Artificial Intelligence in Pancreatic Intraductal Papillary Mucinous Neoplasm Imaging: A Systematic Review

Artificial Intelligence in Pancreatic Intraductal Papillary Mucinous Neoplasm Imaging: A Systematic Review | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Artificial Intelligence in Pancreatic Intraductal Papillary Mucinous Neoplasm Imaging: A Systematic Review View ORCID Profile Muhammad Ibtsaam Qadir , Jackson A. Baril , Michele T. Yip-Schneider , Duane Schonlau , Thi Thanh Thoa Tran , C. Max Schmidt , View ORCID Profile Fiona R. Kolbinger doi: https://doi.org/10.1101/2025.01.08.25320130 Muhammad Ibtsaam Qadir 1 Weldon School of Biomedical Engineering, Purdue University , West Lafayette, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Muhammad Ibtsaam Qadir Jackson A. Baril 2 Division of Surgical Oncology, Department of Surgery, Indiana University School of Medicine , Indianapolis, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Michele T. Yip-Schneider 2 Division of Surgical Oncology, Department of Surgery, Indiana University School of Medicine , Indianapolis, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Duane Schonlau 3 Department of Radiology, Indiana University School of Medicine , Indianapolis, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Thi Thanh Thoa Tran 2 Division of Surgical Oncology, Department of Surgery, Indiana University School of Medicine , Indianapolis, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site C. Max Schmidt 2 Division of Surgical Oncology, Department of Surgery, Indiana University School of Medicine , Indianapolis, IN, USA 4 Department of Biochemistry and Molecular Biology, Indiana University School of Medicine , Indianapolis, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Fiona R. Kolbinger 1 Weldon School of Biomedical Engineering, Purdue University , West Lafayette, IN, USA 5 Regenstrief Center for Healthcare Engineering (RCHE), Purdue University , West Lafayette, IN, USA 6 Department of Biostatistics and Health Data Science, Richard M. Fairbanks School of Public Health, Indiana University , Indianapolis, IN, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Fiona R. Kolbinger For correspondence: fkolbing{at}purdue.edu Abstract Full Text Info/History Metrics Data/Code Preview PDF ABSTRACT Background Based on the Fukuoka and Kyoto international consensus guidelines, the current clinical management of intraductal papillary mucinous neoplasm (IPMN) largely depends on imaging features. While these criteria are highly sensitive in detecting high-risk IPMN, they lack specificity, resulting in surgical overtreatment. Artificial Intelligence (AI)-based medical image analysis has the potential to augment the clinical management of IPMNs by improving diagnostic accuracy. Methods Based on a systematic review of the academic literature on AI in IPMN imaging, 1041 publications were identified of which 25 published studies were included in the analysis. The studies were stratified based on prediction target, underlying data type and imaging modality, patient cohort size, and stage of clinical translation and were subsequently analyzed to identify trends and gaps in the field. Results Research on AI in IPMN imaging has been increasing in recent years. The majority of studies utilized CT imaging to train computational models. Most studies presented computational models developed on single-center datasets (n=11,44%) and included less than 250 patients (n=18,72%). Methodologically, convolutional neural network (CNN)-based algorithms were most commonly used. Thematically, most studies reported models augmenting differential diagnosis (n=9,36%) or risk stratification (n=10,40%) rather than IPMN detection (n=5,20%) or IPMN segmentation (n=2,8%). Conclusion This systematic review provides a comprehensive overview of the research landscape of AI in IPMN imaging. Computational models have potential to enhance the accurate and precise stratification of patients with IPMN. Multicenter collaboration and datasets comprising various modalities are necessary to fully utilize this potential, alongside concerted efforts towards clinical translation. Download figure Open in new tab Highlights Artificial Intelligence holds promise in the field of IPMN by augmenting IPMN detection, differentiation of different types of pancreatic cysts, stratifying malignant progression risk, and automating the analysis of IPMN imaging through computational cyst segmentation. The majority of studies related to AI-based analysis of IPMN imaging use single-center patient cohorts of less than 250 patients to develop and validate computational models and consider imaging as the only data modality. Reporting transparency of existing studies on AI in IPMN imaging is limited and there remains a scarcity of comprehensive, multimodal approaches as well as clinical translation. Introduction Pancreatic cancer is one the most aggressive cancers, with a 5-year overall survival rate that remains under 15% [ 1 , 2 ]. Pancreatic ductal adenocarcinoma (PDAC) is the most prevalent solid tumor of the pancreas and constitutes more than 85% of all pancreatic cancer incidences [ 3 , 4 ]. The majority of patients receive a diagnosis at a late and incurable stage of the disease. Some pancreatic cysts, often incidentally detected on imaging [ 5 ], are potential precursors to pancreatic cancer, presenting a window of opportunity for early detection and cancer prevention. The most common type of pancreatic cyst is Intraductal papillary mucinous neoplasm (IPMN) [ 6 ]. About 10% of the general population over 50 years of age are estimated to carry IPMN [ 7 ]. The frequency of invasive carcinoma in resected branch duct (BD-)IPMN and main duct (MD-)IPMN ranges from 10 to 45% [ 8 – 10 ] and from 40 to 75% [ 10 – 12 ], respectively. In patients with BD-IPMN, the 5-year and 15-year malignant progression rates are 3.3% and 12%, respectively [ 13 ]. Of all PDACs, 8-10% are estimated to derive from IPMN [ 14 , 15 ]. The management of IPMN primarily depends on imaging features, cytology, and clinical variables such as CA19-9 serum levels and pancreatitis [ 7 , 16 ]. Patients with MD-IPMN and pancreatic duct dilation to a diameter of 10 mm or more or a nodule of at least 5 mm are generally recommended surgical resection; in patients with BD-IPMN, imaging features such as main duct dilation to a diameter of 5-9 mm, enhancing mural nodule less than 5 mm, cyst size ≥ 30 mm or a cyst growth rate ≥ 2.5 mm/year are considered indicative of higher malignant potential warranting consideration for surgical resection [ 10 ]. Patients carrying BD-IPMN without such “worrisome features” are recommended surveillance every 6 or 12 months, including imaging and a clinical examination [ 10 ]. Given the substantial risks associated with pancreatic surgery including mortality [ 17 ] and morbidity [ 18 ] such as post-pancreatectomy diabetes