue Laser Imaging, BLI)和窄带成像(Narrow Band Imaging, NBI),它们比传统的白光成像更有效。人工智能辅助评估可以提供一种更好的客观方法,以提高诊断准确性并避免不必要的活检 [6] [7]。

预测胃癌的浸润深度(T分期)非常重要,因为它是决定EGC治疗方法和预后的重要因素。不涉及淋巴血管浸润且浸润深度不超过500 μm粘膜下层的早期肿瘤可以通过内窥镜切除术单独治疗 [8]。内窥镜检查或EUS上观察到的肿瘤的总体发现用于确定EGC的浸润深度。一些研究报道,常规内镜检查在预测EGC侵袭深度方面与EUS相当 [9] [10]。据报道,使用常规内镜检查的浸润深度的总体准确性在69%至79%之间 [9] [11]。在分化EGC的深度预测评分研究中,肿瘤大小超过30 mm,明显的发红,不均匀的表面和边缘升高与更深的粘膜下癌有关 [12]。然而,仅通过内窥镜检查很难确定胃癌深度,当内窥镜切除可能是一种有效的治疗方法时,一些患者可能会接受手术。为了克服这些问题,已经研究了利用AI来确定入侵深度。久保田等人回顾性地收集了来自344名接受手术或内窥镜切除的患者的902张常规内窥镜胃癌图像,以使用反向传播算法进行训练和验证,以确定浸润深度。检测侵袭深度的总体准确率为64.7%,其中T1期为77.2% (T1a为68.9%,T1b为63.6%),T2期为49.1%,T3期为51.0%,T4期为55.3%。这种计算机辅助系统提出了一种使用AI通过内窥镜确定癌症浸润深度的新方法 [13]。Zhu等人.使用来自胃癌患者的790张图像进行训练,另外203张图像来验证ResNet50。据报道,总体准确率为89.2%,明显高于经验丰富的内镜医生77.5%的总体准确率。AI的AUC为0.94 (95%CI: 0.90~0.97),敏感性、特异性、PPV和NPV分别为76.5%、95.6%、89.7%和89.0% [14]。

上消化道内镜检查对幽门螺杆菌胃炎的诊断特别重要,因为幽门螺杆菌感染与胃癌发生密切相关。然而,研究表明,传统的图像增强内窥镜检查(IEE)联合白光成像(WLI)并不能诊断胃粘膜的各种炎症 [15]。研究已经证明了人工智能在预测幽门螺杆菌感染状态以诊断胃炎方面的能力。人工智能是通过独立电气、蓝色激光成像(Blue Laser Imaging, BLI)和生命周期集成电路(Life cycle integrated circuit, LCI)高效创建的。这些研究评估了白光成像(WLI)和内窥镜检查(IEE)对幽门螺杆菌胃炎的诊断准确性,发现使用具有放大功能的WLI对幽门螺杆菌感染的诊断准确率为83.8%。一项试点研究表明,基于BLI和LCI的人工智能显示出诊断幽门螺杆菌的出色能力。BLI高光和LCI的灵敏度分别比使用WLI高96.7%和10% [16]。

2.2. 在结直肠癌中的应用研究

在结直肠癌中人工智能主要用于结直肠癌筛查、结直肠息肉检测、结直肠癌术前淋巴结分期三个方面。

大约60%~70%的具有临床症状的结直肠癌(CRC)患者在疾病的晚期被诊断出来 [17]。然而,早期检测可以改善患者的临床结局,避免治疗延误,能够使CRC发病率和死亡率降低 [18]。结直肠癌是一种高度可预防的疾病,常规筛查似乎是降低这种恶性肿瘤发病率的重要步骤 [19]。从正常粘膜到癌前生长,再到恶性病变的改变需要近15至20年的时间。息肉-癌序列进展缓慢,结直肠息肉最终可能需要10年或更长时间才能转化为恶性结构 [20]。已经开发出有效的筛查方法来识别异常组织,这些组织可能提示癌前病变或早期肿瘤 [21] [22] [23]。可用于CRC筛查的方法包括侵入性(结肠镜检查和柔性乙状结肠镜检查)和微创(胶囊内窥镜检查)技术,成像检查(计算机断层扫描结肠造影),血液和粪便检查,例如愈创木脂粪便潜血试验(FOBT),粪便免疫化学试验(FIT)和多靶点粪便DNA (MT-sDNA)试验 [24] [25]。机器学习算法可用作非侵入性和具有成本效益的方法,使用个人健康数据在广大群众中筛查CRC患病风险 [26]。

