You are here

  • Home
  • Archive
  • Volume 70, Issue 7
  • Utilisation of artificial intelligence for the development of an EUS-convolutional neural network model trained to enhance the diagnosis of autoimmune pancreatitis

Article Text

Article menu
Download PDFPDF
Pancreas
Original research
Utilisation of artificial intelligence for the development of an EUS-convolutional neural network model trained to enhance the diagnosis of autoimmune pancreatitis
  1. Neil B Marya1,
  2. Patrick D Powers2,
  3. Suresh T Chari3,
  4. Ferga C Gleeson1,
  5. Cadman L Leggett1,
  6. Barham K Abu Dayyeh1,
  7. Vinay Chandrasekhara1,
  8. Prasad G Iyer1,
  9. Shounak Majumder1,
  10. Randall K Pearson1,
  11. Bret T Petersen1,
  12. Elizabeth Rajan1,
  13. Tarek Sawas1,
  14. Andrew C Storm1,
  15. Santhi S Vege1,
  16. Shigao Chen4,
  17. Zaiyang Long4,
  18. David M Hough4,
  19. Kristin Mara5,
  20. Michael J Levy1
  1. 1 Gastroenterology and Hepatology, Mayo Clinic, Rochester, Minnesota, USA
  2. 2 Independent Researcher, Chelsea, Massachusetts, USA
  3. 3 Gastroenterology, Hepatology and Nutrition, University of Texas MD Anderson Cancer Center, Houston, Texas, USA
  4. 4 Diagnostic Radiology, Mayo Clinic, Rochester, Minnesota, USA
  5. 5 Biomedical Statistics and Informatics, Mayo Clinic Rochester, Rochester, Minnesota, USA
  1. Correspondence to Dr Michael J Levy, Gastroenterology and Hepatology, Mayo Clinic, Rochester, MN 55905, USA; levy.michael{at}mayo.edu

Abstract

Objective The diagnosis of autoimmune pancreatitis (AIP) is challenging. Sonographic and cross-sectional imaging findings of AIP closely mimic pancreatic ductal adenocarcinoma (PDAC) and techniques for tissue sampling of AIP are suboptimal. These limitations often result in delayed or failed diagnosis, which negatively impact patient management and outcomes. This study aimed to create an endoscopic ultrasound (EUS)-based convolutional neural network (CNN) model trained to differentiate AIP from PDAC, chronic pancreatitis (CP) and normal pancreas (NP), with sufficient performance to analyse EUS video in real time.

Design A database of still image and video data obtained from EUS examinations of cases of AIP, PDAC, CP and NP was used to develop a CNN. Occlusion heatmap analysis was used to identify sonographic features the CNN valued when differentiating AIP from PDAC.

Results From 583 patients (146 AIP, 292 PDAC, 72 CP and 73 NP), a total of 1 174 461 unique EUS images were extracted. For video data, the CNN processed 955 EUS frames per second and was: 99% sensitive, 98% specific for distinguishing AIP from NP; 94% sensitive, 71% specific for distinguishing AIP from CP; 90% sensitive, 93% specific for distinguishing AIP from PDAC; and 90% sensitive, 85% specific for distinguishing AIP from all studied conditions (ie, PDAC, CP and NP).

Conclusion The developed EUS-CNN model accurately differentiated AIP from PDAC and benign pancreatic conditions, thereby offering the capability of earlier and more accurate diagnosis. Use of this model offers the potential for more timely and appropriate patient care and improved outcome.

  • autoimmune disease
  • pancreas
  • chronic pancreatitis
  • pancreatic cancer
  • pancreatitis

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study. Data sharing not applicable as this is not a clinical trial. Please contact the corresponding author—MJL (levy.michael@mayo.edu)—regarding requests for any additional information sharing.

https://doi.org/10.1136/gutjnl-2020-322821

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Data availability statement

    Data sharing not applicable as no datasets generated and/or analysed for this study. Data sharing not applicable as this is not a clinical trial. Please contact the corresponding author—MJL (levy.michael@mayo.edu)—regarding requests for any additional information sharing.

    View Full Text

    Footnotes

    • Contributors NM: Conception and design; analysis and interpretation of the data; dataset management; drafting of the article; critical revision of the article; data extraction; final approval of article. PDP: Model development; drafting of article segments relative to model development. STC: Critical revision of the article; data extraction; final approval of article. FG: Critical revision of the article; final approval of article. CLL: Critical revision of the article; final approval of article. BKAD: Critical revision of the article; data extraction; final approval of article. VC: Critical revision of the article; final approval of article. PGI: Critical revision of the article; final approval of article. SM: Critical revision of the article; final approval of article. RKP: Critical revision of the article; final approval of article. BTP: Critical revision of the article; final approval of article. ER: Critical revision of the article; final approval of article. TS: Critical revision of the article; final approval of article. ACS: Critical revision of the article; final approval of article. SSV: Critical revision of the article; final approval of article. SC: Critical revision of the article; final approval of article. ZL: Critical revision of the article; final approval of article. DH: Critical revision of the article; final approval of article. KM: Critical revision of the article; statistical analysis; final approval of article. MJL: Conception and design; analysis and interpretation of the data; drafting of the article; critical revision of the article; data extraction; final approval of article.

    • Funding This publication was made possible by the Mayo Clinic CTSA through grant number UL1TR000135 from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH). This publication was also made possible by a pilot grant from the Mayo Clinic Ultrasound Research Center.

    • Competing interests PDP: is an independent researcher who was compensated by Mayo Clinic grant funds to participate in AI model research.

    • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.