The DOI matches its Crossref record; 108 of the 118 references it cites were found in a registry or an index; 11 of the 21 sources its claims cite were checked for support.
Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma
Open accessAbstract
Review of AI applications to pancreatic cystic lesions and adenocarcinoma.
Overview
CoChat Summary
## TL;DR
This review surveys published AI/ML work on pancreatic cancer screening, pancreatic cystic lesion (PCL) risk stratification, treatment response, and prognosis, emphasizing studies from 2013–2023. Overall, the paper concludes that AI performs best for PCL classification and risk stratification, while screening models for predicting who will develop PDAC are less reliable.
## Key findings
- In screening-oriented models, performance was modest: the review notes that existing AI models for the general population had AUCs “hovering around 0.60,” while a primary-care logistic regression model achieved AUCs of 65.6% for patients under 60 and 60.9% for older patients, and was estimated to enable earlier diagnosis for around 60% of pancreatic tumors.
- For PCL detection/risk stratification, several models performed strongly: a radiomics random forest model in 214 resected cyst patients reached AUC 0.940 for distinguishing five cystic neoplasm types, compared with AUC 0.895 for academic radiologists; a fused radiomics–DL model reached AUC 0.92 for serous cystadenoma diagnosis and 0.97 for differentiating MCNs from IPMNs.
- EUS-based AI also showed high accuracy: a CNN using 28 EUS cysts achieved AUC 1.0 for identifying mucinous cysts, with overall accuracy 98.5%, sensitivity 98.3%, and specificity 98.9%; another CNN using 43 patients with histologically proven IPMN classified advanced neoplasia with 99.6% accuracy, exceeding guideline accuracies of 51.8%–70.3% in that cohort.
- For prognostication and treatment-related prediction, performance was promising but more variable: a random forest model predicted 2-year overall survival with AUC 0.76 and 1-year recurrence-free survival with AUC 0.74; another model reported AUC 90% with 87% sensitivity and 80% specificity for above- vs below-median survival.
## Methods
- The paper is a narrative review of primary literature on novel AI algorithms for PDAC and precancerous pancreatic lesions.
- The authors searched PubMed and Embase for combinations of “artificial intelligence,” “machine learning,” “radiomics,” or “deep learning” with “PDAC,” “pancreatic cystic lesions,” or “endoscopic ultrasound.”
- Included studies were published from 2013–2023 and had to report accuracy-type metrics such as AUC, accuracy, sensitivity, or specificity; case reports, protocols, meta-analyses, and review articles were excluded.
## Limitations
- The paper states that many imaging-based AI studies are retrospective, single-center, and not externally validated, which limits generalizability and reproducibility.
- It also states that models may be biased toward higher-risk patients because histopathology from surgical resection is the reference standard, so included cohorts often consist of people who already required surgery.
- The text makes evident that several cited studies had small sample sizes and manual preprocessing requirements; these are stated in the paper as limitations that reduce scalability and may contribute to overfitting.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures


Verification
Claims the paper makes
8 Valid · 0 Not valid · 7 Not checked · 20 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“While the goal of surgery is to resect lesions with advanced neoplasia (high-grade dysplasia and/or adenocarcinoma), approximately 42–63% of resected IPMNs were found to have only low-grade dysplasia [12,13,14,15,16,17].”
