海军军医大学学报  2024, Vol. 45 Issue (12): 1463-1469   PDF    
胰腺囊性肿瘤恶变风险影像学诊断进展与困境
袁小涵, 边云     
海军军医大学(第二军医大学)第一附属医院放射诊断科, 上海 200433
摘要: 随着影像学检查的普及和人均寿命的延长,胰腺囊性肿瘤(PCN)的检出率越来越高。不同亚型的PCN具有不同的恶变风险,对PCN恶变风险进行准确分层能够为患者提供正确的监测方案并指导手术决策。近年来多项指南明确了多个影像学特征(囊肿大小、壁结节和主胰管管径)为PCN恶变的危险因素,但单个特征衡量标准不一且诊断能力有限,而综合多个特征的模型诊断能力表现欠佳。本文围绕指南中影像学危险因素在PCN恶变预测中的价值及影像组学和人工智能的应用进展进行综述,旨在为PCN的影像学研究提供方向,提高术前PCN危险分层的准确性。
关键词: 胰腺囊性肿瘤    影像组学    人工智能    X线计算机体层摄影术    磁共振成像    
Imaging diagnosis of malignant risk of pancreatic cystic neoplasms: advances and difficulties
YUAN Xiaohan, BIAN Yun     
Department of Radiology, The First Affiliated Hospital of Naval Medical University (Second Military Medical University), Shanghai 200433, China
Abstract: With the popularity of imaging examination and the increase in average life expectancy, the detection rate of pancreatic cystic neoplasm (PCN) is also increasing. Different subtypes of PCN have different risks of malignancy, therefore, accurate stratification of the malignant potential is crucial for providing surveillance plans and making surgical decision. In recent years, many guidelines have identified several imaging features (cyst size, mural nodules, and main pancreatic duct diameter) as risk factors for PCN malignancy. However, the measurement standards for individual feature are not uniform and their diagnostic capabilities are limited; models that integrate multiple features perform poorly in diagnosis. This article reviews the value of imaging risk factors in PCN malignancy prediction and the application of radiomics and artificial intelligence, providing direction for further imaging research and improving the accuracy of preoperative PCN risk stratification.
Key words: pancreatic cystic neoplasms    radiomics    artificial intelligence    X-ray computed tomography    magnetic resonance imaging    

胰腺囊性肿瘤(pancreatic cystic neoplasm,PCN)是指起源于胰腺导管上皮和/或间质组织的囊性肿瘤性病变。PCN分为黏液性肿瘤和非黏液性肿瘤两大类。黏液性PCN主要包括导管内乳头状黏液性肿瘤(intraductal papillary mucinous neoplasm,IPMN)和黏液性囊性肿瘤(mucinous cystic neoplasm,MCN),非黏液性PCN最常见的有浆液性囊腺瘤(serous cystic neoplasm)、囊性神经内分泌肿瘤(cystic neuroendocrine neoplasm)和实性假乳头状肿瘤(solid pseudopapillary tumor)。其中,IPMN和MCN患者是胰腺癌高风险人群,也是早癌筛查的重点人群。国内外多项指南[1-6]的核心内容包括了PCN危险因素的监测和手术指征,提出的PCN危险因素大部分基于影像学特征,可见围绕PCN恶变预测的影像学研究已成为热点。此外,影像组学和人工智能(artificial intelligence,AI)技术也广泛应用于PCN的诊断、鉴别诊断、良恶性预测及预后评估。本文就PCN的影像学研究进展进行综述。

