Aidoc
Aidoc is a clinical AI platform that gives healthcare providers faster decision support across radiology, vascular, neuro, and cardiology care. It pairs specialty AI models with a workflow layer for care coordination and patient management.
Medical imaging and diagnostics tools read scans and signals, flagging findings for priority review and speeding up triage in radiology and cardiology.
Aidoc is a clinical AI platform that gives healthcare providers faster decision support across radiology, vascular, neuro, and cardiology care. It pairs specialty AI models with a workflow layer for care coordination and patient management.
Heartflow provides non-invasive, personalized cardiac testing to improve the diagnosis and management of coronary artery disease.
Lunit develops medical AI software for cancer screening and precision oncology, helping clinicians detect cancer earlier and analyze tumor biology.
Medical imaging and diagnostics AI reads images and physiological signals for two main jobs: moving suspicious cases up the reading list (triage and pre-screening), and assisting detection and quantification (lesions, volumes, degree of stenosis). Users are radiology departments, cardiology and screening programs.
Two positions must be separated first: a tool that only reorders a worklist and a tool that makes a diagnostic claim sit under different regulation. The former is usually treated as workflow software; the latter requires device clearance in most markets, and clearance is indication-specific — "this AI is approved" does not mean it can read everything.
First, prioritisation: surfacing findings such as hemorrhage, pneumothorax or large-vessel pathology earlier so urgent cases wait less. Second, detection and marking: highlighting suspicious regions and producing measurements as a second pair of eyes during read. Third, report assistance: drafting structured reports and auto-filling measurements. Fourth, quality control and follow-up: comparing prior studies and tracking lesion change.
The gains are speed and consistency: triage can genuinely change the timeline for critical cases, and quantification reduces reader variability. The limits need spelling out. Performance is indication-specific — a model that is strong on one finding says nothing about the rest. False positives produce alert fatigue, and once readers start ignoring prompts the value drops to zero. Integration is the largest engineering hurdle; connecting to PACS, reporting systems and worklists often takes longer than the model itself. Responsibility stays with the reading clinician, and reimbursement paths are frequently missing, so the hospital has to justify the spend on its own terms.
Common structures are per study, per site or per scanner license, and enterprise agreements. Research-oriented or modality-specific products are sometimes annual subscriptions.