Symposium

MOOSE 2026

Machine-learning in Ophthalmology & Ocular Science Exchange

Monday, November 16, 2026 · 5:00 to 6:50 PM
Auditorium G, Dartmouth Hitchcock Medical Center

Download the flyerJoin remotely, email us

Twenty-six students present their research in two-minute talks, each followed by a minute for questions. Faculty score the talks during the reception, and the top three receive awards at 6:45 PM. Hosted and moderated by Benton Chuter, MD, MS, and William Gensheimer, MD.

Program

  1. 5:00 PMOpening remarksBenton Chuter, MD, MS2 min
  2. 5:0301Per-eye decay curves beat a neural operator for cell density forecasts in CPTS and CDS-SMASAugust Mondragon with T. Daley, F. B. Kayarian, M. Jablonski, W. Gensheimer, B. Chuter
  3. 5:0602Machine learning on Castro-Luna Pentacam data discriminates subclinical keratoconus better than the published modelNoelle Fedor with S. Azam, C. Joe, L. Larned, W. Gensheimer, B. Chuter
  4. 5:0903Interpretable measurement beats prior AI for donor endothelial counts in CPTSKatie Cox with D. Zhu, C. Joe, F. B. Kayarian, W. Gensheimer, B. Chuter
  5. 5:1204Single-scan corneal stiffness is not fully identifiable, and shared-mesh tests overstate its recoveryBrian Lihari with A. Lihari, W. Gensheimer, B. Chuter
  6. 5:1505Endothelial cell density and corneal thickness are inversely correlated within eyes after PKP and DSAEKAidan Lihari with B. Lihari, W. Gensheimer, B. Chuter
  7. 5:1806Public CAMG meibography yields first reference intervals for meibomian gland structure in childrenChloé Lawrence with A. Kangsumrith, A. Isaacs, W. Gensheimer, B. Chuter
  8. 5:2107Automated corneal nerve length on Lagali mosaics correlates well with tracing but agrees less tightlyBailey Millis with G. Park, A. Tavasli, F. B. Kayarian, R. A. Badian, W. Gensheimer, B. Chuter
  9. 5:2408Deep learning endothelial cell counts on the Freiburg specular set do not reach clinical acceptabilityJenna Jeon with S. Bhandary, C. Ye, C. Joe, W. Gensheimer, B. Chuter
  10. 5:2709A digital twin fitted to CPTS and tested on CDS separates surgical and chronic endothelial cell lossShad Hassan with A. Shaadab, L. Cano-Bartolo, Z. Usman, S. Shah, F. B. Kayarian, B. Millis, M. Jablonski, W. Gensheimer, B. Chuter
  11. 5:3010A corneal foundation model counts dendritic cells and keratocytes on CORN nearly as well as supervised AIAlanna Polyak with B. Han, H. Kaaua, M. Hoefer, W. Gensheimer, B. Chuter
  12. 5:3311A GSE77938 keratoconus gene-expression classifier does not reliably transfer across tissue or cohortTaylor Wolfe with L. Galton, S. Yoo, W. Gensheimer, B. Chuter
  13. 5:3612One network matches a published model ensemble for meibomian gland segmentation on MGD-1KTheo Kapogiannis with C. Joe, L. Saxena, W. Gensheimer, B. Chuter
  14. 5:3913Data splits and shortcuts overstate AI for keratitis on the Essalat–Le confocal datasetBrandon Horn with A. Shukla, T. Zima, H. Smith, F. B. Kayarian, W. Gensheimer, B. Chuter
  15. 5:42 PMBreak8 min
  16. 5:5014Corneal confocal AI inherits most of its keratitis accuracy on Essalat–Le from a general modelKevin He with S. De Andrade, A. Huang, R. A. Badian, W. Gensheimer, B. Chuter
  17. 5:5315Off-the-shelf image features match the keratoconus screening benchmark on Al-Timemy Pentacam mapsAlanna Polyak with W. Gensheimer, B. Chuter
  18. 5:5616General AI models match fundus-trained ones on Retina-SLO and RAVIR laser retinal scansLibby Galton with T. Surakiatchanukul, W. Gensheimer, B. Chuter
  19. 5:5917One self-supervised transformer matches or outperforms established methods for visual field imputation and forecastingDiya Patel with C. Joe, I. Allen, W. Gensheimer, B. Chuter
  20. 6:0218Disease labels, not cross-modal alignment, cut corneal AI label needs across MCOA, SLID and SUSTech-SYSUAsli Tavasli with C. Joe, A. Kakkera, W. Gensheimer, B. Chuter
  21. 6:0519A systematic review of open globe injury detection from eye photographsHelena Kaaua with Y. Mustafa, T. Nguyen, B. Millis, G. Richardson, K. Alozie, F. B. Kayarian, W. Gensheimer, B. Chuter
  22. 6:0820Donor endothelial grading and specular images did not measurably improve graft-failure prediction in CPTSChelsea Joe with M. Jablonski, W. Gensheimer, B. Chuter
  23. 6:1121Transplant center identity improves corneal graft failure prediction in CPTSZofia Cieslak with C. Joe, K. Cox, W. Gensheimer, B. Chuter
  24. 6:1422The Cornea Preservation Time Study's cell-loss baseline predicts its own exposureShoaib Jamil with C. Joe, Y. Mustafa, T. Kapogiannis, W. Gensheimer, B. Chuter
  25. 6:1723Corneal graft risk models show validation gaps and lose accuracy for new surgeons in CPTSYuseph Mustafa with C. Joe, S. Jamil, T. Kapogiannis, W. Gensheimer, B. Chuter
  26. 6:2024Papilledema AI accuracy on the Salaheldin OCT release depends on how the data are splitLori Saxena with T. Surakiatchanukul, W. Gensheimer, B. Chuter
  27. 6:2325A few hundred local readings adapt automated endothelial cell density estimates to the VisionGift eye bankChelsea Joe with M. Jablonski, J. Clover, W. Gensheimer, B. Chuter
  28. 6:2626Published data splits inflate retinopathy of prematurity AI results across RIDIRP, A-Fundus and HVDROPDBMin Young Kim with A. Kakkera, I. Allen, C. Saint James, Z. Cieslak, T. Surakiatchanukul, W. Gensheimer, B. Chuter
  29. 6:29 PMClosing remarksWilliam Gensheimer, MD2 min
  30. 6:31 PMReception and networkingStudents and faculty network while talks are scored14 min
  31. 6:45 PMAwards for the top three talksDrs. Chuter and Gensheimer5 min
MOOSE 2026 symposium logo, a round green badge with an antlered mascot in glasses