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 1. Introduction > 1.1. Challenges in Predicting Progression to PDAC
What the sources say
- Supportedcites [12] Valsangkar, 2012, 851 resected cystic tumors of the pancreas: A 33-year experience at the Massachusetts General Hospital, Surgery source
The Discussion reports that 41% of the 326 resected IPMNs were adenomas with low-grade dysplasia, approximately the lower end of the cited 42–63% range. This paper supports the low-end estimate; the sentence cites multiple studies for the range. (LLM, conf=0.88)
- Could not verifycites [13] Gaujoux, 2011, Cystic lesions of the pancreas: Changes in the presentation and management of 1,424 patients at a single institution over a 15-year time period, J. Am. Coll. Surg. sourceExperimental — accuracy not yet measured
The Results section reports 114 initially resected IPMNs and discusses non-invasive IPMNs, but it does not give the proportion with low-grade dysplasia. Searches for “low-grade dysplasia” and the 42–63% range found no passage in our copy establishing that figure; the cited Table 3 figures were not available in the returned passages (page unavailable). (LLM, conf=0.68)
- Partly supportedcites [14] Scheiman, 2015, American gastroenterological association technical review on the diagnosis and management of asymptomatic neoplastic pancreatic cysts, Gastroenterology sourceExperimental — accuracy not yet measured
In the 'Interval Growth in Pancreatic Cysts and Risk of Adenocarcinoma' section, the technical review pools 99 studies of 9,249 surgically resected IPMNs and reports an overall high-risk (HGD/carcinoma in situ) rate of 42% (95% CI, 39%-45%), concluding that 58% (95% CI, 55%-61%) had low-risk (eg, low-grade dysplasia) or benign lesions. So this source supplies a figure inside the cited 42-63% band (58%), but neither endpoint: its 42% is the high-risk rate, not a low-grade-dysplasia rate, and it never reports 63%. It also groups benign lesions with low-grade dysplasia and explicitly cautions that the 58% was inferred (studies reporting only the high-risk group were assumed to be low risk otherwise), so the claim's 'only low-grade dysplasia' framing overstates it. (LLM, conf=0.75)
- Not checked — no full text of the sourcecites [15] Sahora, 2013, 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, Ann. Surg. source
- Not checked — no full text of the sourcecites [16] Sharib, 2018, Surgical overtreatment of pancreatic intraductal papillary mucinous neoplasms: Do the 2017 International Consensus Guidelines improve clinical decision making?, Surgery source
- Supportedcites [17] Marchegiani, 2021, Progression vs cyst stability of branch-duct intraductal papillary mucinous neoplasms after observation and surgery, JAMA Surg. source
Our copy reports a resected cohort of 292 pathologically proven BD IPMNs after ≥12 months of surveillance in which 107 (36.6%) harboured HGD or invasive cancer and the remainder — 185 patients, 63.4% — had low-grade dysplasia only (Results, Table 4: 'LGD (n = 185)' vs 'HGD (n = 63)'; Discussion: 'of the 185 patients (63.4%) ... none harbored a cyst with either HGD or invasive cancer at final pathological examination'). That value is the upper end of the 42–63% range the citing sentence attributes to its cluster of references, so this source supports the share it contributes; the only narrowing detail is that its denominator is branch-duct (and mixed) IPMNs resected after documented preoperative surveillance, not all resected IPMNs. (LLM, conf=0.85)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Hanania et al. reported similarly promising findings, with a logistic regression model achieving an AUC of 0.96 (sensitivity 97% and specificity 88%).”
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 3. Results > 3.2. Detection and Risk Stratification of Pancreatic Cystic Lesions
What the sources say
- Supportedcites [54] Hanania, 2016, Quantitative imaging to evaluate malignant potential of IPMNs, Oncotarget source
The Results report that a cross-validated logistic-regression model distinguished pathologically confirmed high-grade from low-grade IPMNs with an AUC of 0.96, sensitivity of 97%, and specificity of 88%. Those are the figures attributed to Hanania et al.; the study describes this as an assessment of malignant potential. (LLM, conf=0.94)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“While the rates of malignancy in IPMNs is well characterized (ranging from 57 to 92% for MD-IPMN and 6 to 46% for BD-IPMNs), few reliable tools exist for predicting the risk of an individual lesion progressing to carcinoma [49].”
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 3. Results > 3.2. Detection and Risk Stratification of Pancreatic Cystic Lesions
What the sources say
- Supportedcites [49] Machado, 2015, Intraductal Papillary Mucinous Neoplasm of Pancreas, N. Am. J. Med. Sci. source
The section on indications for resection of MD-IPMN states that there are no consistent predictive factors for malignancy. The preoperative-predictors section likewise says that distinguishing benign from malignant IPMN remains a concern despite advances in imaging. Together, these support the claim that reliable prediction for an individual lesion is limited; page numbers are not available in our copy. (LLM, conf=0.87)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“In comparison, “worrisome features”, as defined by consensus guidelines (main duct dilation, cyst size > 3 cm, cyst wall thickening, mural nodules, or acute pancreatitis), yielded an AUC of only 0.54.”
