BACKGROUND: About 30-40% of PDAs can be missed on CTs due to subtle imaging findings. Unfortunately, PDA can advance from being undetectable to widely metastatic in a matter of months which underscores the critical need to develop novel tools such as AI to improve its detection. Most current AI-based PDA detection approaches require manual segmentations that are not clinically scalable. Our objective was to develop a 3D Convolutional Neural Network (CNN) for fully automated detection of PDA on standard-of-care CTs.

METHODS: A fully-automated 3D-CNN was trained on a dataset of 696 portal venous phase diagnostic CTs with PDA and 1080 control CTs with normal pancreas. Model was evaluated on (a) test subset of 409 PDA and 829 control CTs; (b) simulated cohort with high pre-test probability of PDA (1-5% over 3-years), which matched the case-control distribution in high-risk groups (glycemically-defined new-onset diabetes and high ENDPAC score); (c) multi-institutional public dataset of 194 PDA and 80 control CTs; and (d) a cohort of 102 prediagnostic CTs (i.e. CTs with normal-appearing pancreas incidentally acquired 3–36 months before clinical diagnosis of PDA) and 134 control CTs.

RESULTS: In the test subset, the model correctly classified 360 (88%) PDA CTs and 783 (94%) normal CTs [accuracy (mean; 95% CI) 0·92 (0·91-0·94); AUROC 0·97 (0·96-0·98), sensitivity 0·88 (0·85-0·91), specificity 0·95 (0·93-0·96)]. Performance was comparable across tumor stages (sensitivity 0·80, 0·87, 0·95 and 1 on T1 through T4 stages, respectively), hypodense versus isodense tumors (sensitivity: 0·90 vs. 0·82), age, sex, CT slice thicknesses & vendors (all p >0·05), and generalizable on both the simulated cohort [accuracy 0·95 (0·94-0·95), AUROC 0·97 (0·94-0·99)] and public dataset [accuracy 0·86 (0·82-0·90), AUROC 0·90 (0·86-0·95), sensitivity 0·88, specificity 0·93] including for T1 and isodense tumors (sensitivity 0·76 and 0·88, respectively). Despite being trained on diagnostic CTs, the model could detect PDA at the prediagnostic stage from normal pancreas [accuracy 0·84 (0·79-0·88), AUROC 0·90 (0·87-0·94), sensitivity 0·75 (0·66-0·83), specificity 0·90 (0·85-0·95)] at a median interval of 475 days (range:93-1082 days) between prediagnostic CTs and the clinical diagnosis of PDA. 

CONCLUSION: A 3D-CNN developed on a large dataset shows high accuracy and generalizability for fully-automated detection of PDA (including small and isodense tumors) on diagnostic CTs as well as for early detection of PDA from normal pancreas on prediagnostic CTs at a substantial lead time prior to clinical diagnosis. Prospective validation with blood-based biomarkers could enable AI-augmented early detection of PDA in high-risk subjects.

AI-Powered Fully automated Early Detection of Pancreatic Ductal Adenocarcinoma on Standard-of-care CT Scans.

Clinical and Translational Research 1

technical paper

Welcome to the largest national, multi-specialty research event **featuring research from 750 medical students, residents and fellows, and international medical graduates.**


#### Event Details
The 2022 American Medical Association Research Challenge includes a poster symposium hall and a semifinals hall with the top scored research. Log in at any time throughout the event to:
* **View posters and presentations** on a variety of topics and specialties
* **Score posters in the semifinals** to help decide the top five finalists (AMA member exclusive opportunity)
* **Connect with participants** and provide feedback to help advance their research

At the conclusion of this event, the AMA will announce the top five finalists who will advance to the AMA Research Challenge finals, present their research to an elite panel of judges and compete to win a $10,000 grand prize sponsored by Laurel Road.
 

#### Attend Live Forums
**Oct. 20, 6 — 7 p.m. Central**
 **How to plan a career in academic medicine** 
 Learn how to plan a career in academic medicine and get your questions answered. [Access Here](https://underline.io/events/363/lounge)
 
 **Speakers**
![](https://assets.underline.io/uploads/markdown_image/1/image/d5adb133dfc58eb0a6d469e9cedcd5d4.jpg)
Ricardo Correa, MD, EdD
![](https://assets.underline.io/uploads/markdown_image/1/image/5f9437ab87861538bf02b9ca7c0d926e.jpg)
Stuart M. Greenstein, MD
![](https://assets.underline.io/uploads/markdown_image/1/image/05b1b637c99286d558c371f8e8b185ce.jpg)
Stephanie Strohbeen, MD
 
 **Oct. 22, 11 a.m. — Noon Central**
 **How to get published**
 Learn tips to get your research published and get your questions answered. [Access Here](https://underline.io/events/363/lounge)
 
 **Speakers**
![](https://assets.underline.io/uploads/markdown_image/1/image/8ce6235e957d31cb5e6f9097cc6f6972.jpg)
Jessica Li, MD/PhD Candidate
![](https://assets.underline.io/uploads/markdown_image/1/image/9950ef2e0237f54cb3d8280334d42491.jpg)
Charles Lopresto, DO

### Sponsor
AMA would like to thank Laurel Road for sponsoring a $10,000 grand prize for the winner of the AMA Research Challenge. Laurel Road is not involved in the judging or scoring process to select the winner. The winner will be announced on Dec. 7 during the AMA Research Challenge finals. Payment of the grand prize will be directly from Laurel Road. See the AMA Research Challenge Official Rules for terms and conditions.

![](https://assets.underline.io/uploads/markdown_image/1/image/611e331d4a0eba0920f240cfc04b4d77.png)

**There is no registration fee for this event! **

Register to explore research in a variety of topics. AMA members get exclusive access to score posters in the semifinals.

**Please use the link below to register for free if you have not already done so:**

https://www.ama-assn.org/about/events/ama-research-challenge

Registration Is Required

AMA Research Challenge 2022

Welcome to the largest national, multi-specialty research event featuring research from 750 medical students, residents and fellows, and international medical graduates.


Garima Suman

1

31

SHORT BIO

Presentations

AI-Powered Fully automated Early Detection of Pancreatic Ductal Adenocarcinoma on Standard-of-care CT Scans.