AIRA Matrix
AIRA Matrix, from AIRA Matrix Private Limited of Thane, India, sells deep learning image analysis for pathology to drug developers and laboratories. For preclinical work, AIRATox screens and triages toxicology study slides with artifact detection, anomaly scores against control groups and abnormality quantification, built on a self supervised model trained on 130,000 unlabeled whole slide images across six species from more than 40 labs; PredTox predicts in vivo organ toxicity from compound data, including SMILES, on a desktop platform; and AIRADHI is a browser based image management system with audit trails and electronic signatures for 21 CFR Part 11 work.
For clinical research, AIRAProstate detects tumor and assigns Gleason patterns, ISUP grade groups, cribriform pattern and perineural invasion on core needle biopsies and radical prostatectomies, with AIRAStrat and AIRAPredict for risk stratification and prediction, and AIRAQc checks whole slide images for processing and scanning artifacts, accepting or flagging each slide against thresholds a lab sets. Products are labeled Research Use Only unless stated otherwise, run in the cloud or on premise, and work with multiple scanners.
The algorithms reach labs through Sectra's Amplifier marketplace, Techcyte Fusion, Gestalt PathFlow and Lumea, and image analysis is also sold as a service, pay per use or by subscription. Peer reviewed studies with Johns Hopkins, Radboud University Medical Center, Experimental Pathology Laboratories and LabCorp cover its prostate grading and toxicologic pathology models. Chaith Kondragunta is chief executive.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Deep learning is the engine of most products: AIRAProstate grades prostate tissue, AIRAQc finds artifacts in whole slide images, AIRATox triages and quantifies findings in toxicology studies on a self supervised foundation model, and PredTox predicts organ toxicity from compound data with an AutoML framework.
AIRADHI, the image management system, carries value on its own: study and slide management, a viewer, peer review, reporting, audit trails and electronic signatures for GLP work. Custom AI development and image analysis sold as a service sit beside the products.
The products assist a pathologist or scientist rather than sign out. AIRAProstate runs in automatic or manual mode and is described as an assistive tool, not a replacement for pathologists; AIRATox triages slides and quantifies findings for a toxicologic pathologist to interpret; and AIRAQc accepts or flags each slide for rescanning against thresholds the lab sets for each artifact and tissue type.
What happens when a slide is wrongly accepted or a finding missed, and how often that occurs, is not described.
Training data is characterized product by product. AIRATox runs on the company's own self supervised foundation model, trained on 130,000 unlabeled whole slide images across six species, more than 200 abnormalities, several stains, scanner types and magnifications from more than 40 labs; AIRAProstate was trained on data from 7,000 patients at six centers in the US, UK, Europe and India; PredTox uses a customizable AutoML framework with federated learning support.
No model versions, release history or update practice are published, so a lab cannot tell when a model changed.
The chain is short at the model layer: the company trains its own models, including the self supervised foundation model behind AIRATox, and no outside model provider is described. On premise and desktop deployments keep data inside the customer's site.
For the cloud deployment, no hosting provider or subprocessor is named, and the image management service that receives uploaded slides is the company's own.
Peer reviewed studies stand behind the prostate and toxicology models. With Johns Hopkins, the company's grading algorithm was studied for predicting grade reclassification on active surveillance (JNCI, 2024) and metastatic outcomes after radical prostatectomy; Radboud University Medical Center included it in an evaluation of Gleason grading algorithms on outside data (Modern Pathology, 2024); and toxicologic pathology papers with Experimental Pathology Laboratories (rodent cardiomyopathy scoring, 2024) and LabCorp (spermatogenic staging in macaque testes, 2024) cover the preclinical side.
Company staff coauthor each paper, and the studies are retrospective. Product pages add up to 92% accuracy and a quadratic kappa of up to 0.96 for AIRAProstate, and 95% sensitivity with 96% specificity for AIRAQc, without the dataset behind each figure.
Product pages say patient information is encrypted and data protection follows industry standards, and PredTox runs on the desktop so compound information stays inside the customer's organization.
The only privacy policy covers the website, contacts, events and job applicants. Nothing says how long slide images and results are kept on the cloud deployment, whether customer slides train the models, or how images are de identified, and no safety handling for a wrong result is described.
No HIPAA, business associate or GDPR position is stated for the products. The privacy policy covers the website and the people who contact or apply to the company, and the clinical product FAQ says only that data protection follows industry standards.