mellitus [ 19 ], recommendations for surgical resection are generally determined by experts who evaluate each case using all available diagnostic modalities [ 20 ]. The Fukuoka [ 10 ] and Kyoto international consensus guidelines [ 21 ] represent the gold standard for IPMN management; however, while they are sensitive, they lack specificity [ 22 ], Consequently, patient risk stratification is often not accurate, either overestimating malignant transformation risk, resulting in surgical overtreatment [ 23 , 24 ], or underestimating the malignant potential, resulting in surveillance of malignant IPMN (undertreatment) [ 25 ]. Artificial Intelligence (AI), particularly deep learning, can detect characteristic patterns in large medical imaging datasets such as radiological, histopathological, and endoscopic imaging, that are beyond what humans can identify [ 26 , 27 ]. Similarly, traditional natural language processing (NLP) models and, more recently, large language models (LLMs), facilitate the interpretation of written reports, which may support clinical decision-making and personalization of patient care [ 28 , 29 ]. The complex management of IPMN patients is an ideal use case for AI-based computational tools, which could contribute to clinical decision-making and risk stratification based on available data, particularly imaging as the primary non-invasive source of information about IPMN malignant potential. This systematic review provides an overview of the scientific literature on AI in IPMN imaging. We summarize relevant clinical and technological aspects of published studies, including patient selection criteria and cohort sizes, prediction targets, data modalities, methodological characteristics, performance metrics, translational considerations, and data availability. By highlighting the research trends and gaps, this review aims to showcase the potential of computational methods in the context of the clinical management of IPMN. Methods Search Strategy We performed this systematic review following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [ 30 ] and AMSTAR (A MeaSurement Tool to Assess systematic Reviews) [ 31 ] guidelines. To cover the interdisciplinary literature on AI in IPMN imaging, our search strategy included four literature databases (PubMed, IEEE Xplore, Embase, and CINAHL [Cumulative Index to Nursing and Allied Health Literature]), and three preprint servers (ArXiv, medRxiv, and bioRxiv). Three authors with experience in medical AI research (MIQ, JAB and FRK) and a librarian focusing on biomedical literature contributed to the development of the search strategy. Medical subject headings (MeSH) terms were used to specify the search term. PubMed, IEEE Xplore, Embase, CINAHL, and ArXiv were searched on January 31, 2024, without language restrictions, for literature published since database inception, using the following search term: ((“Pancreatic Neoplasms”[MeSH Terms]) OR (“Pancreatic Cyst”[MeSH Terms]) OR (“Pancrea* Neoplasm*”[Title/Abstract])) AND ((“Diagnostic Imaging”[MeSH Terms]) OR (diagnosis[MeSH Terms]) OR (Record*[Title/Abstract]) OR (Tomography[Title/Abstract]) OR (Radiolog*[Title/Abstract]) OR (methods[MeSH Terms]) OR (“Electronic Medical Record”[Title/Abstract]) OR (“EMR”[Title/Abstract]) OR (Report*[Title/Abstract])) AND ((“Artificial Intelligence”[MeSH Terms]) OR (“Deep Learning”[Title/Abstract]) OR (“Natural Language Processing”[MeSH Terms]) OR (“neural network*”[Title/Abstract]) OR (“Machine Learning”[Title/Abstract]) OR (AI[Title/Abstract])). The format of the search term was modified per the databases’ requirements. MedRxiv and bioRxiv were searched on January 31, 2024, using the following, shorter search term to comply with the character restrictions of the respective search tools: (“Learning” OR “Artificial Intelligence”) AND (“Pancreas” OR “Pancreatic”) AND (“Imaging” OR “EMR”). Study Selection Duplicate studies were removed automatically and manually. All studies were independently screened by two raters: one physician (JAB) with three years of experience in IPMN management and research and one engineer (MIQ) with two years of experience in AI and computational medical image analysis, using the Covidence platform (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia, https://www.covidence.org/ ). Screening results were blinded until the completion of each reviewer’s individual screening. The screening process was carried out in two iterations: First, screening based on title and abstract, and second, full-text screening. In each iteration, conflicts between the two reviewers were resolved by independent assessment of a third reviewer (FRK). The inclusion criteria were: (i) analysis of IPMN (and synonyms: mucinous cyst, mucinous cystic lesion) imaging, (ii) development of computational models (machine learning, deep learning, or NLP) for the analysis of medical images and/or imaging reports, (iii) imaging modality being Endoscopic Ultrasonography (EUS) or tomography, i.e. Magnetic Resonance Imaging (MRI) or Computed Tomography (CT), (iv) original research, (v) analysis of human data, and (vi) the article being in English. Non-original articles and publications (i) not analyzing any data from patients with IPMN or (ii) not developing computational models for data analysis (i.e., statistical methods, manually curated imaging features), were excluded. Data Extraction and Analysis One reviewer with expertise in AI and deep learning-based medical image analysis (MIQ) extracted the following data from each included study based on a data extraction protocol ( Supplementary Table 1 ): (i) year of publication, (ii) prediction targets, (iii) ground truth source, (iv) number of centers in the study, (v) patient selection criteria, (vi) total number of patients on which the AI model was developed, (vii) number of patients with IPMN on which the AI model was developed, (viii) number of patients in the training set, (ix) number of patients in test/validation set, (x) validation procedure, (xi) data/imaging type, (xii) computational methodology, (xiii) model performance, (xiv) stage of clinical translation, and (xv) data availability. To assess the risk of bias, two reviewers (MIQ and JAB) independently evaluated the studies using PROBAST [ 32 ] (prediction model risk of bias assessment tool), with a third reviewer (FRK) to resolve discordances. The overall risk of bias for each study was categorized as low, high, or unclear, based on the evaluations across four domains: participants, predictors, outcomes, and analysis. We systematically categorized the studies into four use cases based on the prediction targets: (i) detection, (ii) segmentation, (iii) differential diagnosis, and (iv) risk stratification. Detection encompasses the identification of the presence of pancreatic cysts including IPMN. Segmentation refers to the precise localization of the pancreatic cyst. This use case