目前,结直肠息肉最重要的诊断方法是使用结肠镜。据估计,在50岁以上的筛查人群中,癌前息肉的患病率将超过50% [27]。腺瘤是最常见的癌前息肉。腺瘤检出率(ADR)是结肠镜医生检测腺瘤能力的指标。然而,结肠镜医生的ADR从7%到53%不等 [28]。许多研究表明,在筛查结肠镜检查中具有较高ADR的内窥镜医生可以更有效地保护患者免受随后的结肠癌风险 [28] [29]。Corley等人评估了136名结肠镜医生进行的314,872结肠镜检查。结果显示,若ADR每上升1.0%,那么CRC的风险就会下降3.0%。然而,利用结肠镜检查漏诊腺癌的发生率仍然很高,估计在6%至27%之间 [28] [30]。因此,需要新技术来增加结肠镜检查期间的ADR。近年来,越来越多的学者研究了AI在结肠息肉诊疗中的运用 [31]。

通常,放射科医师对CRC进行临床分期,评估结直肠癌患者的对比增强计算机断层扫描(CT)图像,此外,还评估直肠癌患者的磁共振成像(MRI)。CT和MRI的分期准确性受多种因素影响,例如设备性能,标准化成像方案,报告的放射科医生的经验以及患者特定的因素。总体而言,已发表的系列报道显示,CT诊断淋巴结转移的准确率为70%,使用MRI的准确率为69% [32] [33]。当前的诊断模式准确性有待提高,但能够通过使用人工智能(AI)模型来克服。支持AI的放射组学涉及使用高级计算算法从医学图像中提取大量研究者定义的特征 [34]。虽然放射组学模型已被用于预测CRC的淋巴结转移并取得了部分成功,但Ding等人和Wang等人先前的研究表明,深度学习算法有可能识别可能逃避常规放射学和统计方法的更微妙的模式 [35] [36] [37]。深度学习是一种复杂的技术,包括使用卷积神经网络(CNN)一种来自学基于图像的有用表示的算法,从而绕过提取手动设计特征的步骤 [38]。近年来,放射组学列线图和深度学习模型已经开始对放射诊断做出有意义的贡献 [39]。

3. 人工智能应用面临的挑战和未来发展方向

但是,AI技术的应用存在一些限制。首先,用于训练和验证的高质量数据集很少。其次,人工智能算法大多是一个黑匣子,它们缺乏对内部运作的普遍理解。此外,在医学中实施人工智能引发了许多伦理困境,这些困境跨越了医学教育,研究和实践领域。这些困境会影响患者护理,医生的角色以及参与医学领域的人的角色——政府,监管机构,保险公司,付款人和其他提供者。人工智能的作用仍在不断发展,并具有独立于人类发挥作用的独特潜力。人工智能在医学中的作用应该是透明的,以便其益处超过其潜在危害。人工智能还存在一些道德和安全问题,例如在获得患者同意后使用人工智能并确定谁应对误诊或错误治疗负责。此外,几乎任何新技术都会导致一定程度的依赖。医生对人工智能的依赖可能会误导他们,使他们容易受到伤害并受到诉讼。因此,使用人工智能可能会导致决策效率下降。过度依赖人工智能,也被称为“自动化偏见”,是一种短期症状。长期症状是技能丧失。医生的技能被技术完全或部分取代,导致可能失去自信或能力 [40]。