1 PCN的危险因素

目前关于PCN的国际指南主要包括2015年美国胃肠病学协会(American Gastroenterological Association,AGA)制定的《无症状胰腺囊肿诊断和治疗指南》[1],2017年国际胰腺病学会制定的《修订版胰腺导管内乳头状黏液性肿瘤诊断福冈国际共识》[2],2017年美国放射学会(American College of Radiology,ACR)制定的《胰腺偶发囊肿管理——美国放射学会偶发病变委员会白皮书》[3],2018年欧洲胰腺囊性肿瘤研究组制定的《欧洲胰腺囊性肿瘤循证指南》[4],2018年美国胃肠病学会(American College of Gastroenterology,ACG)制定的《胰腺囊肿的诊断和治疗指南》[5],以及2022年中国国家消化病临床研究中心制定的《中国胰腺囊性肿瘤诊断指南(2022年)》[6]。这些指南对PCN的可疑特征和高危特征均进行了定义,基本围绕着PCN的大小、强化壁结节、主胰管扩张程度、囊壁厚度及肿瘤增长速率等,但有所不同(表 1)。

表 1 国际指南对胰腺囊性肿瘤可疑特征和高危特征的定义 Tab 1 Definition of worrisome features and high-risk stigmata of pancreatic cystic neoplasms in international guidelines

2 影像学方法选择

与CT相比,MRI/磁共振胰胆管成像(magnetic resonance cholangiopancreatography,MRCP)可以更灵敏地识别病灶与胰管系统的连通、实性成分或壁结节的存在,以及多灶性PCN,尤其在分支胰管型IPMN(branch duct-IPMN,BD-IPMN)的鉴别诊断上具有独特优势[7-8]。国内外多项指南[1-6]均推荐将MRI作为PCN的首选检查手段。

欧洲指南建议将内镜超声(endoscopic ultrasound,EUS)作为其他放射学检查的补充,EUS在评价胰腺实质和囊性成分方面具有高度准确性,有助于诊断具有手术指征的PCN[4]。用EUS判断IPMN高危征象的效果(AUC=0.733)与2名医师使用增强CT(AUC=0.792、0.830)、MRI(AUC=0.742、0.776)预测恶性IPMN的效果相似(P>0.05)[9]。EUS还能检测出MRI/MRCP遗漏的壁结节,尤其是增强EUS,其特异度为80%,灵敏度为100%[10]。此外,EUS引导的细针穿刺(fine-needle aspiration)具有在整个胰腺长度上采样的能力,可以对固体成分进行细胞学采样,也可用于囊肿液体抽吸以进行蛋白质、代谢物、分子和细胞学评价,尤其增加了对没有可疑特征的小BD-IPMN的诊断效能[2, 11]。尽管如此,由于EUS是一项侵入性检查且诊断准确性与CT和MRI/MRCP相当,不推荐作为诊断PCN的一线检查方法。

3 影像学特征的诊断困境 3.1 肿瘤大小

肿瘤大小是PCN患者手术决策的关键参数,但在文献中尚没有统一的推荐手术的截断值。修订后的福冈共识[2]和ACR白皮书[3]将囊肿≥30 mm作为可疑特征,推荐EUS引导的细针穿刺进一步检查。而欧洲指南[4]则放宽了标准,将囊肿≥40 mm设置为IPMN和MCN手术的相对适应证。AGA指南[1]和ACG指南[5]选择了更激进的诊断标准,将囊肿≥30 mm作为推荐手术的高危特征。

多项将行手术切除的BD-IPMN患者纳入分析的研究发现,囊肿≥30 mm对BD-IPMN伴高级别异型增生(high grade dysplasia,HGD)或浸润性癌(invasive cancer)的阳性预测值为27%~33%,良性与恶性组间囊肿大小差异无统计学意义(P>0.05)[12-14]。Del Chiaro等[15]对444例IPMN患者进行了随访,结果表明在不存在其他危险因素的情况下,囊肿<40 mm的BD-IPMN患者可继续选择随访观察。现有研究大多用肿瘤最大面长径与短径的平均值来代表肿瘤大小,但肿瘤往往并不是规则球形,有学者提出测量肿瘤三维体积可能具有更高的准确性。Pozzi Mucelli等[16]研究表明,无论是BD-IPMN还是混合型IPMN(mixed-type IPMN,MT-IPMN)患者,囊肿体积与肿瘤恶性度都不相关(OR=1.01,95% CI 0.99~1.02,P=0.12)。这表明单独的囊肿大小并不适合作为IPMN恶变的危险因素,需要寻找其他可疑特征对患者进行分层[17]

