Teleradiology, PACS and RIS
RadioLens is a teleradiology platform for imaging centres and hospitals in Bangladesh. The archive that keeps the images, the bridge that moves them, the worklist that decides who reads them, the viewer they are read in and the report that comes out are one system, built by us and run by us. Bring your own radiologists, or let ours read for you.
What actually happens
In most centres here the scanner is an island. Someone types the patient’s name at the console, the images stay on the machine or leave on a burned disc, and nothing downstream knows the scan happened until a person says so.
RadioLens closes that loop in both directions. The patient the front desk registered appears on the scanner’s own screen, so the name is never retyped. The images travel by themselves into the archive. The study lands on a worklist, is assigned by rule rather than by whoever is free, and the report that comes out is attached to the same patient it started with.
The only manual step left is the one that needs a radiologist: reading the study.
Four ways it is used
The same platform carries all four, and a site can move between them without changing systems. Start with the archive and add reading later, or send everything to us from day one.
You send the studies and our radiologists report them. This is the largest use by a wide margin: around seven of every eight reports finalized in the last three months came from a reader covering many sites rather than one of a hospital's own doctors.
The hospital keeps its own radiologists and uses the platform as its archive, its worklist and its reporting system. We supply no clinical opinion at all. Plenty of sites run this way, and some run it alongside sending overflow to us.
No reporting contract and no remote reading. The router stores your studies, serves the worklist to the scanner and lets any browser on site open an image. Sold on its own, and often the first thing a centre buys.
A vendor rents the whole platform, brings their own hospitals and radiologists, and trades under their own name. Either as a tenant inside our deployment, or as a separate instance of their own.
The reading service
A teleradiology contract is judged on one thing: how long a referring doctor waits. Most of this market answers that question with an adjective. We would rather answer it with a measurement, taken from the system that did the work.
More than fifty radiologists reported through the platform in the last three months. The worklist routes a study to one of them by rule, so a report is not waiting on whoever happens to open the list. What a referring doctor sees is a report that lands while the patient is often still in the building.
Worth stating plainly: about nineteen in twenty studies here are plain X-ray, with computed tomography a real but much smaller share. This is a high-volume radiography service, and we will tell you honestly where your case mix fits before you sign anything.
The report
Reporting is usually the part where software gives up and hands everyone a locked document. If a second person needs to touch a report, they wait, or they work in a copy and somebody reconciles it afterwards.
Here the report is a shared document. Where a study belongs to a group, or where a transcriptionist is typing for a radiologist who is dictating, both people are in the same text at the same time and can see where the other one is working. The edits merge. There is no lock to wait for and no version to reconcile.
Two limits, stated rather than discovered. The live session is not the system of record: the report is saved into the information system as drafts and versions, and that is what survives. And on the live system today the pairing actually in use is a radiologist with a transcriptionist. Shared editing between radiologists works, and almost nobody works that way.
Getting the words down
Reporting volume is limited by typing, not by reading. So there are two ways to get out of the keyboard, and they work together rather than being alternatives.
You can dictate straight into the report. Where the browser can recognise speech it does, and where it cannot the audio goes to our own server instead, so it behaves the same on a machine that has no speech engine at all.
Or a transcriptionist works with you. They are attached to you, not to the hospital, and they work inside your own templates and shortcuts, with the report authored under your name and the trail recording that they typed it. Their screen follows whatever report you have open, so there is nothing to coordinate and no file to hand over. If you would rather not be shadowed, leave a voice note instead: it is tied to the study you were looking at, and it waits in their inbox.
Two boundaries worth stating. A study assigned to a group of radiologists is outside this feature completely, by design. And a call can only be recorded when everyone else on it is your own transcriptionist, so two radiologists talking cannot be recorded at all.
The information system
Two hundred and four database migrations and four hundred and seventy-four tests sit behind this. It is the largest part of the platform and the website has never mentioned it.
Artificial intelligence, stated honestly
Every radiology product claims AI now, and most of them mean one of two very different things. We would rather tell you which one we mean.
Ours is trained in-house and it works on the paperwork, not the diagnosis. It recognises what was scanned so the study files itself against the right procedure. It reads the markers burned into an image. It segments a region once a person has pointed at it. And it types while a radiologist dictates.
There is no model in this platform that detects a finding, and we have not trained one. The only one that ever went near that territory was a prototype we left switched off.
It ships disabled and each site decides. On our own platform it is switched on and it does fill in the anatomy and laterality it reads, which is the point of it. Nothing it fills in is final: a study it gets wrong lands in the reconciliation queue, where a person changes it, and that correction is what trains the next version.
The piece that goes in first
The router is a product in its own right, and for a lot of centres it is the only thing they need. It is one application on one machine at your site. It receives from the scanners, keeps the studies, serves the worklist back to the consoles, and delivers each study wherever your rules say, including to a system you already own.
A centre can run it with no teleradiology, no remote reading and no subscription to anything. And if reading studies remotely becomes useful later, the bridge is already installed, so that becomes a configuration change instead of a migration.
Straight answers