included publications developing a segmentation method for pixel- or voxel-wise delineation of IPMN or other pancreatic cysts in the radiological images. Differential diagnosis concentrates on differentiating between types of pancreatic cysts including cysts other than IPMN. Risk stratification includes classifying the IPMN and other pancreatic cysts into IPMN with low-grade dysplasia (benign cysts) or IPMN with high-grade dysplasia (malignant cysts). To provide a comprehensive account of the previous research, we used descriptive statistics to examine the data for each category. Key findings are summarized to highlight trends and gaps in the current literature. Study Registration This systematic review was prospectively registered at OSF on January 4, 2024 ( https://doi.org/10.17605/OSF.IO/NVGXB ) and at PROSPERO on August 13, 2024 (CRD42024575960). The protocol was not amended or changed. Results Search Results The search strategy, when queried across all the databases and preprint servers, resulted in 1041 publications (PubMed: n = 558, Embase: n = 199, CINAHL: n = 51, IEEE Xplore: n = 45, bioRxiv: n = 137, medRxiv: n = 47, arXiv: n = 4, Figure 1 ). After the removal of duplicates (n = 125), 916 unique publications were screened. Download figure Open in new tab Figure 1: Flowchart of the systematic review and meta-analysis according to the PRISMA 2020 statement for reporting systematic reviews. Based on the systematic review of the publications on AI in IPMN imaging, 25 articles published between 2018 and 2023 were included in this analysis. Based on title and abstract screening, 40 studies were selected for full-text screening. A total of 15 publications were excluded based on their prediction target (n = 6), data modality (n = 6) or for not developing any AI model (n = 3). Specifically, studies involving deep learning-based MRI reconstruction were excluded, as these fell outside the scope of diagnosis or risk stratification of IPMN. Additionally, studies using imaging modalities other than MRI, CT, or EUS were excluded to align with consensus guidelines for IPMN management [ 10 , 21 ] and American College of Radiology’s (ACR) Appropriateness Criteria for pancreatic cysts [ 33 ]. This resulted in the inclusion of 25 publications in the systematic review ( Figure 1 ). Study Characteristics Studies were categorized based on the analyzed imaging modality. The majority of publications on AI in IPMN imaging (n = 15, 60%) analyzed CT imaging, one-third of the studies (n = 8, 32%) employed MRI imaging, while only a few (n = 2, 8%) were based on EUS imaging ( Figure 2 ). Download figure Open in new tab Figure 2: Published studies by year of publication and prediction target. The studies are further categorized by the imaging modality used. The radius of each circle is proportional to the number of publications, with colors indicating imaging modality. We scrutinized the publication trend over time, both collectively and for each prediction target. All included studies were conducted between 2018 and 2023. The number of publications increased over time, with more than two-thirds (n = 17, 68%) of all the studies published within the last three years (2021-2023), and the highest number of studies published in 2023 (n = 8, 32%, Figure 2 ). Based on the prediction target, approximately one-fourth of the included studies (n = 5, 20%) focused on detection, a couple of studies (n = 2, 8%) on segmentation, while the majority of the studies performed differential diagnosis (n = 9, 36%) and risk stratification (n = 10, 40%) of pancreatic cysts and IPMN ( Figure 3 ). Download figure Open in new tab Figure 3: Overview of research on AI in IPMN imaging. (A) Distribution of publications on AI in IPMN imaging based on prediction target (IPMN/cyst detection, cyst segmentation, differential diagnosis, risk stratification) and stage of clinical translation: internal validation (monocentric studies), external validation (multicentric studies), prospective clinical evaluation. Each dot represents a single publication, with the dot color indicating the imaging modality used. (B) Size of the total patient cohort and the patient cohort with IPMN. Colors indicate the imaging modality used, and symbol shapes indicate the number of centers involved in the study. One study evaluating segmentation did not document the number of IPMN patients in the underlying cohort. (C) Performance metrics - Accuracy (Acc), Area Under the Receiver Operating Curve (AUC), Sensitivity (Sen), and Specificity (Spec) - for the proposed model across prediction targets in each study. Colors represent the imaging modalities used. For segmentation, the DSC is presented. Not all studies reported every metric. Current Landscape and Clinical Translation of AI in IPMN Imaging Regarding the stage of clinical translation, over two-thirds of the studies (n = 17, 68%) conducted internal validation, 7 (28%) carried out external validation, and only one study (4%) proceeded to a prospective clinical evaluation ( Figure 3 A ). Nearly half of the studies (n = 11, 44%) were single-center studies, and one-third (n = 8, 32%) were multicenter studies, with most of these being two-center studies (n = 5, 20%). The remaining studies (n = 6, 24%) did not document the number of centers explicitly ( Figure 3 B , Table 1 ). View this table: View inline View popup Table 1: Summary and study characteristics of publications included in this systematic review The majority of the studies (n = 18, 72%) investigated cohorts of less than 250 patients. Three studies (12%) included over 1000 patients in total. Most of the studies had an IPMN patient population of less than 100 (n = 13, 52%), and ten studies (40%) within the range of 100-200, while only one study (4%) included over 200 patients with IPMN ( Figure 3 B ). Only one-third of the included studies (n = 8, 32%) mainly focused on IPMN, with the remaining studies (n = 17, 58%) including IPMN patients as a subset of total patients ( Supplementary Table 2 ). In the studies with a large overall patient cohort size, i.e., over 1000, only a small subset of patients, under 7%, had a pathologically confirmed IPMN ( Figure 3 B , Table 1 ). The diagnosis of IPMN or other pancreatic cysts was established through surgical pathology reports in most of the studies (n = 21, 84%), while in the other studies (n = 4, 16%) annotations provided by radiologists were considered as ground truth labels ( Table 1 ). The methodologies exhibit heterogeneity, yet a majority of studies used Convolutional Neural Networks (CNN) to develop the prediction model. About half of the included studies evaluate these models on independent datasets, and an almost equal number of studies present cross-validation results ( Table 1 ). Studies focused on IPMN/cyst detection predominantly employed the U-Net architecture for pancreas region of interest (ROI) segmentation. Alternatively, some