与首次检出时囊肿大小相比,随访期间的PCN生长速率可能是预测恶变风险的更准确参数。一项纳入201例囊肿<30 mm且无其他危险因素的BD-IPMN患者的研究结果表明,与生长速率<2 mm/年的囊肿相比,生长速率≥2 mm/年的囊肿恶变风险更高(5年恶变风险:45.5% vs 1.8%)[18]。Kwong等[19]发现当生长速率为2~<5 mm/年时,BD-IPMN恶变的HR为11.4(95 % CI 2.2~58.6)(P=0.004),当生长速率≥5 mm/年时HR为19.5(95 % CI 2.4~157.8),恶变风险均高于生长速率<2 mm/年的BD-IPMN(P=0.004、0.005)。但是,也有研究表明在BD-IPMN患者中,无论是囊肿生长速率≥5 mm/年还是增长速率(总生长大小/肿瘤初始大小)≥30%/年都与囊肿恶变无关[20]

综上所述,无论是单纯的囊肿大小还是囊肿的生长速率,与BD-IPMN恶变风险的相关性都存在争议,并不适合作为判断肿瘤恶变的标准。但是,上述研究大多局限于回顾性研究及有限的样本量,未来需要更有力的前瞻性证据证实肿瘤大小和生长速率在评估BD-IPMN恶变风险中的重要性。

3.2 壁结节和实性成分

壁结节是指突出于囊壁/分隔或胰管壁的实性成分,也称之为实性成分。相对于PCN囊肿的大小,壁结节和实性成分的存在更能预测IPMN及MCN恶变[21]。一项meta分析显示,壁结节对IPMN恶变风险的阳性预测值为62.2%,且其是唯一可靠预测各种类型IPMN-HGD/浸润性癌的因子(标准化均差为0.79)[22]。然而,目前尚未确定强化壁结节大小的最佳截断值。AGA指南[1]、ACR白皮书[3]和ACG指南[5]认为实性成分或强化壁结节无论其大小,只要出现即可作为手术指征。修订后的福冈共识[2]和欧洲指南[4]将强化壁结节≥5 mm作为PCN明确的手术指征。Harima等[23]的研究结果认为,将壁结节≥8.8 mm为诊断标准预测恶性BD-IPMN最准确,灵敏度为100%,特异度为86%。CT、MRI及EUS上强化壁结节或实性成分的存在对预测IPMN恶变的效能相似(OR分别为1.8、1.36、1.47,P>0.05)[9]

除了对壁结节大小的探究,Niiya等[24]纳入17例BD-IPMN患者进行的小样本研究探讨了壁结节位置与BD-IPMN恶性度的相关性,结果表明壁结节靠近主胰管一侧的BD-IPMN恶性比例高于壁结节远离主胰管侧的BD-IPMN,且靠近主胰管的壁结节组织学亚型多为肠型,而远离主胰管侧的壁结节多为胃型。虽然该研究样本量很小,缺乏说服力,但为将来的研究提供了一定的参考。

3.3 主胰管扩张

主胰管扩张是主胰管型IPMN(main duct-IPMN,MD-IPMN)或同时累及主胰管和分支胰管的MT-IPMN发生恶变的高危因素。尽管主胰管管径≥10 mm是公认的高危特征,但是作为手术指征其截断值仍存在争议。Del Chiaro等[25]研究表明,主胰管扩张至5~9.9 mm(n=286)与IPMN-HGD(OR=2.74,95% CI 1.80~4.16)和浸润性癌(OR=4.42,95% CI 2.55~7.66)的风险增加有关;当主胰管扩张至>10 mm(n=150)时发生IPMN-HGD(OR=6.57,95% CI 3.94~10.98)和浸润性癌(OR=15.07,95% CI 8.21~27.65)的风险更高;而主胰管管径为5~7 mm被认为是区分良恶性IPMN的最佳指标。Crippa等[26]将312例MD-IPMN和MT-IPMN患者纳入研究,结果显示胰头处主胰管管径≥9 mm(AUC=0.66,OR=2.6)和胰体尾处主胰管管径≥7 mm(AUC=0.70,OR=4.3)是IPMN恶变的独立危险因素;他们进一步研究发现,在没有其他可疑特征或高危特征的情况下,胰头处主胰管管径≤8 mm和体尾部主胰管管径≤6 mm的肿瘤恶变风险较低。