尽管存在这些挑战,但在精准医疗时代,人工智能在胃肠道肿瘤管理中的可预见未来是有希望的。它可以作为内镜诊断的工具进行探索,以便在早期发现疾病并预测侵袭深度。它还可用于对癌症类型进行分类,并通过数字病理学识别对药物治疗的反应。人工智能辅助系统将参与手术治疗的整个过程。然而,人工智能不会完全取代医生——人类和机器和谐地一起工作是产生最佳性能的理想状态。

NOTES

*通讯作者。

参考文献

[1] Gullo, I., Grillo, F., Mastracci, L., et al. (2020) Precancerous Lesions of the Stomach, Gastric Cancer and Hereditary Gastric Cancer Syndromes. Pathologica, 112, 166-185.
https://doi.org/10.32074/1591-951X-166
[2] Chetroiu, D., Pop, C.S., Filip, P.V., et al. (2021) How and Why do We Screen for Colorectal Cancer? Journal of Medicine and Life, 14, 462-467.
https://doi.org/10.25122/jml-2021-0192
[3] Bhinder, B., Gilvary, C., Madhukar, N.S., et al. (2021) Ar-tificial Intelligence in Cancer Research and Precision Medicine. Cancer Discovery, 11, 900-915.
https://doi.org/10.1158/2159-8290.CD-21-0090
[4] Alagappan, M., Brown, J.R.G., Mori, Y., et al. (2018) Artifi-cial Intelligence in Gas-Trointestinal Endoscopy: The Future Is Almost Here . World Journal of Gastrointestinal Endos-copy, 10, 239-249.
https://doi.org/10.4253/wjge.v10.i10.239
[5] Rees, C.J. and Koo, S. (2019) Artificial Intelligence-Upping the Game in Gastrointestinal Endoscopy? Nature Reviews Gastroenterology & Hepatology, 16, 584-585.
https://doi.org/10.1038/s41575-019-0178-y
[6] Jin, P., Ji, X., Kang, W., et al. (2020) Artificial Intelligence in Gastric Cancer: A Systematic Review. Journal of Cancer Research and Clinical Oncology, 146, 2339-2350.
https://doi.org/10.1007/s00432-020-03304-9
[7] Kim, J.H., Nam, S.J. and Park, S.C. (2021) Usefulness of Arti-ficial Intelligence in Gastric Neoplasms. World Journal of Gastroenterology, 27, 3543-3555.
https://doi.org/10.3748/wjg.v27.i24.3543
[8] Japanese Gastric Cancer Association (2011) Japanese Gastric Can-cer Treatment Guidelines 2010 (Ver. 3). Gastric Cancer, 14, 113-123.
https://doi.org/10.1007/s10120-011-0042-4
[9] Choi, J., Kim, S.G., Im, J.P., et al. (2010) Comparison of Endo-scopic Ultrasonography and Conventional Endoscopy for Prediction of Depth of Tumor Invasion in Early Gastric Cancer. Endoscopy, 42, 705-713.
https://doi.org/10.1055/s-0030-1255617
[10] Pei, Q., Wang, L., Pan, J., et al. (2015) Endoscopic Ultrasonography for Staging Depth of Invasion in Early Gastric Cancer: A Meta-Analysis. Journal of Gastroenterology and Hepatology, 30, 1566-1573.
https://doi.org/10.1111/jgh.13014
[11] Sano, T., Okuyama, Y., Kobori, O., Shimizu, T., et al. (1990) Early Gastric Cancer. Digestive Diseases and Sciences, 35, 1340-1344.
https://doi.org/10.1007/BF01536738
[12] Abe, S., Oda, I., Shimazu, T., et al. (2011) Depth-Predicting Score for Differentiated Early Gastric Cancer. Gastric Cancer, 14, 35-40.
https://doi.org/10.1007/s10120-011-0002-z
[13] Kubota, K., Kuroda, J., Yoshida, M., et al. (2012) Medical Image Analysis: Computer-Aided Diagnosis of Gastric Cancer Invasion on Endoscopic Images. Surgical Endoscopy, 26, 1485-1489.
https://doi.org/10.1007/s00464-011-2036-z
[14] Zhu, Y., Wang, Q.C., Xu, M.D., et al. (2019) Application of Convolutional Neural Network in the Diagnosis of the Invasion Depth of Gastric Cancer Based on Conventional Endos-copy. Gastrointestinal Endoscopy, 89, 806-815.
