Cerebral DSA

What it does

Cerebral DSA reads a digital subtraction angiography study of the head and neck: the AP and lateral runs of each injected vessel (internal and external carotid, vertebral, common carotid), diagnostic or recorded during a mechanical thrombectomy. The runs of one study are grouped and read together, phase by phase, from the arterial through the capillary to the venous phase.

The result is a per-territory read. It maps the cerebral arterial tree on every frame, locates a large-vessel occlusion, grades reperfusion on the expanded and modified TICI scales after thrombectomy, lists aneurysms with size, neck width, dome-to-neck ratio and parent vessel, describes arteriovenous shunts (AVM nidus or dural fistula), grades intracranial and extracranial stenosis by the WASID and NASCET methods, and measures vessel-diameter change on follow-up studies after subarachnoid haemorrhage.

Intended use

Decision support for physicians who perform or read cerebral angiography: a structured second read for the resident or general radiologist reading a diagnostic DSA in a smaller centre, and a documented reperfusion grade for the operator closing a stroke case. Every aneurysm, shunt and occlusion comes with its measurements and the frames that show it.

Who it is for

Neurointerventionalists, interventional and diagnostic neuroradiologists, stroke neurologists, neurosurgeons and radiology residents. Stroke registries and research teams use the same structured output to record occlusion site and final eTICI consistently.

Accepted input

• DSA runs: multi-frame X-Ray Angiographic Image objects (including Enhanced XA), subtracted or native, or a zip of the single-frame instances of one run.
• Whole studies: a patient CD or study folder with a DICOMDIR; runs are grouped by study and duplicates skipped.
• Thrombectomy cases: pre- and post-thrombectomy runs of the treated territory in the same study.
• Vasospasm follow-up: the follow-up DSA, with the baseline study in prior_files when it exists.

Requirements per run

Requirement	Why
At least 16 frames	Phase detection needs the contrast passage over time
Not a 3D rotational acquisition	Rotational runs cannot be read as fixed AP or lateral projections
Capillary and venous phases recorded	TICI grading and early venous filling need the late frames
Pixel spacing with source distances (or a magnification factor)	Millimetre sizes for aneurysms and diameters; without it, values stay in pixels and are flagged

Aneurysm sizes come from calibrated 2D projections and can read smaller than on 3D rotational angiography; the result says which projection each measurement comes from.

Optional context

The clinical question (acute stroke, subarachnoid haemorrhage, known aneurysm or AVM, follow-up), the timing after haemorrhage for vasospasm studies, and the treated vessel in thrombectomy cases can be passed in clinical_context.

The result

One structured result per study, with findings, measurements, scores and evidence frames for every section.

Section	Content
Vessel segmentation	Cerebral arterial tree mask on every frame and a minimum-intensity projection per run
Large-vessel occlusion	Occlusion site (ICA terminus, M1, M2, basilar) and collateral filling of the territory
Reperfusion grade (mTICI/eTICI)	Grade from pre- and post-thrombectomy runs, with the reperfused fraction of the territory
Aneurysms	Each aneurysm with location, size, neck width, dome-to-neck ratio and parent vessel
AVM and dural fistula	Detection of an AVM nidus or dural fistula, nidus size, arterial feeders and venous drainage pattern
Stenosis	Percent diameter stenosis of intracranial or extracranial segments, with the method and reference segment used
Vasospasm	Diameter change per segment against baseline or reference values, with severity
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

Standards

• DICOM: SR Measurement Report TID 1500; SEG for the vessel masks, coded with CID 12105/12106 Intracranial Cerebral Vessel; Secondary Capture key frames and overlay video.
• Reperfusion: mTICI consensus (Zaidat, Stroke 2013) and eTICI (Liebeskind, J Neurointerv Surg 2019).
• Stenosis: WASID method (Samuels, AJNR 2000) for intracranial arteries; NASCET method, against the normal distal cervical internal carotid artery, for the extracranial carotid.
• AVM: nidus size and venous drainage are two of the three Spetzler–Martin elements (J Neurosurg 1986); eloquence of the adjacent brain is not visible on DSA, so the run reports the two elements without a total grade.

Reading limits

Overlapping vessel loops and infundibula can mimic small aneurysms, misregistration on subtracted frames can mimic filling defects, and a reperfusion grade needs the capillary phase. Findings affected by these limits carry a quality flag with the run and frames concerned.