Listening carefully
to the heart.
Harkheart runs an agent team across your 7T cardiac MR queue — Intake scores every study, QA motion-corrects every cine, and a Subspecialist Reader signs off on the findings automated reading still misses. Listening carefully, page only when the heart asks for it.
- Sites
- ~10 US
- Modal
- 7T cardiac
- Coverage
- 24 / 7
Three agents. One queue. Zero pages you don't need.
Each agent owns one stage and writes back to a shared audit log, so every page, override, and correction is traceable end-to-end.
Intake
Triage at the head of the queue
Scores every incoming study against acquisition parameters, referral urgency, and prior-flag severity so on-call attention lands on the right case first.
- Acquisition grading
- Severity id
- Referral text
QA
Motion-correct + image-quality gate
Applies peer-reviewed DL motion correction to every cine, then runs image-quality checks — SNR, ghosting, RVS visibility — before a study leaves the pipeline.
- DL motion-correction
- SNR / ghosting
- RVS visibility
Subspecialist Reader
Pages when findings exceed automated confidence
Frames the finding only when automated reading is short of confidence — right-ventricular, stress-perfusion, and any QC-fail case. Without that signal, no page goes out.
- Confidence-gated paging
- Subspecialty routing
- Audit trail
Score urgency. Motion-correct every cine. Run image-quality checks. File findings into the structured report.
Sign off on right-ventricular studies, stress-perfusion, and any QC-fail case Harkheart escalated.
Tell us about your 7T program.
Four fields — hospital, contact, role, weekly study volume. We onboard one 7T program at a time and reply inside two business days with a pilot scope and a quote. The agent team listens the whole time.
- We route Intake-quarantined studies through QA, and only page a subspecialty reader when the finding warrants it.
- No public waitlist spam — every reply comes from the founder with a real pilot scope.
Horizontal platforms skip this. We don't.
The clinical signal from 7T cine is exceptional — and so is the cost of a wasted scan or a missed RV finding. Harkheart stacks three things the platforms can't economically offer at this scale.
T1/T2 mapping, ECV, strain, perfusion quantification on every cine.
Motion correction recovers wall border definition on arrhythmias and breath-hold failure; structured quantification rolls into the report without a second pass through the workstation.
AI-reimbursement-aware billing
Per-study invoices carry the AI-augmented read codes the payer expects — no manual re-coding, no write-offs.
Multi-agent, shared audit log
Shared workflow between Intake, QA, and the Subspecialist Reader. Every override timestamped.
Fewer wasted scans. Faster reports. Same night-shift headcount.
Subscribe per shift, or pay per study.
Most 7T programs subscribe per shift; light-volume or pilot sites can pay per study. Intake, motion-correct QA, and Subspecialist Reader paging are bundled in.
Detailed plan tiers live on the pricing page; per-study quotes come back inside the pilot reply. Either way, the agent team is in the loop from the first study.