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 3. Results > 3.2. Detection and Risk Stratification of Pancreatic Cystic Lesions
What the sources say
- Supportedcites [53] Permuth, 2016, Combining radiomic features with a miRNA classifier may improve prediction of malignant pathology for pancreatic intraductal papillary mucinous neoplasms, Oncotarget source
The Introduction defines the consensus-guideline worrisome features as main-duct dilation, cyst size >3 cm, thickened cyst walls, non-enhanced mural nodules, or acute pancreatitis. Table 3 in the Results reports an AUC of 0.54 (95% CI: 0.38–0.69) for worrisome features alone in this study cohort. (LLM, conf=0.94)
plainn = 38- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Its predictive ability has sometimes surpassed that of trained specialists or international guidelines.”
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 4. Discussion
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
5 more claims
Create a free account to see all 35 claims and to verify the 1 reference still open.
- It may state two separate claims as one. 17
- Probably not a substantial claim. 12
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 3. Results > 3.2. Detection and Risk Stratification of Pancreatic Cystic Lesions
“The deep learning protocol achieved a sensitivity and specificity of 75% and 78%, respectively, with an AUC of 0.90 (95% confidence interval 0.71–0.85) for detecting high-grade dysplasia or cancer.”
§ Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma > 3. Results > 3.2. Detection and Risk Stratification of Pancreatic Cystic Lesions
“with an AUC of 0.90 (95% confidence interval 0.71–0.85)”
Citations
Cites
Siegel, R. L., Miller, K. D., & Jemal, A. (2020) · CA: A Cancer Journal for Clinicians · doi:10.3322/caac.21590 · Cited by 20,444
Rahib, L., Smith, B. D., Aizenberg, R., Rosenzweig, A. B., Fleshman, J. M., & Matrisian, L. M. (2014) · Cancer Research · doi:10.1158/0008-5472.can-14-0155 · Cited by 7,094
Pancreatic cancer: A review of clinical diagnosis, epidemiology, treatment and outcomes
McGuigan, A., Kelly, P., Turkington, R. C., Jones, C., Coleman, H. G., & McCain, R. S. (2018) · World Journal of Gastroenterology · doi:10.3748/wjg.v24.i43.4846 · Cited by 1,908
Park, W., Chawla, A., & O’Reilly, E. M. (2021) · JAMA · doi:10.1001/jama.2021.13027 · Cited by 1,882
Revisions of international consensus Fukuoka guidelines for the management of IPMN of the pancreas
Tanaka, M., Fernández-del Castillo, C., Kamisawa, T., Jang, J. Y., Levy, P., Ohtsuka, T., Salvia, R., Shimizu, Y., Tada, M., & Wolfgang, C. L. (2017) · Pancreatology · doi:10.1016/j.pan.2017.07.007 · Cited by 1,797
Confocal laser endoscopy for diagnosing intraepithelial neoplasias and colorectal cancer in vivo
Kiesslich, R., Burg, J., Vieth, M., Gnaendiger, J., Enders, M., Delaney, P., Polglase, A., McLaren, W., Janell, D., Thomas, S., et al. (2004) · Gastroenterology · doi:10.1053/j.gastro.2004.06.050 · Cited by 843
ACG Clinical Guideline: Diagnosis and Management of Pancreatic Cysts
Elta, G. H., Enestvedt, B. K., Sauer, B. G., & Lennon, A. M. (2018) · American Journal of Gastroenterology · doi:10.1038/ajg.2018.14 · Cited by 719
Frequent detection of pancreatic lesions in asymptomatic high-risk individuals
Canto, M. I., Hruban, R. H., Fishman, E. K., Kamel, I. R., Schulick, R., Zhang, Z., Topazian, M., Takahashi, N., Fletcher, J., Petersen, G., et al. (2012) · Gastroenterology · doi:10.1053/j.gastro.2012.01.005 · Cited by 648 · cited 1×
“Despite extensive research into novel predictors and effect sizes of known risk factors, there is currently no unified recommendation on a high-risk population (outside of those with hereditary cancer syndromes) that may benefit from dedicated screening [34].”