https://doi.org/10.1016/j.gie.2018.11.011
[15] Bessède, E., Arantes, V., Mégraud, F., et al. (2017) Diagnosis of Helicobacter pylori Infection. Helicobacter, 22, e12404.
https://doi.org/10.1111/hel.12404
[16] Nakashima, H., Kawahira, H., Kawachi, H., et al. (2018) Artificial Intelligence Diagnosis of Helicobacter pylori Infection Using Blue Laser Imaging-Bright and Linked Color Imaging: A Single-Center Prospective Study. Annals of Gastroenterology, 31, 462-468.
https://doi.org/10.20524/aog.2018.0269
[17] Mandel, J.S., Bond, J.H., Church, T.R., et al. (1993) Re-ducing Mortality from Colorectal Cancer by Screening for Fecal Occult Blood. Minnesota Colon Cancer Control Study. New England Journal of Medicine, 328, 1365-1371.
https://doi.org/10.1056/NEJM199305133281901
[18] Maida, M., Macaluso, F.S., Ianiro, G., et al. (2017) Screen-ing of Colorectal Cancer: Present and Future. Expert Review of Anticancer Therapy, 17, 1131-1146.
https://doi.org/10.1080/14737140.2017.1392243
[19] Lauby-Secretan, B. (2018) Eine internationale Sichtweise auf das Darmkrebs-Screening. Tumor Diagnostik & Therapie, 39, 221-222.
https://doi.org/10.1055/a-0583-8941
[20] Morson, B. (1974) The Polyp-Cancer Sequence in the Large Bowel. Proceedings of the Royal Society of Medicine, 67, 451-457.
https://doi.org/10.1177/00359157740676P115
[21] Hewitson, P., Glasziou, P.P., Irwig, L., et al. (2007) Screening for Colorectal Cancer Using the Faecal Occult Blood Test, Hemoccult. Cochrane Database of Systematic Reviews, No. 1, CD001216.
https://doi.org/10.1002/14651858.CD001216.pub2
[22] Doubeni, C.A., Corley, D.A., Quinn, V.P., et al. (2018) Effectiveness of Screening Colonoscopy in Reducing the Risk of Death from Right and Left Colon Cancer: A Large Community-Based Study. Gut, 67, 291-298.
https://doi.org/10.1136/gutjnl-2016-312712
[23] Brenner, H., Stock, C. and Hoffmeister, M. (2014) Effect of Screening Sigmoidoscopy and Screening Colonoscopy on Colorectal Cancer Incidence and Mortality: Systematic Review and Meta-Analysis of Randomised Controlled Trials and Observational Studies. BMJ, 348, Article No. 2467.
https://doi.org/10.1136/bmj.g2467
[24] Smith, R.A. andrews, K.S., Brooks, D., et al. (2019) Cancer Screening in the United States, 2019: A Review of Current American Cancer Society Guidelines and Current Issues in Cancer Screening. CA: A Cancer Journal for Clinicians, 69, 184-210.
https://doi.org/10.3322/caac.21557
[25] Issa, I.A. and Noureddine, M. (2017) Colorectal Cancer Screening: An Updated Review of the Available Options. World Journal of Gastroenterology, 23, 5086-5096.
https://doi.org/10.3748/wjg.v23.i28.5086
[26] Nartowt, B.J., Hart, G.R., Muhammad, W., et al. (2020) Robust Machine Learning for Colorectal Cancer Risk Prediction and Stratification. Fron-tiers in Big Data, 3, Article No. 6.
https://doi.org/10.3389/fdata.2020.00006
[27] Leufkens, A.M., Van Oijen, M.G.H., Vleggaar, F.P., et al. (2012) Factors Influencing the Miss Rate of Polyps in a Back-to-Back Colonoscopy Study. Endoscopy, 44, 470-475.