studies used a sequential model setup with CNN-based feature extraction followed by machine learning classifiers for final detection. Reported sensitivity across these studies generally exceeded 80%. Variants of U-Net architecture optimized specifically for pancreatic cyst segmentation were commonly utilized for segmentation tasks. The reported dice similarity coefficient (DSC) for segmentation ranged from 0.42 to 0.80. For differential diagnosis, approximately half of the studies (n=4) extracted radiomic features and utilized machine learning classifiers, achieving diagnostic accuracies between 67 and 94%. The remaining studies (n=5) employed CNNs directly for image classification, with reported accuracies varying from 72 to 92%. For risk stratification, most studies implemented CNNs, with some incorporating image fusion techniques to integrate information from multiple MRI sequences. The average reported accuracy for these studies reached 85% ( Figure 3 C , Table 1 , Supplementary Table 2 ) Risk of bias analysis using the PROBAST tool [ 32 ] identified methodological shortcomings in most included works on AI in IPMN imaging ( Table 1 , Supplementary Table 3 ). Discussion Recent advances in AI for analyzing medical imaging and imaging reports have demonstrated capabilities to improve clinical tasks such as diagnosis, prognosis, therapeutic planning, and monitoring disease progression [ 59 ]. Similarly, such advanced computational methods hold promise in improving the clinical management of IPMN by enhancing diagnostic accuracy and treatment selection. We conducted a systematic review of existing publications on AI in IPMN imaging, with an emphasis on evaluating their patient and data characteristics, methodology, application areas, and stages of clinical translation. Our analysis indicates an increasing number of publications on AI-based IPMN diagnosis and risk stratification. Nonetheless, only a limited number of studies had a primary focus on IPMN [ 38 , 41 – 44 , 50 , 51 , 56 ]; instead, most works included smaller numbers of IPMN patients in addition to other patient cohorts, i.e., with other types of pancreatic cysts. A prevalent challenge identified across all the publications is limited data availability, as most studies trained the AI models on small single-center datasets and did not publish data alongside the scientific analysis, resulting in an overall scarcity of high-quality PDAC datasets [ 60 ] and a notable absence of publicly available IPMN datasets. This is in contrast with the availability of datasets for other types of cancer, for instance, breast cancer [ 61 , 62 ] and brain tumor [ 63 ] imaging. Moreover, the reporting quality appears inadequate as several publications do not document the exact number of centers involved, which compromises the validity and reproducibility of the study [ 64 ] as well as the adoption of new technology [ 65 ]. Our risk of bias analysis using the PROBAST tool [ 32 ] mirrored such methodological shortcomings in most included works on AI in IPMN imaging. Particularly frequent sources of bias were small cohort sizes and unclear patient selection criteria. The stratification of IPMN malignant progression based on imaging is challenging, with estimated diagnostic accuracies of imaging ranging from 60 to 80% in the scientific literature [ 66 ]. The established consensus guidelines for IPMN management and stratification recommend imaging, laboratory results, and physical examinations to stratify the risk of malignant progression [ 10 , 21 ]. However, the low specificity of existing methods [ 22 ] frequently leads to unnecessary surgery or surveillance of cystic lesions at high risk of malignancy. Computational analysis of pancreatic imaging can predict morphological types of IPMN noninvasively and bridge a crucial clinical gap by identifying patterns that could stratify and correlate with malignant progression in patients with IPMN. Most studies included in this systematic review proposing AI-based stratification of type of dysplasia based on imaging reported accuracies ranging from 60 to 99.6%, with most between 75 and 90% ( Table 1 , Supplementary Table 2 ). We found CT to be the most frequently used imaging type. However, per the American College of Radiology’s Appropriateness Criteria for pancreatic cysts, the preferred imaging modality for initial evaluation is an abdominal MRI without and with intravenous contrast with magnetic resonance cholangiopancreatography (MRCP). EUS is only recommended if worrisome features are observed, and the cyst size is greater than 2.5 cm [ 33 ]. The preference for CT in literature is likely because it is more cost-effective, expeditious, and widely available in nearly all medical facilities [ 67 ], whereas MRI and EUS are protracted, costly, and comparatively less available [ 68 ]. It is noteworthy that the majority of studies rely solely on (one type of) imaging data; however, integrating multimodal clinical data could further improve model performance and enhance patient selection for surgical resection. Methodologically, in deep learning-based studies, CNNs predominated in ROI segmentation and feature extraction, with UNet-based models, notably nnUNet [ 69 ], being the most frequently employed. In radiomics-based studies, which represented a relatively small proportion of studies investigating differential diagnosis, radiomic features (e.g., global histogram, first-order, second-order, wavelet features, etc.) are extracted and subsequently fed into classical machine learning algorithms (e.g., random forest and support vector machine) for prediction or classification. Withal, drawing definitive conclusions on the most effective AI methodology is challenging due to the dearth of large high-quality data and heterogeneity in prediction targets. While few studies directly compare different approaches on the same task or dataset, most publications report the performance of a single model on a closed- or restricted-access dataset, complicating reproducibility and comparison. Models for language processing, particularly LLMs like GPT-4 [ 70 ] and Med-PaLM 2 [ 71 ], are increasingly being used in the field of medicine, owing to their potential to enhance patient care [ 72 ]. Despite our search strategy being inclusive of NLP and imaging reports, no such study met the inclusion criteria. However, the few retrieved studies employing language-based methods centered on the identification of patients with pancreatic cysts using rule-based algorithms [ 73 – 75 ]. One study used the pre-trained BioBERT model for extracting information from radiology reports in the context of pancreatic cysts, albeit without fine-tuning [ 76 ]. Therefore, the potential of modern LLMs in IPMN care, including vision-language models, remains to be elucidated. Optimization methods like Retrieval-Augmented Generation, which allows integrating authoritative knowledge such