Childhood Cancer Data Initiative Annual Symposium (Abstract Registration): Submission #60

Abstract Submission for Poster Presentation
-------------------------------------------
Abstract Title:: Generalizable Pediatric Sarcoma Histopathology Classification with Multi-Institutional Machine Learning
Abstract::
Intro: Digitization of histopathology slides has allowed for the use of computational machine learning and artificial intelligence (AI)–based approaches to aid in diagnostics. These tools could be especially helpful for classifying pediatric sarcoma subtypes, which are rare and heterogeneous, and whose diagnoses often require costly genetic and molecular testing that may not be available to every patient. These machine learning models offer great promise but come with the caveat of being prone to overfitting to an individual institution’s microscope, scanner, and staining protocol, which can affect performance in real-world clinical settings where instruments and protocols differ across hospitals. Therefore, it can be difficult to make these models generalizable for use globally if they are not trained on a large and diverse dataset.

Methods: We have curated over 700 H&E images from four institutions spanning over 10 different sarcoma subtypes. We utilize an in-house, open-source pipeline for stain normalization, focus checking, and cropping to harmonize the images. AI models are then used to extract features from these images, which can be used for downstream SAMPLER-based machine learning.

Results: We achieve state-of-the-art results in classifying images as rhabdomyosarcoma vs non-rhabdomyosarcoma soft tissue sarcomas (AUC 0.969 ± 0.026), alveolar vs embryonal rhabdomyosarcoma (AUC 0.961 ± 0.021), and Ewing sarcoma (AUC 0.929). Importantly, our models generalize well when tested on data from previously unseen institutions, outperforming similar methods.

Conclusion: Our pipeline is well suited for additional collaboration and could be a tool to help bridge access to clinical resources globally.


Abstract:: {Empty}
Authors::
1. First Name: Adam
   Middle Initial: H.
   Last Name: Thiesen
   Organization: The Jackson Laboratory, UConn School of Medicine
2. First Name: Sergii
   Last Name: Domanskyi
   Degree(s): Ph.D.
   Organization: The Jackson Laboratory
3. First Name: Ali 
   Last Name: Foroughi pour
   Degree(s): Ph.D.
   Organization: St. Jude Children's Research Hospital
4. First Name: Jingyan
   Last Name: Zhang
   Organization: Johns Hopkins University
5. First Name: Todd
   Middle Initial: B.
   Last Name: Sheridan
   Degree(s): M.D.
   Organization: Hartford Healthcare
6. First Name: Steven
   Middle Initial: B.
   Last Name: Neuhauser
   Organization: The Jackson Laboratory
7. First Name: Alyssa
   Middle Initial: E.
   Last Name: Stetson
   Degree(s): M.D. MPH
   Organization: Massachusetts General Hospital Department of Surgery
8. First Name: Katelyn
   Last Name: Dannheim
   Degree(s): M.D.
   Organization: Massachusetts General Hospital Department of Pathology
9. First Name: Danielle
   Middle Initial: B.
   Last Name: Cameron
   Degree(s): M.D. MPH
   Organization: Massachusetts General Hospital 
10. First Name: Shawn 
    Last Name: Anh
    Degree(s): M.D. Ph.D.
    Organization: University of Pennsylvania Department of Surgery
11. First Name: Hao
    Last Name: Wu
    Degree(s): M.D. Ph.D.
    Organization: Yale School of Medicine Department of Pathology
12. First Name: Emily 
    Middle Initial: R.
    Last Name: Christison-Lagay
    Degree(s): M.D.
    Organization: Yale School of Medicine Department of Surgery
13. First Name: Carol
    Middle Initial: J.
    Last Name: Bult
    Degree(s): Ph.D.
    Organization: The Jackson Laboratory
14. First Name: Jeffrey
    Middle Initial: H.
    Last Name: Chuang
    Degree(s): Ph.D.
    Organization: The Jackson Laboratory, UConn School of Medicine
15. First Name: Jill 
    Middle Initial: C.
    Last Name: Rubinstein
    Degree(s): M.D. Ph.D.
    Organization: The Jackson Laboratory, Hartford Healthcare, UConn School of Medicine

Presenting Author:: Adam Thiesen
Institution:: The Jackson Laboratory
Email Address:: adam.thiesen@jax.org