https://doi.org/10.1055/s-0031-1291666
[28] Corley, D.A., Jensen, C.D., Marks, A.R., et al. (2014) Adenoma Detection Rate and Risk of Colorectal Cancer and Death. New England Journal of Medicine, 370, 1298-1306.
https://doi.org/10.1056/NEJMoa1309086
[29] Coe, S.G. and Wallace, M.B. (2013) Assessment of Adenoma Detection Rate Benchmarks in Women versus Men. Gastrointestinal Endoscopy, 77, 631-635.
https://doi.org/10.1016/j.gie.2012.12.001
[30] Ahn, S.B., Han, D.S., Bae, J.H., et al. (2012) The Miss Rate for Colorectal Adenoma Determined by Quality-Adjusted, Back-to-Back Colonoscopies. Gut and Liver, 6, Article No. 64.
https://doi.org/10.5009/gnl.2012.6.1.64
[31] Wang, K.W. and Dong, M. (2020) Potential Applications of Artificial Intelligence in Colorectal Polyps and Cancer: Recent Advances and Prospects. World Journal of Gastroenterology, 26, 5090-5100.
https://doi.org/10.3748/wjg.v26.i34.5090
[32] Iannicelli, E., Di Renzo, S., Ferri, M., et al. (2014) Accuracy of High-Resolution MRI with Lumen Distention in Rectal Cancer Staging and Circumferential Margin Involvement Predic-tion. Korean Journal of Radiology, 15, 37-44.
https://doi.org/10.3348/kjr.2014.15.1.37
[33] Fernandez, L.M., Parlade, A.J., Wasser, E.J., et al. (2019) How Re-liable Is CT Scan in Staging Right Colon Cancer? Diseases of the Colon & Rectum, 62, 960-964.
https://doi.org/10.1097/DCR.0000000000001387
[34] Koçak, B., Durmaz, E.Ş., Ateş, E., et al. (2019) Radiomics with Artificial Intelligence: A the Copyediting, Typesetting, Pagination and Proofreading Process, Which May Lead to Differences between This Version and the Version of Record.
[35] Wang, H., Wang, H., Song, L., et al. (2019) Auto-matic Diagnosis of Rectal Cancer Based on CT Images by Deep Learning Method. 2019 12th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI), Suzhou, 19-21 October 2019, 1-5.
https://doi.org/10.1109/CISP-BMEI48845.2019.8965731
[36] Ding, L., Liu, G., Zhang, X., et al. (2020) A Deep Learning Nomogram Kit for Predicting Metastatic Lymph Nodes in Rectal Cancer. Cancer Medicine, 9, 8809-8820.
https://doi.org/10.1002/cam4.3490
[37] Bedrikovetski, S., Dudi-Venkata, N.N., Maicas, G., et al. (2021) Artificial Intelligence for the Diagnosis of Lymph Node Metastases in Patients with Abdominopelvic Malignancy: A Systematic Review and Meta-Analysis. Artificial Intelligence in Medicine, 113, Article ID: 102022.
https://doi.org/10.1016/j.artmed.2021.102022
[38] Lundervold, A.S. and Lundervold, A. (2019) An Overview of Deep Learning in Medical Imaging Focusing on MRI. Zeitschrift für Medizinische Physik, 29, 102-127.
https://doi.org/10.1016/j.zemedi.2018.11.002
[39] Benjamens, S., Dhunnoo, P. and Mesko, B. (2020) The State of Artificial Intelligence-Based FDA-Approved Medical Devices and Algorithms: An Online Database. NPJ Digital Medi-cine, 3, 1-8.
https://doi.org/10.1038/s41746-020-00324-0
[40] Abdullah, Y.I., Schuman, J.S., Shabsigh, R., et al. (2021) Ethics of Artificial Intelligence in Medicine and Ophthalmology. The Asia-Pacific Journal of Ophthalmology, 10, 289-298.
https://doi.org/10.1097/APO.0000000000000397