as clinical treatment guidelines in addition to its training data sources before generating outputs, may yield additional benefits over generalist LLMs in medical scenarios [ 77 , 78 ]. Despite various available guidelines, a standard PDAC screening examination for the general population has yet to be established. For example, routine screening mammograms are the most sensitive way to screen for breast cancer, offering relatively affordable and accessible options for screening. Multiple AI-based breast cancer screening tools have received Food and Drug Administration (FDA) approval [ 79 ] According to our review, only one study initiated a prospective clinical evaluation [ 36 ], and one study [ 39 ] extended the application to develop an end-to-end system [ 80 ] for the diagnosis of pancreatic lesions, which highlights the paucity of clinical translation and the absence of FDA-approved AI tools for screening IPMN and other pancreatic cysts. The limitations of this systematic review are largely related to the reporting standards of the included publications. Based on the large variation in study designs, prediction targets ( Table 1 ), and reported performance metrics, an adequate summarization of model performances across all included studies was not feasible. We therefore report a complete list of metrics ( Table 1 , Supplementary Table 2 ). We also limited our assessment of the reporting quality of included studies to a general risk of bias assessment using the PROBAST [ 32 ] tool and a detailed review of the most relevant quality criteria for AI-based studies including details on the patient cohort, methodological aspects, and the validation procedure. The reason for this was that existing tools to assess the reporting quality of studies included in systematic reviews (i.e., QUADAS [ 81 ] for systematic reviews of diagnostic accuracy studies) are not suitable for the present type of studies. Therefore, we could not perform a formal assessment of the reporting quality of the included studies. In conclusion, this systematic review provides a comprehensive overview of the academic literature on AI in IPMN imaging and highlights the potential of computational image analysis models for the complex management of IPMN. Most existing studies rely on small, single-center patient cohorts and imaging data only. To exploit the full potential of AI for improving diagnostic accuracy and patient outcomes, it will be necessary to incorporate multicentric and multimodal clinical data and undertake robust efforts to translate retrospective research into clinical practice. Data Availability All data produced in the present work are contained in the manuscript. Author Contributions Conceptualization: MIQ, JAB, MTYS, FRK; Methodology: MIQ, JAB, MTYS, FRK; Software: MIQ, JAB, FRK; Validation: MIQ, JAB, FRK; Formal analysis: MIQ, JAB, FRK; Investigation: MIQ, JAB, MTYS, DS, TTTT, CMS, FRK; Resources: MTYS, CMS, FRK; Data Curation: MIQ, JAB, FRK; Writing - Original Draft: MIQ, FRK; Writing - Review & Editing: MIQ, JAB, MTYS, DS, TTTT, CMS, FRK; Visualization: MIQ, FRK; Supervision: MTYS, CMS, FRK; Project administration: MTYS, TTTT, CMS, FRK; Funding acquisition: MTYS, CMS, FRK. Competing Interests FRK declares unpaid advisory roles for Perspectum, Inc., Oxford, UK; Radical Healthcare, Inc., San Francisco, CA; and the Surgical Data Science Collective (SDSC), Washington, DC. CMS declares an unpaid advisory role for Perspectum, Inc., Oxford, UK. All other authors declare no competing interests. Supplementary Material View this table: View inline View popup Download powerpoint Supplementary Table 1: Data extraction template The table presents a structured data extraction template used to extract the data from the included studies on AI in IPMN imaging. View this table: View inline View popup Supplementary Table 2: Data extracted from the included publications on AI in IPMN imaging Abbreviations used in the table: MPD: Main Pancreatic Duct; MCN: Mucinous Cystic Neoplasm; SCA: Serous Cystadenoma; SPN: Solid Pseudopapillary Neoplasm; SCN: Serous Cystic Neoplasm; SPT: Solid Pseudopapillary Tumor; MCA: Mucinous Cystadenoma; EUS-FNA: Endoscopic Ultrasound Fine Needle Aspiration; ICD: International Classification of Diseases; ND: Not Documented; CNN: Convolutional Neural Network; SVM: Support Vector Machine; GAN: Generative Adversarial Network; RF: Random Forest; ROI: Region of Interest; ViT: Vision Transformer; ANN: Artificial Neural Network; ML: Machine Learning; NB: Naive Bayes; KNN: K-Nearest Neighbor; LDA: Linear Discriminant Analysis; DL: Deep Learning, MRCP: Magnetic Resonance Cholangiopancreatography. Performance Metrics: DLM: Deep Learning Model; RM: Radiomics Model; Sen: Sensitivity; Spec: Specificity; Acc: Accuracy; Pre: Precision; Rec: Recall; AUC/AUROC: Area Under Receiver Operating Characteristic Curve; AUPR: Area Under Precision-Recall Curve; FPR: False Positive Rate; DSC: Dice Similarity Coefficient/Dice Score. View this table: View inline View popup Supplementary Table 3: Risk of bias assessment using PROBAST Tool Download figure Open in new tab Supplementary Figure 1: Performance Metrics Across Prediction Targets For detection, differential diagnosis, and risk stratification accuracy (Acc), area under the receiver operating curve (AUC), sensitivity (Sen), and specificity (Spec) are presented. For segmentation, the DSC is presented. — indicates that the metric was not reported. PRISMA 2020 Checklist View this table: View inline View popup Acknowledgments and Funding The authors gratefully acknowledge assistance from Jing Lu (Purdue University, Libraries and School of Information Studies, West Lafayette, IN) in developing the search strategy for this systematic review and from Sarwesh Rauniyar (Indiana University School of Medicine, Indianapolis, IN) in developing the data extraction protocol. We thank Michaela Unger (Else Kröner Fresenius Center for Digital Health, Dresden, Germany) for providing a visualization code template for Figure 2 . CMS and FRK receive support from the Indiana Clinical and Translational Sciences Institute funded, in part, by Grant Number UM1TR004402 from the National Institutes of Health, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award. Furthermore, FRK receives support from the German Cancer Research Center (CoBot 2.0) and the Joachim Herz Foundation (Add-On Fellowship for Interdisciplinary Life Science). