Spine CT

What it does

Spine CT reads the cervical, thoracic and lumbar spine, either from a dedicated spine CT or from the spine reformats of a trauma or body CT. The run first segments and labels every vertebra from C1 to L5 and the sacrum, with the vertebral bodies and intervertebral discs, so that each later finding is tied to a named level.

On that labelled spine it looks for fractures and describes their morphology (compression, burst, distraction, translation) with an AO Spine thoracolumbar type at thoracic and lumbar levels, assesses alignment and canal compromise, and measures the canal and foramina at every level for degenerative narrowing. In the emergency department this is the read where a missed cervical fracture costs the most; in clinic it is a measured map of stenosis before surgery.

Intended use

Decision support for the clinician clearing the spine after trauma, often the emergency physician or radiology resident reading a head and cervical spine CT together at night, and for the spine surgeon planning decompression or fixation. Each fracture and narrowing is reported at a named level, with its measurements.

Who it is for

Emergency and musculoskeletal radiologists, emergency physicians, trauma teams, spine surgeons, neurosurgeons and orthopaedic surgeons. Developers use the labelled vertebrae and level-based findings in trauma worklists, surgical planning software or spine registries.

Accepted input

• Dedicated spine CT of any region, or a trauma pan-scan or chest-abdomen CT whose spine is reformatted from the body volume.
• NIfTI with explicit spacing is accepted for the vertebra segmentation.

Requirements

Requirement	Why
Bone-kernel reconstruction	Fracture lines and cortical breaks are read on the sharp kernel
Thin axial slices allowing sagittal reformats	Height loss, alignment and canal measurements are sagittal
Image Position and Orientation (Patient)	Slices are ordered geometrically, which matters for reformats of long body scans

Transitional anatomy (lumbarised or sacralised segments, cervical ribs) is flagged when detected, because it can shift level numbering. Shoulder artefact at C7–T1, osteophytes, vascular channels and Mach bands are handled as quality flags and known mimics.

Optional context

The mechanism of injury, neurological symptoms, known previous fractures or fixation hardware and the clinical question (trauma clearance or degenerative work-up) go in clinical_context.

The result

A sectioned JSON result in which every finding carries its vertebral level, with masks in DICOM SEG.

Section	Content
Spine fractures	Fracture per vertebral level, morphology (compression, burst, distraction, translation), AO Spine type at thoracolumbar levels, canal compromise
Vertebrae and discs	Masks and labels for C1–L5 and the sacrum, vertebral bodies and intervertebral discs
Canal and foraminal stenosis	Canal anteroposterior diameter and area per level, foraminal narrowing, disc-height loss, facet hypertrophy flags
Draft report	Opt-in (options.report): an English narrative that interprets the result for the reader (findings, impression, limitations); every number is checked against the findings

The trauma read also states alignment and canal compromise at each injured level.

Standards

• DICOM: SR TID 1500 Measurement Report with one measurement group per level; SEG (label map) for vertebrae and discs; key images of each fracture; Encapsulated PDF and FHIR DiagnosticReport.
• Classification: AO Spine thoracolumbar classification system (2013) for thoracic and lumbar fractures; Genant semiquantitative grading for vertebral body height loss.

Known mimics

Old healed fractures, osteophytes and degenerative endplate change can resemble acute injury, and vascular channels or Mach bands can mimic a fracture line. Each fracture therefore comes with its level, key image and morphology. For a head and cervical spine trauma study, the Head CT model reads the brain and skull in a separate run.