Reference guide to the terminology used in AI CT reporting, teleradiology, and medical imaging workflows — from DICOM and HL7 to sensitivity/specificity and verified accuracy. Each term is defined in plain language with clinical and operational context.
A service that delivers complete, ready-to-sign CT radiology reports using AI-generated report drafts reviewed by our in-house European radiologist before delivery. Distinct from AI detection tools, which produce alerts but no complete structured report. The output of AI CT reporting is a complete, ready-to-sign preliminary report — reviewed by our in-house European radiologist and formatted identically to traditional teleradiology reports.
The international standard for transmitting, storing, retrieving, and sharing medical imaging data. CT studies are stored and transmitted as DICOM files. AI CT reporting services receive studies via DICOM C-STORE — your PACS pushes the study to the service exactly as it would push to any other DICOM destination.
A set of international standards for the exchange and integration of clinical and administrative data in healthcare systems. AI CT reporting services return preliminary reports via HL7 ORU (Observation Result Unsolicited) messages to your Radiology Information System — exactly how traditional teleradiology services return reports.
Medical imaging technology providing storage and access to CT, MRI, and other radiology images from multiple source machines. Your PACS is the source system that sends CT studies to AI CT reporting services via DICOM. Compatible systems include Sectra IDS7, Fujifilm Synapse, Intelerad InteleOne, Ambra Health, Carestream Vue, and any DICOM-compliant system.
Software that manages radiology workflow, scheduling, patient data, and report distribution. AI CT reporting services deliver preliminary reports directly to your RIS via HL7, where they appear alongside any other radiology result. No workflow changes are required at the RIS level.
The electronic transmission of radiological images and data from one location to another for remote interpretation and diagnosis. Traditional teleradiology employs remote radiologists who read scans and dictate reports manually. AI CT reporting is a form of teleradiology where AI generates the complete report draft and our in-house European radiologist reviews before delivery.
A category of AI radiology products that analyze CT or MRI scans for specific high-acuity findings (e.g., intracranial hemorrhage, pulmonary embolism, aortic dissection) and send alerts to reprioritize the radiologist worklist. Detection tools do not produce a complete report — the radiologist still reads the full scan and dictates the report from scratch after seeing the alert. Designed primarily for hospital emergency departments.
A large-scale AI model trained on broad data that can be adapted to a range of downstream tasks. In AI CT reporting, a foundation model is trained on volumetric CT data to identify and characterize findings across multiple anatomical regions and pathology categories — rather than being limited to a single finding type like a narrow AI detection model.
A hierarchical vision transformer architecture particularly effective for analyzing 3D volumetric medical imaging data. Swin transformers process CT data as overlapping 3D patches rather than 2D slices, preserving spatial relationships across the full scan volume. Used in AI CT reporting foundation models for volumetric CT interpretation.
Clinical performance metrics for diagnostic tests. Sensitivity = the proportion of actual positive cases correctly identified (true positive rate). Specificity = the proportion of actual negative cases correctly identified (true negative rate). xAID achieves 95% accuracy verified by peer-reviewed studies, outperforming unaided radiologists across multiple pathology categories. Full study data is available at xaid.ai/accuracy.
The unweighted mean of F1 scores across all classes — used to evaluate classification performance when classes are imbalanced. F1 is the harmonic mean of precision and recall. A high macro F1 indicates strong performance across all pathology categories, treating each category equally regardless of how common it is in the training data.
xAID achieves 95% accuracy verified by independent peer-reviewed studies. Full methodology, study details, and results are available at xaid.ai/accuracy. No traditional teleradiology provider or AI detection tool publishes equivalent clinical evidence.
A CT radiology report that is complete, structured, and requires no further dictation or editing by the reviewing radiologist before signature. xAID achieves 95% accuracy verified by peer-reviewed studies, meaning the vast majority of reports are ready to sign as delivered. Distinct from a report draft that requires substantial radiologist editing before it can be signed.
A legally required contract under HIPAA between a covered entity (e.g., imaging center) and a business associate (e.g., AI CT reporting service) that handles protected health information (PHI). The BAA defines how PHI must be handled, secured, and reported in case of breach. xAID signs a BAA before the first study is processed — as standard, not as a formality.
A premium pricing tier charged by traditional teleradiology providers for CT studies read outside standard business hours — typically nights, weekends, and holidays. Industry-standard surcharges range from 30–100% above the base daytime per-study rate. For a center reading 500 CTs per month with 40% after-hours volume at a $60 base rate and 50% surcharge, the annual surcharge cost reaches approximately $72,000. AI CT reporting services like xAID charge a flat per-study rate 24/7 — no after-hours surcharge.
The time from a CT study being received to delivery of the completed report. ACR guideline for routine CT: 24 hours. xAID typical TAT: 2–12 hours from DICOM receipt to report delivery, 24/7. Traditional teleradiology TAT: typically 30 minutes–4 hours for STAT reads; 1–12 hours for routine reads. TAT can degrade during peak demand periods for traditional teleradiology; AI CT reporting TAT is consistent because AI processing begins immediately on receipt.
A pricing model where the customer pays per completed CT report rather than a monthly subscription, seat license, or annual enterprise contract. xAID uses per-study pricing at narrow AI rates — with no volume minimums, no setup fees, and radiologist review included in the per-study price. Traditional teleradiology pricing is also per-study, but at higher base rates and with after-hours surcharges.
An international standard for information security management systems (ISMS). ISO 27001 certification means an organization has implemented a systematic approach to managing sensitive information, including regular third-party audits. xAID is ISO 27001 certified — a standard relevant to healthcare AI vendors because it covers how patient data is handled beyond the minimum HIPAA requirements.
AI detection tools analyze CT scans for specific high-acuity findings and send alerts or reprioritize the worklist — but the radiologist still reads the full scan and dictates the complete report from scratch. AI CT reporting services (xAID) generate the complete structured CT report using AI, which our in-house European radiologist then reviews before delivery. The key difference: detection tools produce alerts; AI CT reporting services produce complete ready-to-sign preliminary reports.
DICOM integration means your PACS sends CT studies to the AI CT reporting service via the same DICOM C-STORE protocol it uses to route studies anywhere else in your network. No custom software or middleware is required. Most PACS administrators can configure a new DICOM destination in 30–60 minutes. For AI CT reporting, you add xAID as a DICOM destination; studies that meet your routing rules (body part, modality) push to xAID automatically. Preliminary reports are returned via HL7 to your RIS.
xAID achieves 95% accuracy verified by peer-reviewed studies. No traditional teleradiology provider publishes equivalent clinical accuracy data. Full study details are available at xaid.ai/accuracy.
Sensitivity is the proportion of actual positive findings that the AI correctly identifies. Specificity is the proportion of true negatives — cases without a finding — that the AI correctly classifies as negative. A high sensitivity means very few real findings are missed. A high specificity means very few false alarms. Both matter: low sensitivity means missed diagnoses; low specificity means unnecessary follow-up. xAID achieves 95% accuracy verified by peer-reviewed studies, with AI-assisted detection consistently outperforming unaided radiologists across pathology categories.
Free 5-study pilot. Send CT studies, receive preliminary reports in 2–12 hours, compare against your current workflow. No integration required to start.