BackgroundScheiman, J. M., Hwang, J. H., & Moayyedi, P. (2015) · Gastroenterology · doi:10.1053/j.gastro.2015.01.014 · Cited by 437
Valsangkar, N. P., Morales-Oyarvide, V., Thayer, S. P., Ferrone, C. R., Wargo, J. A., Warshaw, A. L., & Fernández-del Castillo, C. (2012) · Surgery · doi:10.1016/j.surg.2012.05.033 · Cited by 418
Research works printed but not linked: 1
- Moulin, 2022, [Diabetes and pancreatic cancer—When diabetes indicates a pancreatic cancer], Rev. Med. Suisse, 18, 1994
Cited by
Gonda, T. A., Cahen, D. L., & Farrell, J. J. (2024) · New England Journal of Medicine · doi:10.1056/nejmra2309041 · Cited by 136
Diagnosis and Management of Pancreatic Cysts
Gardner, T. B., Park, W. G., & Allen, P. J. (2024) · Gastroenterology · doi:10.1053/j.gastro.2024.02.041 · Cited by 90
Artificial Intelligence in Pancreatic Imaging: A Systematic Review
Podină, N., Gheorghe, E. C., Constantin, A., Cazacu, I., Croitoru, V., Gheorghe, C., Balaban, D. V., Jinga, M., Țieranu, C. G., & Săftoiu, A. (2025) · United European Gastroenterology Journal · doi:10.1002/ueg2.12723 · Cited by 48
The application of artificial intelligence in EUS
Zhang, D., Wu, C., Yang, Z., Yin, H., Liu, Y., Li, W., Huang, H., & Jin, Z. (2024) · Endoscopic Ultrasound · doi:10.1097/eus.0000000000000053 · Cited by 30
Pancreatic Adenocarcinoma: Imaging Modalities and the Role of Artificial Intelligence
Anghel, C., Mugur, G. C., Anghel, D. A., Rusu-Munteanu, G. I., Dumitru, R. L., & Lupescu, I. G. (2024) · doi:10.20944/preprints202401.0904.v1 · Cited by 23
Metadata
- Authors
- Jiang, J., Chao, W.-L., Culp, S., & Krishna, S. G.
- Affiliations
- The Ohio State University Wexner Medical Center (US)
- The Ohio State University (US)
- Corresponding author
- Krishna, Somashekar G.
- Venue
- Cancers
- Publisher
- MDPI AG
- Published
- 2023-04-22
- Volume
- 15
- Issue
- 9
- Pages
- 2410
- ISSN
- 2072-6694 (ISSN-L)
- PMC
- PMC10177524
- OpenAlex
- W4366975505
- Cited by
- 38
- References
- 118
- Citations by year
- 2026: 9 · 2025: 12 · 2024: 14 · 2023: 3
- FWCI
- 0.80
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Cancers (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- eScholarship (California Digital Library) (submittedVersion, cc-by)
- Funders
- National Cancer Institute (R01CA279965)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
- Health Sciences › Medicine › Oncology › Pancreatic and Hepatic Oncology Research
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
- Keywords
- artificial intelligence, pancreatic cysts, pancreatic ductal adenocarcinoma, pancreatic intraepithelial neoplasia, intraductal papillary mucinous neoplasm, pancreatic cancer screening, risk stratification, premalignant pancreatic lesions, endoscopic ultrasound-guided fine-needle aspiration, cyst fluid analysis, outcome prediction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma. https://app.cochat.ai/papers/10.3390/cancers15092410
- The paper
- Jiang, J., Chao, W.-L., Culp, S., & Krishna, S. G. (2023). Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma. Cancers. https://doi.org/10.3390/cancers15092410
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.