Open data and reproducible benchmarks for spine imaging AI.

The consortium is a collaborative, student-driven research effort — clinicians, computer scientists, and medical students building computational spine imaging together. Each contributor's scholarly contributions to OSC are listed on their card.

OpenSpineConsortium is also a medical-education model. The same open, asynchronous framework that produces our datasets turns students with little coding or imaging background into contributors with conference-ready scholarship — an outcome we measure rather than assume.

A free, hands-on morning on building spine-imaging AI — from reproducible Python and SLURM workflows to running, annotating, and training your own neural networks on the WSU grid. Bring a laptop and follow along.

Where the work gets presented, and by whom. Almost every project here is student-led from first commit to podium — the consortium exists to put medical students in front of an audience with something of their own to say.

Michigan Association of Neurological Surgeons — Annual Meeting

August 14–16, 2026 Crystal Mountain, Michigan

↓  Agenda (PDF) ↓  Slides (PPTX, 12 MB)

Concurrent session for advanced practitioners

1. Invited

   Gregory Schwing, MD, PhD Saturday 15 August, 1:45 pm · Arctic Bay

   When the Cord Is Compressed: Neuroradiology Essentials for Surgical Decision-Making in Degenerative Spine Disease

Preston M. Hickey Medical Student Symposium

August 7, 2026 Wayne State University School of Medicine · Detroit, MI

↓  Program (PDF)

Oral presentation awards

1. 1st

   Ashley Schehr M2, Wayne State University School of Medicine

   Automated CT-Based Cobb Angle Measurement for Scoliosis Evaluation: A Computational Imaging Pipeline Across 1,156 Cases

2. 3rd

   Razanne Zaghloul M2, Wayne State University School of Medicine

   Automated Measurement of Lumbar Intervertebral Disc Heights from CT Segmentation Masks: A Preliminary Validation Study

Consortium projects presented

• Ashley Schehr M2

  Automated CT-Based Cobb Angle Measurement for Scoliosis Evaluation: A Computational Imaging Pipeline Across 1,156 Cases

• Razanne Zaghloul M2

  Automated Measurement of Lumbar Intervertebral Disc Heights from CT Segmentation Masks: A Preliminary Validation Study

• Hassan Saad M2

  Automated TRALL Angle Computation from CT Segmentation: A Scalable Pipeline for Lumbar Lordosis Quantification

• Jerick Kim M2

  Straightening Out the Curve: Automated Analysis of Lordosis Distribution Index Measurement Using CT Segmentation Masks

• Dane Hubers M2

  CT-Based Automated Morphometric Analysis of the Lumbar Spinal Canal: Normative Torg Ratio and Canal Diameter Values Across L1–L5

• Jason Fahmy M2

  Automated Quantification of Lumbar Spondylolisthesis Using CT-Derived Vertebral Landmarks

• Namrata Vedagarbha M3

  Opportunistic Screening for Vertebral Degeneration Using Automated Analysis of Vertebral Hounsfield Units

Seven consortium projects were presented at the symposium, all of them built on the shared CT segmentation pipeline. Titles are as printed in the official program. Consortium members also present their own unrelated work; this list is the consortium’s, not a roster of everything its members did.

American Association of Clinical Anatomists — 43rd Annual Meeting

June 12–15, 2026 Mayo Clinic, Rochester, Minnesota

Tech Fair oral presentation

1. Oral

   Gregory Schwing, MD, PhD Wayne State University · Detroit Medical Center

   Harnessing Algorithmic Disagreement for Automated Lumbosacral Transitional Vertebrae Detection

Surfaces rendered straight from the released segmentations, each on a turntable — drag a specimen to turn it. Below them, the shape of the cohort. loading…

Every measure here is count-free: none of them requires knowing which vertebra is which. A transitional vertebra cannot be named from a spine-limited field of view — the same bone is a sacralised L5 or a lumbarised S1 depending only on where the count starts — so these describe shape and intervals instead.

Surfaces are decimated for the browser and are not the release geometry; the labels themselves are published at full resolution on HuggingFace. Images originate from the public TCIA CT COLONOGRAPHY collection.

11 of the 802 records carry surgical instrumentation, in four kinds. Every implant is labelled as its own structure rather than left inside the bone that a segmenter had absorbed it into — each one below is rendered from the released segmentation.

Every case was read by a radiologist before it was labelled. Filter on hardware_labelled in the manifest before any measurement that depends on a gap between bones: a cage-bridged interspace reads as “no gap” exactly as a congenitally fused transitional vertebra does. In 9 cases the femoral head itself is an implant, which is the landmark pelvic incidence and pelvic tilt are measured from.