Albert Pham
Founder & CEO
Radiology resident (diagnostic imaging) in Ho Chi Minh City. Started ApiVision after seeing firsthand how reporting workload affects radiologists and patients.
Early-stage · In development · Ho Chi Minh City
AI that reads alongside Vietnam's radiologists.
We're building tools that help radiology teams flag urgent studies sooner, add a second look to every image, and draft structured reports for the radiologist to review and sign.
The context
More images, the same number of hands.
More medical imaging is being done in Vietnam every year, and the number of trained radiologists isn't keeping up. Reports are often written under time pressure, and their structure varies from one hospital to the next, and sometimes from one reader to the next. Urgent findings can end up waiting in the same queue as routine ones.
We think software can take on the repetitive work around a diagnosis (sorting, gathering context, drafting), so radiologists can spend their attention on the diagnosis itself.
How it works
Built around the radiologist's workflow
-
Study arrives
A new study comes in from the hospital's PACS, with no extra steps for staff.
-
Imaging AI flags and sorts
A detection model marks suspected findings and moves likely urgent studies up the worklist.
-
Draft report prepared
A reporting layer built on Anthropic's Claude pulls in prior studies and clinical history, applies standard criteria (Lung-RADS, BI-RADS, Fleischner) and drafts a structured report in Vietnamese and English.
-
Radiologist decides
The radiologist reviews the images and the draft, edits as needed, and signs off.
A licensed radiologist makes every diagnosis.
Who it's for
What we're designing for
These are design goals for tools that are still in development, not descriptions of a finished product.
For radiologists
Urgent studies first
A worklist ordered by suspected urgency, so time-critical cases can be read sooner.
A draft, not a blank page
Prior studies and history gathered in one place, with a structured draft ready to edit.
Consistent criteria
Classification systems applied the same way every time, and always open to the radiologist's judgment.
For hospitals
Fits existing systems
Designed to connect to the PACS a department already uses, not replace it.
Standardized reports
A consistent report structure across readers and sites, which helps audit and teaching.
Governed data
Built for de-identification and for Vietnam's personal data protection requirements.
For referring clinicians
Clearer reports
Findings, impressions and follow-up recommendations in a predictable structure.
Vietnamese and English
Bilingual reports for local care teams and cross-border consultations.
Ask about a report
A Q&A tool for questions about a signed report, with answers grounded in what the radiologist wrote.
Products
What we're building
-
In development
Reporting Assistant
Draft reports that follow published guidelines and a consistent structure, plus a Q&A tool clinicians can use to ask about a study or a report.
-
Research
Chest X-ray AI
A detection model we plan to develop with Vietnamese hospital partners, with ethics approval and on de-identified data.
-
Planned
CT & Worklist Prioritization
A later expansion to CT and urgency-based worklist sorting, including PACS integration and the path to Ministry of Health device registration.
Our principles
How we work
-
Physician in the loop
The software suggests and the radiologist decides. Nothing reaches a patient record without a physician's sign-off.
-
Privacy by design
During development we use only de-identified or synthetic data, and we design to comply with Vietnam's personal data protection law.
-
Evidence before claims
We'll publish validation results before we make any claim about performance.
-
Built for Vietnam
Reports in Vietnamese, designed around how local hospitals actually work, and developed with Vietnamese clinicians.
Team
Who's behind it
We're looking for clinical and engineering collaborators. If you work in radiology, medical imaging, or healthcare software in Vietnam, we'd like to hear from you.
Let's talk
We'd like to hear from hospitals, clinicians and early supporters who want to help shape radiology AI for Vietnam.
Ho Chi Minh City, Vietnam