Modalities

What Verityn reads.

Verityn supports a focused set of modalities, each validated individually before release. A narrow scope, honestly stated, is more useful than a broad claim that outruns the evidence.

Validation approach

One modality at a time, each held to evidence.

Each modality is trained, validated, and reviewed as a distinct product. A model for chest radiography does not carry over its validation to CT head.

Separate training per modality

Datasets, architectures, and training runs are scoped to each modality. Chest radiography and CT head are distinct models with distinct evaluations.

Held-out validation on retrospective data

Each modality is evaluated on de-identified, retrospective datasets not used in training. Sensitivity, specificity, and calibration are reported per finding class.

Clinical evaluation report

Before a modality is made available, a clinical evaluation report is completed under the quality management system. Indications and limitations are stated explicitly.

Post-market surveillance

Once live, each modality is monitored via a post-market surveillance programme. Flag rates and clinician dismiss rates are reviewed on a defined schedule.

Supported modalities

What Verityn reads today.

Findings are illustrative of the categories the model is designed to detect. Per-finding performance is documented in the clinical evaluation for each modality.

Chest radiography

CR / DX

The highest-volume plain film, and where a tireless second read earns its place first.

Findings flagged

  • Pneumothorax
  • Consolidation
  • Pleural effusion
  • Nodule and mass
  • Rib fracture
Supported

Musculoskeletal radiography

CR / DX

Hairline and corner fractures are among the most common and most missed findings on plain film.

Findings flagged

  • Subtle fracture
  • Effusion
  • Dislocation
  • Foreign body
Supported

CT chest

CT

Volumetric review with localisation across slices and a per-finding confidence value.

Findings flagged

  • Pulmonary nodule
  • Pulmonary embolism signs
  • Consolidation
  • Lymphadenopathy
Supported

CT head

CT

Prioritisation matters most here. A high-confidence bleed can be moved to the top of the list.

Findings flagged

  • Intracranial haemorrhage
  • Midline shift
  • Mass effect
  • Early ischaemic change
Supported
Requesting a modality

Your caseload shapes the roadmap.

Clinical need, evidence availability, and the feasibility of rigorous validation determine what enters the roadmap. If you have a need that is not listed, contact our clinical team.

A modality request is reviewed by our clinical affairs team alongside our head of quality. We consider the following before committing to a scope:

  • Availability of a suitable retrospective dataset for training and validation
  • Clarity of the intended population and intended purpose
  • Feasibility of an independent reader study to validate performance
  • Regulatory pathway in the target jurisdiction
  • Whether the clinical need is sufficiently distinct from an existing modality
Illustrative roadmap

Modalities under evaluation.

The following modalities are in early evaluation or exploratory research. This is an illustrative roadmap, not a commitment to a release date or a clinical claim for any listed modality.

Abdominal radiography

CR / DX

Under evaluation

Mammography

MG

Under evaluation

CT abdomen and pelvis

CT

Under evaluation

Dental panoramic radiography

DX

Exploratory

Modality names are illustrative. Inclusion on this list is not a clinical claim or a commitment to availability. All modalities are subject to the same validation process before release.

Evidence

Performance is stated, not implied.

Sensitivity, specificity, and calibration figures for each supported modality are in the validation summary. Numbers on this page are illustrative. Real performance is in the documentation, under a clinical evaluation.

A modality you need. A read you can trust.

Start with chest radiography on the Pilot plan, or talk to us about your caseload and what Verityn can do for your team.