BD-IPMN也会出现主胰管扩张,这种扩张多是因病灶分泌大量黏液至主胰管所致,主胰管扩张能否用于预测BD-IPMN的恶变风险存在争议[27-29]。Crippa等[30]报告的144例BD-IPMN手术患者中,主胰管扩张至5~9 mm(n=7)的患者术后病理证实均未发生HGD/浸润性癌。而在Robles等[13]的研究中,主胰管扩张至5~9 mm(n=44)与BD-IPMN恶变独立相关(OR=3.395,95% CI 1.349~8.543,P=0.009)。

一项纳入了所有类型IPMN(n=1 688)并进行中位时间为60个月随访的研究表明,在随访过程中,30例出现主胰管扩张的患者中仅有1例(3%)发生恶变;主胰管扩张的患者与主胰管未扩张的患者5年累积恶变风险差异无统计学意义(4% vs 1.2%,P=0.448)[31]。由此可见,主胰管扩张作为IPMN患者手术指征的证据不够。未来还需更有力的前瞻性研究证实主胰管扩张在IPMN恶性预测中的作用及探讨最佳诊断阈值。

3.4 各指南诊断PCN恶变的能力

在Lekkerkerker等[32]纳入115例接受胰腺切除术的PCN患者的研究中,按照福冈共识、欧洲指南和AGA指南决定手术的准确率分别为54%、53%、59%;AGA指南可使多数良性病变患者避免进行不必要的手术,但会遗漏12%的IPMN-HGD/浸润性癌患者。另一项包括21项研究3 723例PCN患者的meta分析[33]评估了福冈共识和AGA指南的高危特征对PCN的诊断效能,结果显示,福冈共识和AGA指南的诊断准确性相似,但总体不令人满意,福冈共识预测晚期肿瘤的合并灵敏度和合并特异度分别为0.67和0.64,AGA指南的合并灵敏度和合并特异度分别为0.59和0.77。将高危特征与患者年龄(>60岁)、血清糖类抗原19-9水平(>37 U/mL)和急性胰腺炎病史相结合建立的模型,对恶性IPMN的诊断特异度高达90.1%,但灵敏度不高(50.7%)[34]

4 影像组学和AI助力PCN诊断

PCN的术前风险分层仍是亟待解决的难题。利用AI提取影像组学特征是近几年新兴的图像评估方法,其核心是对图像中肉眼难以识别的纹理特征进行分析。目前大部分PCN的影像组学研究基于CT图像。Chakraborty等[35]在回顾性研究中利用影像组学于术前对103例BD-IPMN患者进行了良恶性预测,联合囊肿的纹理特征及囊壁和囊肿强化程度建立的影像组学模型预测恶性BD-IPMN的AUC值为0.78;加入年龄、性别、囊肿大小、囊肿内实性成分等参数后模型的诊断效能得到了提高(AUC值为0.81)。Cui等[36]基于MRI图像将9个影像组学特征与糖类抗原19-9及主胰管管径相结合建立了模型,用于预测BD-IPMN恶变风险,其列线图诊断病灶恶变的AUC值为0.903(训练集)和0.884(外部验证集)。另有多项研究表明,影像组学模型可获得优于指南的风险分层能力,且与单纯影像学参数或单纯临床参数建立的模型相比,影像组学与临床参数的结合能明显提高模型鉴别诊断的能力[36-38]