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. References [1]. ↵ Siegel RL , Giaquinto AN , Jemal A. Cancer statistics, 2024 . CA Cancer J Clin 2024 ; 74 : 12 – 49 . OpenUrl CrossRef PubMed [2]. ↵ Quaresma M , Coleman MP , Rachet B. 40-year trends in an index of survival for all cancers combined and survival adjusted for age and sex for each cancer in England and Wales, 1971-2011: a population-based study . Lancet 2015 ; 385 : 1206 – 18 . OpenUrl CrossRef PubMed [3]. ↵ Sarantis P , Koustas E , Papadimitropoulou A , Papavassiliou AG , Karamouzis MV . Pancreatic ductal adenocarcinoma: Treatment hurdles, tumor microenvironment and immunotherapy . World J Gastrointest Oncol 2020 ; 12 : 173 – 81 . OpenUrl CrossRef PubMed [4]. ↵ Schawkat K , Manning MA , Glickman JN , Mortele KJ . Pancreatic Ductal Adenocarcinoma and Its Variants: Pearls and Perils . Radiographics 2020 ; 40 : 1219 – 39 . OpenUrl PubMed [5]. ↵ Chang YR , Park JK , Jang J-Y , Kwon W , Yoon JH , Kim S-W . Incidental pancreatic cystic neoplasms in an asymptomatic healthy population of 21,745 individuals: Large-scale, single-center cohort study . Medicine 2016 ; 95 : e5535 . OpenUrl PubMed [6]. ↵ Nassour I , Choti MA . Types of Pancreatic Cysts . JAMA 2016 ; 316 : 1226 . OpenUrl PubMed [7]. ↵ de la Fuente J , Chatterjee A , Lui J , Nehra AK , Bell MG , Lennon RJ , et al. Long-Term Outcomes and Risk of Pancreatic Cancer in Intraductal Papillary Mucinous Neoplasms . JAMA Netw Open 2023 ; 6 : e2337799 . OpenUrl [8]. ↵ Gentiluomo M , Corradi C , Arcidiacono PG , Crippa S , Falconi M , Belfiori G , et al. Role of pancreatic ductal adenocarcinoma risk factors in intraductal papillary mucinous neoplasm progression . Front Oncol 2023 ; 13 : 1172606 . [9]. Capurso G , Crippa S , Vanella G , Traini M , Zerboni G , Zaccari P , et al. Factors Associated With the Risk of Progression of Low-Risk Branch-Duct Intraductal Papillary Mucinous Neoplasms . JAMA Netw Open 2020 ; 3 : e2022933 . OpenUrl PubMed [10]. ↵ Tanaka M , Fernández-Del Castillo C , Kamisawa T , Jang JY , Levy P , Ohtsuka T , et al. Revisions of international consensus Fukuoka guidelines for the management of IPMN of the pancreas . Pancreatology 2017 ; 17 : 738 – 53 . OpenUrl CrossRef PubMed [11]. Del Chiaro M , Ateeb Z , Hansson MR , Rangelova E , Segersvärd R , Kartalis N , et al. Survival Analysis and Risk for Progression of Intraductal Papillary Mucinous Neoplasia of the Pancreas (IPMN) Under Surveillance: A Single-Institution Experience . Ann Surg Oncol 2017 ; 24 : 1120 – 6 . OpenUrl PubMed [12]. ↵ Marchegiani G , Mino-Kenudson M , Sahora K , Morales-Oyarvide V , Thayer S , Ferrone C , et al. IPMN involving the main pancreatic duct: biology, epidemiology, and long-term outcomes following resection . Ann Surg 2015 ; 261 : 976 – 83 . OpenUrl CrossRef PubMed [13]. ↵ Oyama H , Tada M , Takagi K , Tateishi K , Hamada T , Nakai Y , et al. Long-term risk of malignancy in branch-duct intraductal papillary mucinous neoplasms . Gastroenterology 2020 ; 158 : 226 – 37 .e5. OpenUrl CrossRef PubMed [14]. ↵ Felsenstein M , Noë M , Masica DL , Hosoda W , Chianchiano P , Fischer CG , et al. IPMNs with co-occurring invasive cancers: neighbours but not always relatives . Gut 2018 ; 67 : 1652 – 62 . OpenUrl Abstract / FREE Full Text [15]. ↵ Lewis R , Drebin JA , Callery MP , Fraker D , Kent TS , Gates J , et al. A contemporary analysis of survival for resected pancreatic ductal adenocarcinoma . HPB 2013 ; 15 : 49 – 60 . OpenUrl PubMed [16]. ↵ Agostini A , Piro G , Inzani F , Quero G , Esposito A , Caggiano A , et al. Identification of spatially-resolved markers of malignant transformation in Intraductal Papillary Mucinous Neoplasms . Nat Commun 2024 ; 15 : 2764 . OpenUrl CrossRef PubMed [17]. ↵ Aoki S , Miyata H , Konno H , Gotoh M , Motoi F , Kumamaru H , et al. Risk factors of serious postoperative complications after pancreaticoduodenectomy and risk calculators for predicting postoperative complications: a nationwide study of 17,564 patients in Japan . J Hepatobiliary Pancreat Sci 2017 ; 24 : 243 – 51 . OpenUrl CrossRef PubMed [18]. ↵ Reddy R , Chari ST . IPMN: not a new kid on the block anymore . Gastroenterology 2004 ; 127 : 1853 – 5 . OpenUrl PubMed [19]. ↵ Kwon JH , Kim SC , Shim IK , Song KB , Lee JH , Hwang DW , et al. Factors Affecting the Development of Diabetes Mellitus After Pancreatic Resection . Pancreas 2015 ; 44 : 1296 – 303 . OpenUrl PubMed [20]. ↵ Henn J , Wyzlic PK , Esposito I , Semaan A , Branchi V , Klinger C , et al. Surgical treatment for pancreatic cystic lesions-implications from the multi-center and prospective German StuDoQ|Pancreas registry . Langenbecks Arch Surg 2023 ; 408 : 28 . [21]. ↵ Ohtsuka T , Fernandez-del Castillo C , Furukawa T , Hijioka S , Jang J-Y , Lennon AM , et al. International evidence-based Kyoto guidelines for the management of intraductal papillary mucinous neoplasm of the pancreas . Pancreatology 2023 . doi: 10.1016/j.pan.2023.12.009 . OpenUrl CrossRef [22]. ↵ Heckler M , Michalski CW , Schaefle S , Kaiser J , Büchler MW , Hackert T . The Sendai and Fukuoka consensus criteria for the management of branch duct IPMN - A meta-analysis on their accuracy . Pancreatology 2017 ; 17 : 255 – 62 . OpenUrl CrossRef PubMed [23]. ↵ Crippa S , Pergolini I , Rubini C , Castelli P , Partelli S , Zardini C , et al. Risk of misdiagnosis and overtreatment in patients with main pancreatic duct dilatation and suspected combined/main-duct intraductal papillary mucinous neoplasms . Surgery 2016 ; 159 : 1041 – 9 . OpenUrl PubMed [24]. ↵ Khoury RE , Kabir C , Maker VK , Banulescu M , Wasserman M , Maker AV . What is the Incidence of Malignancy in Resected Intraductal Papillary Mucinous Neoplasms? An Analysis of Over 100 US Institutions in a Single Year . Ann Surg Oncol 2018 ; 25 : 1746 – 51 . OpenUrl CrossRef PubMed [25]. ↵ Hol L , Bruno MJ , Cahen DL . Follow-up of asymptomatic pancreatic cysts in clinical practice: A vignette questionnaire . Pancreatology 2016 ; 16 : 416 – 22 . OpenUrl PubMed [26]. ↵ Obermeyer Z , Emanuel EJ . Predicting the Future - Big Data, Machine Learning, and Clinical Medicine . N Engl J Med 2016 ; 375 : 1216 – 9 . OpenUrl CrossRef PubMed [27]. ↵ Aggarwal R , Sounderajah V , Martin G , Ting DSW , Karthikesalingam A , King D , et al. Diagnostic accuracy of deep learning in medical imaging: a systematic review and meta-analysis . NPJ Digit Med 2021 ; 4 : 65 . [28]. ↵ Yang X , Chen A , PourNejatian N , Shin HC , Smith KE , Parisien C , et al. A large language model for electronic health records . NPJ Digit Med 2022 ; 5 : 194 . [29]. ↵ Clusmann J , Kolbinger FR , Muti HS , Carrero ZI , Eckardt J-N , Laleh NG , et al. The future landscape of large language models in medicine . Commun Med 2023 ; 3 : 141 . [30]. ↵ Page MJ , McKenzie JE , Bossuyt PM , Boutron I , Hoffmann TC , Mulrow CD , et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews . J Clin Epidemiol 2021 ; 134 : 178 – 89 . OpenUrl CrossRef PubMed [31]. ↵ Shea BJ , Reeves BC , Wells G , Thuku M , Hamel C , Moran J , et al. AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both . BMJ 2017 ; 358 : j4008 . OpenUrl FREE Full Text [32]. ↵ Wolff RF , Moons KGM , Riley RD , Whiting PF , Westwood M , Collins GS , et al. PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies . Ann Intern Med 2019 ; 170 : 51 – 8 . OpenUrl CrossRef PubMed [33]. ↵ Expert Panel on Gastrointestinal Imaging , Fábrega-Foster K , Kamel IR , Horowitz JM , Arif-Tiwari H , Bashir MR , et al. ACR appropriateness criteria® pancreatic cyst . J Am Coll Radiol 2020 ; 17 : S198 – 206 . OpenUrl PubMed [34]. Abel L , Wasserthal J , Weikert T , Sauter AW , Nesic I , Obradovic M , et al. Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning . Diagnostics (Basel ) 2021 ; 11 . doi: 10.3390/diagnostics11050901 . OpenUrl CrossRef [35]. Abi Nader C , Vetil R , Wood LK , Rohe M-M , Bône A , Karteszi H , et al. Automatic Detection of Pancreatic Lesions and Main Pancreatic Duct Dilatation on Portal Venous CT Scans Using Deep Learning . Invest Radiol 2023 ; 58 : 791 – 8 . OpenUrl PubMed [36]. ↵ Cao K , Xia Y , Yao J , Han X , Lambert L , Zhang T , et al. Large-scale pancreatic cancer detection via non-contrast CT and deep learning . Nat Med 2023 ; 29 : 3033 – 43 . OpenUrl PubMed [37]. Chu LC , Park S , Soleimani S , Fouladi DF , Shayesteh S , He J , et al. Classification of pancreatic cystic neoplasms using radiomic feature analysis is equivalent to an experienced academic radiologist: a step toward computer-augmented diagnostics for radiologists . Abdom Radiol (NY ) 2022 ; 47 : 4139 – 50 . OpenUrl PubMed [38]. ↵ Corral JE , Hussein S , Kandel P , Bolan CW , Bagci U , Wallace MB . Deep Learning to Classify Intraductal Papillary Mucinous Neoplasms Using Magnetic Resonance Imaging . Pancreas 2019 ; 48 : 805 – 10 . OpenUrl PubMed [39]. ↵ Dmitriev K , Marino J , Baker K , Kaufman AE . Visual Analytics of a Computer-Aided Diagnosis System for Pancreatic Lesions . IEEE Trans Vis Comput Graph 2021 ; 27 : 2174 – 85 . OpenUrl PubMed [40]. Gao X , Wang X . Performance of deep learning for differentiating pancreatic diseases on contrast-enhanced magnetic resonance imaging: A preliminary study . Diagn Interv Imaging 2020 ; 101 : 91 – 100 . OpenUrl PubMed [41]. ↵ Hussein S , Kandel P , Corral JE , Bolan CW , Wallace MB , Bagci U. Deep multi-modal classification of intraductal papillary mucinous neoplasms (IPMN) with canonical correlation analysis . 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018), IEEE ; 2018 , p. 800 – 4 . [42]. Hussein S , Kandel P , Bolan CW , Wallace MB , Bagci U . Lung and Pancreatic Tumor Characterization in the Deep Learning Era: Novel Supervised and Unsupervised Learning Approaches . IEEE Trans Med Imaging 2019 ; 38 : 1777 – 87 . OpenUrl PubMed [43]. Kuwahara T , Hara K , Mizuno N , Okuno N , Matsumoto S , Obata M , et al. Usefulness of Deep Learning Analysis for the Diagnosis of Malignancy in Intraductal Papillary Mucinous Neoplasms of the Pancreas . Clin Transl Gastroenterol 2019 ; 10 : 1 – 8 . OpenUrl [44]. ↵ LaLonde R , Tanner I , Nikiforaki K , Papadakis GZ , Kandel P , Bolan CW , et al. INN: Inflated Neural Networks for IPMN Diagnosis . Medical Image Computing and Computer Assisted Intervention – MICCAI 2019 , Springer International Publishing ; 2019 , p. 101 – 9 . [45]. Li H , Reichert M , Lin K , Tselousov N , Braren R , Fu D , et al. Differential Diagnosis for Pancreatic Cysts in CT Scans Using Densely-Connected Convolutional Networks . 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), IEEE ; 2019 , p. 2095 – 8 . [46]. Liang W , Tian W , Wang Y , Wang P , Wang Y , Zhang H , et al. Classification prediction of pancreatic cystic neoplasms based on radiomics deep learning models . BMC Cancer 2022 ; 22 : 1237 . OpenUrl PubMed [47]. Mazor N , Dar G , Lederman R , Lev-Cohain N , Sosna J , Joskowicz L . MC3DU-Net: a multisequence cascaded pipeline for the detection and segmentation of pancreatic cysts in MRI . Int J Comput Assist Radiol Surg 2024 ; 19 : 423 – 32 . OpenUrl PubMed [48]. Park HJ , Shin K , You M-W , Kyung S-G , Kim SY , Park SH , et al. Deep Learning-based Detection of Solid and Cystic Pancreatic Neoplasms at Contrast-enhanced CT . Radiology 2023 ; 306 : 140 – 9 . OpenUrl PubMed [49]. Qu W , Yang J , Li J , Yuan G , Li S , Chu Q , et al. Avoid non-diagnostic EUS-FNA: a DNN model as a possible gatekeeper to distinguish pancreatic lesions prone to inconclusive biopsy . Br J Radiol 2023 ; 96 : 20221112 . [50]. ↵ Proietto Salanitri F , Bellitto G , Palazzo S , Irmakci I , Wallace M , Bolan C , et al. Neural Transformers for Intraductal Papillary Mucosal Neoplasms (IPMN) Classification in MRI images . 2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), IEEE ; 2022 , p. 475 – 9 . [51]. ↵ Schulz D , Heilmaier M , Phillip V , Treiber M , Mayr U , Lahmer T , et al. Accurate prediction of histological grading of intraductal papillary mucinous neoplasia using deep learning . Endoscopy 2023 ; 55 : 415 – 22 . OpenUrl PubMed [52]. Shen X , Yang F , Yang P , Yang M , Xu L , Zhuo J , et al. A Contrast-Enhanced Computed Tomography Based Radiomics Approach for Preoperative Differentiation of Pancreatic Cystic Neoplasm Subtypes: A Feasibility Study . Front Oncol 2020 ; 10 : 248 . [53]. Si K , Xue Y , Yu X , Zhu X , Li Q , Gong W , et al. Fully end-to-end deep-learning-based diagnosis of pancreatic tumors . Theranostics 2021 ; 11 : 1982 – 90 . OpenUrl CrossRef PubMed [54]. Wang X , Sun Z , Xue H , Qu T , Cheng S , Li J , et al. A deep learning algorithm to improve readers’ interpretation and speed of pancreatic cystic lesions on dual-phase enhanced CT . Abdominal Radiology 2022 ; 47 : 2135 – 47 . OpenUrl PubMed [55]. Watson MD , Lyman WB , Passeri MJ , Murphy KJ , Sarantou JP , Iannitti DA , et al. Use of Artificial Intelligence Deep Learning to Determine the Malignant Potential of Pancreatic Cystic Neoplasms With Preoperative Computed Tomography Imaging . Am Surg 2021 ; 87 : 602 – 7 . OpenUrl PubMed [56]. ↵ Yao L , Zhang Z , Demir U , Keles E , Vendrami C , Agarunov E , et al. Radiomics Boosts Deep Learning Model for IPMN Classification. Machine Learning in Medical Imaging , Springer Nature Switzerland ; 2024 , p. 134 – 43 . [57]. Yuan M , Xia Y , Dong H , Chen Z , Yao J , Qiu M , et al. Devil is in the Queries: Advancing Mask Transformers for Real-world Medical Image Segmentation and Out-of-Distribution Localization . 