7.3 Post-Processing, Reconstruction & PACS

Post-Processing, Reconstruction & PACS

ARRT CI A.5 post processing (e.g., reconstruction) is what you do to a run after the beam is off. A.6 archiving/PACS is what becomes the legal image record, how it travels, and how dose is documented. Pretty pixels that never reach the archive do not help the next operator, the quality committee, or a later lawsuit. Ugly pixels that are the only stored coronary still count.

Quick Answer: Pixel shift fixes small DSA misregistration. Window/level and edge enhancement change display, not photons. Stenosis software is an adjunct, not gospel. 3D reconstruction comes from a rotational run. Cine/DSA is what you archive as the diagnostic series; fluoro-store is a lower-dose extra. PACS and CVIS hold DICOM images. RDSR holds dose. HIPAA still applies when you burn a disk or send a case.

Pixel shift and landmarking

Pixel shift translates (and sometimes rotates) the DSA mask relative to the filled frames so bone edges cancel again after a small breath or table nudge. It is the first rescue for mild misregistration. Limits:

• It cannot create contrast that was never injected.
• It cannot un-overlap two vessels that truly overlie in that projection.
• Large motion, a cough, or a different cardiac phase than the mask usually need a repeat run, not twenty shifts.

Landmarking drops annotations on anatomy — a stenosis, a side branch, a calcium nodule, a sternal-wire reference. Landmarks help the operator and the later reader. They are not a substitute for an unforeshortened projection. A labeled “90%” on a foreshortened RAO does not become true because the text is burned in.

Window/level and edge enhancement

Window and level (contrast and brightness of the display) change how stored pixels look on the monitor. A narrow window can make a faint dissection flap pop; it can also clip contrast stain or a perforated blush into black. A wide window can hide a subtle lumen. Windowing does not add dose and does not change the pixels that were acquired — it changes the lookup table. Document and export a clinically honest window, not a cosmetic one that erases the complication.

Edge enhancement (spatial sharpening, unsharp masking) accentuates vessel borders. Helpful for a fine dissection line or a stent strut. The trap is noise amplification: sharpened quantum mottle and residual motion can mimic dissection or thrombus. If the flap is real, it should survive a less-enhanced window and an orthogonal projection. Do not stent a sharpening artifact.

Stenosis quantification software: adjunct, not gospel

Quantitative coronary angiography (QCA) and similar percent-stenosis tools calibrate from the catheter, draw a lumen, and output a number. Used well, they reduce “looks like 70% to me” variation. They do not replace angiographic judgment or physiology (FFR/iFR/RFR, covered later in procedures).

Why the number lies:

• Foreshortening shrinks apparent length and can understate or distort diameter stenosis.
• Single-plane views miss eccentric plaque; orthogonal runs still matter.
• Calibration off the wrong catheter French size, a foreshortened shaft, or a contrast-filled versus empty lumen error scales every measurement.
• Branch overlap and poor filling confuse edge detection.

Treat the software as an adjunct. If QCA says 40% and the patient has a pressure-wire-positive, unforeshortened 80% in another view, believe the physiology and the better projection. If QCA says 90% in a spider that is still overlapped, acquire a better angle before you commit a left-main strategy to a cartoon diameter.

3D reconstruction from rotational runs

The A.3 spin is the acquisition; A.5 is the reconstruction. The workstation builds a volume you can window as a CT-like angiogram, cut as MPRs, and measure (annulus area and perimeter for TAVR, ostial diameter and depth for LAA). Centerlines and implant-angle predictions are only as good as contrast timing, motion suspension, and completeness of the rotation. Re-window the volume the same way you re-window cine: to see anatomy, not to invent a size the operator wishes the annulus had. Store the volume and the key measurements with the case if they drove a device choice — that reconstruction is then part of the record.

Storing fluoro versus storing cine

Cine / digital acquisition / DSA series are the diagnostic pictures: high dose per frame, intended for interpretation, almost always archived. Fluoro-store (last-image-hold sequences, fluoro capture) is a lower-dose, noisier loop. Useful to document a wire perforation, a spasm that resolved, or the live roadmap. It is not automatically the complete legal angiogram.

Facility policy decides which fluoro loops are kept. Teaching default: archive every diagnostic cine/DSA run that answers a clinical question, including “normal” coronaries; keep fluoro-store when it uniquely documents an event; do not skip archiving the left-coronary cine because a pretty fluoro-store exists. Deleting a diagnostic run to save space is not a radiation-protection technique.

PACS, CVIS, and DICOM

PACS (picture archiving and communication system) is the enterprise image store. A cardiovascular information system (CVIS) is the cath-lab-aware cousin: images plus hemodynamics, procedure logs, inventory, and the structured report, often tightly linked to PACS. You will hear both names; the exam-relevant idea is a retrievable, identified, diagnostic archive, not a pile of unnamed MP4s on a thumb drive.

DICOM (Digital Imaging and Communications in Medicine) is the file and network standard: patient identifiers, study/series/instance UIDs, pixel data, and, when configured, dose objects. Sending “a JPEG on a chat app” strips that wrapper and is not an archive.

Legal record and radiation dose structured reports

The legal image record is the identified, time-stamped, diagnostically adequate series the facility retains under medical-record rules — typically the DICOM cine/DSA (and stored 3D volumes used for sizing), plus the report. If a later operator, a referring surgeon, or a court asks “what did the left main look like?”, the answer is the archived spider and cranial runs, not your memory of last-image-hold.

Radiation dose structured reports (RDSR) are DICOM objects that record fluoro time, cumulative reference-point air kerma, KAP/DAP, and often acquisition-frame counts and system technique. They feed PACS and dose-monitoring software. RDSR is how the lab knows a long CTO crossed a skin-dose threshold. It is not optional decoration; if the system generates RDSR, it belongs with the study. Teaching, not a claim that ARRT publishes a required RDSR vendor: know what it documents and that it is archived with the case.

HIPAA when burning disks and sending images

Exporting a case does not end privacy rules. Protected health information (PHI) rides on every DICOM header (name, medical record number, date of birth, accession). HIPAA expectations in this context:

• Give patients and referring physicians encrypted or otherwise controlled media when policy requires it; treat unlabeled disks as PHI, not scrap.
• Do not email unencrypted clinical images to a personal account or a vendor without a business-associate pathway.
• De-identify teaching files if they leave the covered entity.
• Track who received a disk; a “here’s your heart on a CD” handoff in the hallway still needs identity checks.
• Vendor uploads for 3D planning or research use the same PHI rules as PACS.

Processing table

Scenario and traps

An iliac DSA for large-bore TAVR access shows black-white bone edges after a breath. Pixel-shift once. If the iliac still ghosts, repeat the DSA on suspended respiration; do not wire a false lumen on a shifted mask. The same case’s 3D TAVR spin is reconstructed, measured, and stored with the cine aortogram. RDSR goes with the study. The valve-sizing screenshots emailed from a personal phone are a HIPAA problem even if the measurements were right.

Left-coronary QCA reads 45% in a foreshortened RAO. The LAO cranial is an unforeshortened 80% and iFR is positive. Software is an adjunct — treat the lesion, do not treat the 45%.

Traps: calling edge enhancement a diagnosis of dissection. Deleting cine because fluoro-store “looks fine.” Burning an unencrypted disk with the full DICOM header for a family member. Believing window/level reduced the patient’s dose. Filing RDSR in a trash folder because “dose is a physicist problem.”