与影像组学相比,基于人工神经网络的深度学习实现了对影像学信息“端对端”的自动化分析,能对图像特征进行自动提取和筛选,避免人为干预可能丢失的图像深层信息,还极大地提高了对大批量数据的处理能力。Kuwahara等[39]利用EUS图像对IPMN患者进行了AI评分,该评分对IPMN恶变的准确度可达94%,高于人工诊断的准确度(56%)及强化壁结节的诊断准确度(68%)。基于MRI的深度学习预测模型对IPMN-HGD/浸润性癌诊断的灵敏度及特异度分别为75%和78%,达到了与AGA指南及福冈共识相似的诊断效能(AUC值分别为0.76、0.77、0.78,P=0.90)[40]。用EUS图像进行IPMN患者风险分层,其诊断准确度(99.6%)高于欧洲指南、AGA指南及修订后的福冈共识(51.8%~71.3%)[41]

影像组学和深度学习为PCN的术前风险分层提供了可靠的非侵入性检查手段,有助于患者临床治疗方案及随访策略的制定,然而其结果的不可解释性使其应用受限。Huang等[42]在胃癌的研究中将影像组学特征与临床病理变量相关联,可在某种程度上增加结果的可信度与可解释性。Lee等[43]的研究将训练集中所有急性脑出血患者的图像块进行颜色编码并形成热图,人们可以通过比较CT图像和热图之间的形态相似性来理解模型预测的可靠性。

此外,福冈共识还强调了IPMN的组织学亚型(胃型、肠型和胰胆管型)与恶性潜能的相关性,胰胆管型IPMN具有较高的恶性转化风险,即使没有可疑特征或恶性肿瘤的高风险特征,也可以考虑手术[2]。AI在利用标准影像学图像实现术前对组织学亚型的预测方面有巨大潜能,未来还需在此方面进行大量的研究。

5 小结

影像学是诊断PCN良恶性最重要的手段,但是目前仍存在较多需要解决的问题。例如,虽然指南明确了单个影像学特征(囊肿大小、壁结节和主胰管管径)为恶变的危险因素,但是单个特征衡量标准不一且诊断能力有限;综合多个特征的模型诊断能力表现欠佳;影像组学和AI仍未在PCN上广泛应用。临床迫切需要全新、无创的手段以精准评估IPMN良恶性,使患者得到及时治疗或免于不必要的胰腺手术。未来,寻找更合适的影像组学特征,建立AI诊断模型辅助医生对PCN患者的危险程度进行精准分层,最终提高PCN患者的生活质量,是胰腺多学科医生需要努力的方向。

