onkommon.com
Programme R4
Radiomics & Imaging AI
What do scans carry that human reading does not extract?
Coming soon
The problem
Every CT, MRI and PET contains far more quantitative information than a radiologist can report. Most of it is discarded.
Any population screening programme in India runs into the same wall: there are not enough radiologists to read the volume. Radiologist supply, not scanner supply, is the binding constraint.
The questions
- Can imaging features predict treatment response early enough to change the plan?
- Can AI triage screening imaging where radiologist capacity is the limiting factor?
- Can features be harmonised across the very varied scanner estate found in Indian practice?
What we are contributing
Harmonisation across heterogeneous scanners is an unglamorous problem that Western multi-centre studies can often sidestep and Indian programmes cannot. Solving it is a precondition for any national screening effort.
Our partner network gives access to exactly the varied, real-world scanner estate that makes the problem hard, and therefore the data that makes it solvable.
How the work is done
- Radiomic feature extraction and cross-scanner harmonisation
- Deep learning on volumetric imaging
- Response-assessment and longitudinal change modelling
What it runs on
- Partner imaging under research agreements
- Public imaging cohorts
- Matched clinical outcomes from our own cohort
What it produces
- Response-prediction models
- Screening triage tools
- Harmonisation methods for multi-centre Indian imaging
- Publications
Why it is not just science
How this changes care here
- Enables population screening programmes at scale
- Strengthens public tender bids
- A natural entry point for imaging-industry partnership
Funding fit
AI-in-health grants; public screening tenders; imaging industry collaboration.
What we are looking for
- Imaging centres willing to contribute anonymised studies
- Public health partners for screening design
- Industry collaboration on harmonisation
Deliberately empty
Publications, funding and results
Nothing appears in this section until it exists in writing. No paper in preparation, no grant under review, no result not yet published. That rule applies with particular force to a research page.