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE ; 2023 , p. 23879 – 89 . [58]. Zhang J , Mao Y , Li J , Li Y , Luo J . A metric learning-based method using graph neural network for pancreatic cystic neoplasm classification from CTs . Med Phys 2022 ; 49 : 5523 – 36 . OpenUrl PubMed [59]. ↵ Rajpurkar P , Chen E , Banerjee O , Topol EJ . AI in health and medicine . Nat Med 2022 ; 28 : 31 – 8 . OpenUrl CrossRef PubMed [60]. ↵ Suman G , Patra A , Korfiatis P , Majumder S , Chari ST , Truty MJ , et al. Quality gaps in public pancreas imaging datasets: Implications & challenges for AI applications . Pancreatology 2021 ; 21 : 1001 – 8 . OpenUrl PubMed [61]. ↵ Halling-Brown MD , Warren LM , Ward D , Lewis E , Mackenzie A , Wallis MG , et al. OPTIMAM Mammography Image Database: A Large-Scale Resource of Mammography Images and Clinical Data . Radiol Artif Intell 2021 ; 3 : e200103 . OpenUrl [62]. ↵ Jeong JJ , Vey BL , Bhimireddy A , Kim T , Santos T , Correa R , et al. The EMory BrEast imaging Dataset (EMBED): A Racially Diverse, Granular Dataset of 3.4 Million Screening and Diagnostic Mammographic Images . Radiol Artif Intell 2023 ; 5 : e220047 . OpenUrl PubMed [63]. ↵ RSNA-ASNR-MICCAI-BRATS-2021 - the cancer imaging archive (TCIA) . The Cancer Imaging Archive (TCIA) 2023 . https://www.cancerimagingarchive.net/analysis-result/rsna-asnr-miccai-brats-2021/ (accessed July 15, 2024 ). [64]. ↵ Kolbinger FR , Veldhuizen GP , Zhu J , Truhn D , Kather JN . Reporting guidelines in medical artificial intelligence: a systematic review and meta-analysis . Commun Med 2024 ; 4 : 71 . [65]. ↵ Jayakumar S , Sounderajah V , Normahani P , Harling L , Markar SR , Ashrafian H , et al. Quality assessment standards in artificial intelligence diagnostic accuracy systematic reviews: a meta-research study . NPJ Digit Med 2022 ; 5 : 11 . [66]. ↵ Udare A , Agarwal M , Alabousi M , McInnes M , Rubino JG , Marcaccio M , et al. Diagnostic Accuracy of MRI for Differentiation of Benign and Malignant Pancreatic Cystic Lesions Compared to CT and Endoscopic Ultrasound: Systematic Review and Meta-analysis . J Magn Reson Imaging 2021 ; 54 : 1126 – 37 . OpenUrl PubMed [67]. ↵ Power SP , Moloney F , Twomey M , James K , O’Connor OJ , Maher MM . Computed tomography and patient risk: Facts, perceptions and uncertainties . World J Radiol 2016 ; 8 : 902 – 15 . OpenUrl CrossRef PubMed [68]. ↵ Costache MI , Costache CA , Dumitrescu CI , Tica AA , Popescu M , Baluta EA , et al. Which is the Best Imaging Method in Pancreatic Adenocarcinoma Diagnosis and Staging - CT, MRI or EUS? Curr Health Sci J 2017 ; 43 : 132 – 6 . OpenUrl PubMed [69]. ↵ Isensee F , Jaeger PF , Kohl SAA , Petersen J , Maier-Hein KH . nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation . Nat Methods 2021 ; 18 : 203 – 11 . OpenUrl CrossRef PubMed [70]. ↵ OpenAI , Achiam J , Adler S , Agarwal S , Ahmad L , Akkaya I , et al. GPT-4 Technical Report . arXiv [csCL] 2023 . [71]. ↵ Singhal K , Tu T , Gottweis J , Sayres R , Wulczyn E , Hou L , et al. Towards Expert-Level Medical Question Answering with Large Language Models . arXiv [csCL] 2023 . [72]. ↵ Waisberg E , Ong J , Masalkhi M , Kamran SA , Zaman N , Sarker P , et al. GPT-4: a new era of artificial intelligence in medicine . Ir J Med Sci 2023 ; 192 : 3197 – 200 . OpenUrl PubMed [73]. ↵ Mehrabi S , Schmidt CM , Waters JA , Beesley C , Krishnan A , Kesterson J , et al. An efficient pancreatic cyst identification methodology using natural language processing . Stud Health Technol Inform 2013 ; 192 : 822 – 6 . OpenUrl PubMed [74]. Roch AM , Mehrabi S , Krishnan A , Schmidt HE , Kesterson J , Beesley C , et al. Automated pancreatic cyst screening using natural language processing: a new tool in the early detection of pancreatic cancer . HPB 2015 ; 17 : 447 – 53 . OpenUrl PubMed [75]. ↵ Kooragayala K , Crudeli C , Kalola A , Bhat V , Lou J , Sensenig R , et al. Utilization of Natural Language Processing Software to Identify Worrisome Pancreatic Lesions . Ann Surg Oncol 2022 ; 29 : 8513 – 9 . OpenUrl PubMed [76]. ↵ Yamashita R , Bird K , Cheung PY-C , Decker JH , Flory MN , Goff D , et al. Automated Identification and Measurement Extraction of Pancreatic Cystic Lesions from Free-Text Radiology Reports Using Natural Language Processing . Radiol Artif Intell 2022 ; 4 : e210092 . OpenUrl PubMed [77]. ↵ Unlu Ozan , Shin Jiyeon , Mailly Charlotte J. , Oates Michael F. , Tucci Michela R. , Varugheese Matthew , et al. Retrieval-Augmented Generation–Enabled GPT-4 for Clinical Trial Screening . NEJM AI 2024 ; 1 : AIoa2400181 . OpenUrl [78]. ↵ Kresevic S , Giuffrè M , Ajcevic M , Accardo A , Crocè LS , Shung DL . Optimization of hepatological clinical guidelines interpretation by large language models: a retrieval augmented generation-based framework . NPJ Digit Med 2024 ; 7 : 102 . [79]. ↵ Potnis KC , Ross JS , Aneja S , Gross CP , Richman IB . Artificial Intelligence in Breast Cancer Screening: Evaluation of FDA Device Regulation and Future Recommendations . JAMA Intern Med 2022 ; 182 : 1306 – 12 . OpenUrl PubMed [80]. ↵ Jadhav S , Dmitriev K , Marino J , Barish M , Kaufman AE . 3D Virtual Pancreatography . IEEE Trans Vis Comput Graph 2022 ; 28 : 1457 – 68 . OpenUrl PubMed [81]. ↵ Whiting P , Rutjes AWS , Reitsma JB , Bossuyt PMM , Kleijnen J . The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews . BMC Med Res Methodol 2003 ; 3 : 25 . View the discussion thread. Back to top Previous Next Posted January 09, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Artificial Intelligence in Pancreatic Intraductal Papillary Mucinous Neoplasm Imaging: A Systematic Review Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Artificial Intelligence in Pancreatic Intraductal Papillary Mucinous Neoplasm Imaging: A Systematic Review Muhammad Ibtsaam Qadir , Jackson A. Baril , Michele T. Yip-Schneider , Duane Schonlau , Thi Thanh Thoa Tran , C. Max Schmidt , Fiona R. Kolbinger medRxiv 2025.01.08.25320130; doi: https://doi.org/10.1101/2025.01.08.25320130 Share This Article: Copy Citation Tools Artificial Intelligence in Pancreatic Intraductal Papillary Mucinous Neoplasm Imaging: A Systematic Review Muhammad Ibtsaam Qadir , Jackson A. Baril , Michele T. Yip-Schneider , Duane Schonlau , Thi Thanh Thoa Tran , C. Max Schmidt , Fiona R. Kolbinger medRxiv 2025.01.08.25320130; doi: https://doi.org/10.1101/2025.01.08.25320130 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Radiology and Imaging Subject Areas All Articles Addiction Medicine (568) Allergy and Immunology (863) Anesthesia (300) Cardiovascular Medicine (4435) Dentistry and Oral Medicine (444) Dermatology (382) Emergency Medicine (608) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1509) Epidemiology (15227) Forensic Medicine (30) Gastroenterology (1124) Genetic and Genomic Medicine (6597) Geriatric Medicine (668) Health Economics (997) Health Informatics (4534) Health Policy (1368) Health Systems and Quality Improvement (1613) Hematology (540) HIV/AIDS (1264) Infectious Diseases (except HIV/AIDS) (15916) Intensive Care and Critical Care Medicine (1103) Medical Education (623) Medical Ethics (146) Nephrology (667) Neurology (6599) Nursing (346) Nutrition (998) Obstetrics and Gynecology (1144) Occupational and Environmental Health (957) Oncology (3332) Ophthalmology (974) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (663) Pediatrics (1693) Pharmacology and Therapeutics (691) Primary Care Research (711) Psychiatry and Clinical Psychology (5447) Public and Global Health (9230) Radiology and Imaging (2198) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1196) Rheumatology (593) Sexual and Reproductive Health (712) Sports Medicine (530) Surgery (712) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a00421ee09fa1640',t:'MTc3OTU0MDA2Mg=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();