参考文献
[1]
VEGE S S, ZIRING B, JAIN R, et al. American Gastroenterological Association institute guideline on the diagnosis and management of asymptomatic neoplastic pancreatic cysts[J]. Gastroenterology, 2015, 148(4): 819-822. DOI:10.1053/j.gastro.2015.01.015
[2]
TANAKA M, FERNÁNDEZ-DEL CASTILLO C, KAMISAWA T, et al. Revisions of international consensus Fukuoka guidelines for the management of IPMN of the pancreas[J]. Pancreatology, 2017, 17(5): 738-753. DOI:10.1016/j.pan.2017.07.007
[3]
MEGIBOW A J, BAKER M E, MORGAN D E, et al. Management of incidental pancreatic cysts: a white paper of the ACR Incidental Findings Committee[J]. J Am Coll Radiol, 2017, 14(7): 911-923. DOI:10.1016/j.jacr.2017.03.010
[4]
European Study Group on Cystic Tumours of the Pancreas. European evidence-based guidelines on pancreatic cystic neoplasms[J]. Gut, 2018, 67(5): 789-804. DOI:10.1136/gutjnl-2018-316027
[5]
ELTA G H, ENESTVEDT B K, SAUER B G, et al. ACG clinical guideline: diagnosis and management of pancreatic cysts[J]. Am J Gastroenterol, 2018, 113(4): 464-479. DOI:10.1038/ajg.2018.14
[6]
李兆申, 金震东, 李汛. 中国胰腺囊性肿瘤诊断指南(2022年)[J]. 临床肝胆病杂志, 2023, 39(2): 290-298. DOI:10.3969/j.issn.1001-5256.2023.02.007
[7]
SAHANI D V, KAMBADAKONE A, MACARI M, et al. Diagnosis and management of cystic pancreatic lesions[J]. AJR Am J Roentgenol, 2013, 200(2): 343-354. DOI:10.2214/AJR.12.8862
[8]
JONES M J, BUCHANAN A S, NEAL C P, et al. Imaging of indeterminate pancreatic cystic lesions: a systematic review[J]. Pancreatology, 2013, 13(4): 436-442. DOI:10.1016/j.pan.2013.05.007
[9]
CHOI S Y, KIM J H, YU M H, et al. Diagnostic performance and imaging features for predicting the malignant potential of intraductal papillary mucinous neoplasm of the pancreas: a comparison of EUS, contrast-enhanced CT and MRI[J]. Abdom Radiol, 2017, 42(5): 1449-1458. DOI:10.1007/s00261-017-1053-3
[10]
ALVAREZ-SÁNCHEZ M V, NAPOLÉON B. Contrast-enhanced harmonic endoscopic ultrasound imaging: basic principles, present situation and future perspectives[J]. World J Gastroenterol, 2014, 20(42): 15549-15563. DOI:10.3748/wjg.v20.i42.15549
[11]
ASSARZADEGAN N, THOMPSON E, SALIMIAN K, et al. Pathology of intraductal papillary mucinous neoplasms[J]. Langenbecks Arch Surg, 2021, 406(8): 2643-2655. DOI:10.1007/s00423-021-02201-0
[12]
SAHORA K, MINO-KENUDSON M, BRUGGE W, et al. Branch duct intraductal papillary mucinous neoplasms: does cyst size change the tip of the scale? A critical analysis of the revised international consensus guidelines in a large single-institutional series[J]. Ann Surg, 2013, 258(3): 466-475. DOI:10.1097/SLA.0b013e3182a18f48
[13]
ROBLES E P, MAIRE F, CROS J, et al. Accuracy of 2012 International Consensus Guidelines for the prediction of malignancy of branch-duct intraductal papillary mucinous neoplasms of the pancreas[J]. United European Gastroenterol J, 2016, 4(4): 580-586. DOI:10.1177/2050640615623370
[14]
NGUYEN A H, TOSTE P A, FARRELL J J, et al. Current recommendations for surveillance and surgery of intraductal papillary mucinous neoplasms may overlook some patients with cancer[J]. J Gastrointest Surg, 2015, 19(2): 258-265. DOI:10.1007/s11605-014-2693-z
[15]
DEL CHIARO M, ATEEB Z, HANSSON M R, et al. Survival analysis and risk for progression of intraductal papillary mucinous neoplasia of the pancreas (IPMN) under surveillance: a single-institution experience[J]. Ann Surg Oncol, 2017, 24(4): 1120-1126. DOI:10.1245/s10434-016-5661-x
[16]
POZZI MUCELLI R M, MORO C F, DEL CHIARO M, et al. Branch-duct intraductal papillary mucinous neoplasm (IPMN): are cyst volumetry and other novel imaging features able to improve malignancy prediction compared to well-established resection criteria?[J]. Eur Radiol, 2022, 32(8): 5144-5155. DOI:10.1007/s00330-022-08650-5
[17]
JANG J Y, PARK T, LEE S, et al. Validation of international consensus guidelines for the resection of branch duct-type intraductal papillary mucinous neoplasms[J]. Br J Surg, 2014, 101(6): 686-692. DOI:10.1002/bjs.9491
[18]
KANG M J, JANG J Y, KIM S J, et al. Cyst growth rate predicts malignancy in patients with branch duct intraductal papillary mucinous neoplasms[J]. Clin Gastroenterol Hepatol, 2011, 9(1): 87-93. DOI:10.1016/j.cgh.2010.09.008
[19]
KWONG W T, LAWSON R D, HUNT G, et al. Rapid growth rates of suspected pancreatic cyst branch duct intraductal papillary mucinous neoplasms predict malignancy[J]. Dig Dis Sci, 2015, 60(9): 2800-2806. DOI:10.1007/s10620-015-3679-8
[20]
CHAFIC A E, EL HAJJ I I, DEWITT J, et al. Does cyst growth predict malignancy in branch duct intraductal papillary mucinous neoplasms? Results of a large multicenter experience[J]. Dig Liver Dis, 2018, 50(9): 961-968. DOI:10.1016/j.dld.2018.04.022
[21]
ZHONG N, ZHANG L, TAKAHASHI N, et al. Histologic and imaging features of mural nodules in mucinous pancreatic cysts[J]. Clin Gastroenterol Hepatol, 2012, 10(2): 192-198, 198.e1-198.e2. DOI:10.1016/j.cgh.2011.09.029
[22]
MARCHEGIANI G, ANDRIANELLO S, BORIN A, et al. Systematic review, meta-analysis, and a high-volume center experience supporting the new role of mural nodules proposed by the updated 2017 international guidelines on IPMN of the pancreas[J]. Surgery, 2018, 163(6): 1272-1279. DOI:10.1016/j.surg.2018.01.009
[23]
HARIMA H, KAINO S, SHINODA S, et al. Differential diagnosis of benign and malignant branch duct intraductal papillary mucinous neoplasm using contrast-enhanced endoscopic ultrasonography[J]. World J Gastroenterol, 2015, 21(20): 6252-6260. DOI:10.3748/wjg.v21.i20.6252
[24]
NIIYA F, OHIKE N, NOROSE T, et al. Can the location of the mural nodule indicate benign or malignant in branch duct-type intraductal papillary mucinous neoplasm of the pancreas?[J]. Pancreatology, 2020, 20(7): 1379-1385. DOI:10.1016/j.pan.2020.08.006
[25]
DEL CHIARO M, BECKMAN R, ATEEB Z, et al. Main duct dilatation is the best predictor of high-grade dysplasia or invasion in intraductal papillary mucinous neoplasms of the pancreas[J]. Ann Surg, 2020, 272(6): 1118-1124. DOI:10.1097/SLA.0000000000003174
[26]
CRIPPA S, ALEOTTI F, LONGO E, et al. Main duct thresholds for malignancy are different in intraductal papillary mucinous neoplasms of the pancreatic head and body-tail[J]. Clin Gastroenterol Hepatol, 2022, 20(2): 390-399. DOI:10.1016/j.cgh.2020.12.028
[27]
DAL BORGO C, PERRI G, BORIN A, et al. The clinical management of main duct intraductal papillary mucinous neoplasm of the pancreas[J]. Dig Surg, 2019, 36(2): 104-110. DOI:10.1159/000486869
[28]
RODRIGUEZ J R, SALVIA R, CRIPPA S, et al. Branch-duct intraductal papillary mucinous neoplasms: observations in 145 patients who underwent resection[J]. Gastroenterology, 2007, 133(1): 72-79. DOI:10.1053/j.gastro.2007.05.010
[29]
ATTIYEH M A, FERNÁNDEZ-DEL CASTILLO C, EFISHAT M A, et al. Development and validation of a multi-institutional preoperative nomogram for predicting grade of dysplasia in intraductal papillary mucinous neoplasms (IPMNs) of the pancreas: a report from the pancreatic surgery consortium[J]. Ann Surg, 2018, 267(1): 157-163. DOI:10.1097/SLA.0000000000002015
[30]
CRIPPA S, PEZZILLI R, BISSOLATI M, et al. Active surveillance beyond 5 years is required for presumed branch-duct intraductal papillary mucinous neoplasms undergoing non-operative management[J]. Am J Gastroenterol, 2017, 112(7): 1153-1161. DOI:10.1038/ajg.2017.43
[31]
MARCHEGIANI G, ANDRIANELLO S, MORBIN G, et al. Importance of main pancreatic duct dilatation in IPMN undergoing surveillance[J]. Br J Surg, 2018, 105(13): 1825-1834. DOI:10.1002/bjs.10948
[32]
LEKKERKERKER S J, BESSELINK M G, BUSCH O R, et al. Comparing 3 guidelines on the management of surgically removed pancreatic cysts with regard to pathological outcome[J]. Gastrointest Endosc, 2017, 85(5): 1025-1031. DOI:10.1016/j.gie.2016.09.027
[33]
WU J, WANG Y, LI Z, et al. Accuracy of Fukuoka and American Gastroenterological Association guidelines for predicting advanced neoplasia in pancreatic cyst neoplasm: a meta-analysis[J]. Ann Surg Oncol, 2019, 26(13): 4522-4536. DOI:10.1245/s10434-019-07921-8
[34]
SHIN S H, HAN D J, PARK K T, et al. Validating a simple scoring system to predict malignancy and invasiveness of intraductal papillary mucinous neoplasms of the pancreas[J]. World J Surg, 2010, 34(4): 776-783. DOI:10.1007/s00268-010-0416-5
[35]
CHAKRABORTY J, MIDYA A, GAZIT L, et al. CT radiomics to predict high-risk intraductal papillary mucinous neoplasms of the pancreas[J]. Med Phys, 2018, 45(11): 5019-5029. DOI:10.1002/mp.13159
[36]
CUI S, TANG T, SU Q, et al. Radiomic nomogram based on MRI to predict grade of branching type intraductal papillary mucinous neoplasms of the pancreas: a multicenter study[J]. Cancer Imaging, 2021, 21(1): 26. DOI:10.1186/s40644-021-00395-6
[37]
PERMUTH J B, CHOI J, BALARUNATHAN Y, et al. Combining radiomic features with a miRNA classifier may improve prediction of malignant pathology for pancreatic intraductal papillary mucinous neoplasms[J]. Oncotarget, 2016, 7(52): 85785-85797. DOI:10.18632/oncotarget.11768
[38]
ATTIYEH M A, CHAKRABORTY J, GAZIT L, et al. Preoperative risk prediction for intraductal papillary mucinous neoplasms by quantitative CT image analysis[J]. HPB, 2019, 21(2): 212-218. DOI:10.1016/j.hpb.2018.07.016
[39]
KUWAHARA T, HARA K, MIZUNO N, et al. Usefulness of deep learning analysis for the diagnosis of malignancy in intraductal papillary mucinous neoplasms of the pancreas[J]. Clin Transl Gastroenterol, 2019, 10(5): 1-8. DOI:10.14309/ctg.0000000000000045
[40]
CORRAL J E, HUSSEIN S, KANDEL P, et al. Deep learning to classify intraductal papillary mucinous neoplasms using magnetic resonance imaging[J]. Pancreas, 2019, 48(6): 805-810. DOI:10.1097/MPA.0000000000001327
[41]
SCHULZ D, HEILMAIER M, PHILLIP V, et al. Accurate prediction of histological grading of intraductal papillary mucinous neoplasia using deep learning[J]. Endoscopy, 2023, 55(5): 415-422. DOI:10.1055/a-1971-1274
[42]
HUANG W, JIANG Y, XIONG W, et al. Noninvasive imaging of the tumor immune microenvironment correlates with response to immunotherapy in gastric cancer[J]. Nat Commun, 2022, 13(1): 5095. DOI:10.1038/s41467-022-32816-w
[43]
LEE H, YUNE S, MANSOURI M, et al. An explainable deep-learning algorithm for the detection of acute intracranial haemorrhage from small datasets[J]. Nat Biomed Eng, 2019, 3(3): 173-182. DOI:10.1038/s41551-018-0324-9