AIRAProstate works with patient slides and medical record numbers, and the cloud deployment would hold them, so a hospital or lab in the US would need business associate terms that are not published.
No security attestation is named: no SOC 2, ISO/IEC 27001 or HITRUST, and no trust center. The site links an ISO 9001 certificate, which covers quality management rather than security, and AIRADHI is described as meeting 21 CFR Part 11 requirements with audit trails, electronic signatures and role based access.
The company offers a cloud deployment holding patient slides and study images, so the absence of an outside security audit is the open question.
Products are labeled Research Use Only unless stated otherwise, and Sectra's Amplifier marketplace lists AIRAProstate as research only; no clearance, CE IVD mark or other diagnostic authorization is named.
The clinical pages describe diagnostic use: AIRAProstate is sold under Prostate Diagnostics, for more accurate risk stratification and treatment decisions, and its FAQ says it helps ensure patients receive the most appropriate treatment plan. The preclinical products are research tools by nature, and AIRADHI supports GLP and 21 CFR Part 11 work.
Training sets are described as varied: prostate data from six centers in four regions, and toxicology images across six species, several stains, scanner types and more than 40 labs, and an outside evaluation tested the grading algorithm on data from other sites. That speaks to generalization.
No results are broken down by site, scanner, stain or patient group, and no process for testing a model before release or monitoring it afterward is published.
Results can be checked. AIRADHI keeps a complete audit trail with electronic signatures, analysis runs in automatic or manual mode, annotations and heatmaps show where a finding came from, and AIRAQc shows the artifact behind each flag.
Nothing stands behind the output: the published terms cover only the website, product accuracy figures are given as up to values without a dataset, and no warranty, remedy or error rate with a method is published.
The algorithms run inside other pathology platforms: AIRAProstate is listed on Sectra's Amplifier marketplace, AIRAProstate and AIRAQc are integrated into Techcyte Fusion, AIRAQc is offered through Gestalt PathFlow, and Lumea and Acupath partner on prostate and bladder work. AIRAProstate follows HL7 metadata standards and reviews slides by medical record number, AIRATox exposes configurable APIs for a laboratory information system, and every product is scanner agnostic.
No laboratory information system or EHR is named for writing results back, and integration documentation is not published.
Each product runs in the cloud or on premise, and PredTox is a desktop platform, so a lab can keep images and compound data inside its own walls.
For the cloud option, no hosting provider, region, residency commitment or tenancy model is stated, although the company is based in India and sells to labs in the US, UK and Europe.
Products are sold through a demo, and image analysis as a service is offered pay per use for one or two features on a limited image set or by subscription for broader work, with images uploaded to the company's image management system.
No price, unit rate or subscription tier is published.
Coverage is specified closely: preclinical toxicology studies across common laboratory species, compound toxicity prediction, whole slide image quality control on H&E and IHC, and prostate pathology on core needle biopsies and radical prostatectomies with H&E stain.
Peer reviewed work with Johns Hopkins, Radboud University Medical Center, Experimental Pathology Laboratories and LabCorp stands behind the prostate and toxicology parts. Clinical use is limited by the Research Use Only label, and no named customer reports results on the platform.
Alternatives to AIRA Matrix
Other products that do the same kind of work as AIRA Matrix, and what each one does.
Aiforia sells cloud based deep learning image analysis that pathology laboratories use for cancer diagnostics and that researchers use for preclinical work.
Visiopharm sells pathology image analysis software for diagnostics and research, with its diagnostic products packaged as APPs: deep learning algorithms that score HER2, Ki67, ER and PR and detect metastases in breast cancer, score PD-L1 in non-small cell lung cancer, and assess prostate…
Ibex sells AI applications that read H&E whole slide images for prostate, breast and gastric cancer and return their findings into laboratories' existing image management and laboratory systems, through partners including Roche, Philips, Proscia, Sectra and Fujifilm.
Indica Labs sells HALO AP Dx, an enterprise digital pathology platform that clinical laboratories use for primary diagnosis on digital slides, along with its own clinical AI and the HALO research software used by pharma and research institutes.
Deep Bio builds deep learning algorithms for cancer pathology, led by prostate.
Aiosyn sells AI that analyzes images of stained tissue, to pathology laboratories and to drug companies running animal studies.
Compared With
Each comparison carries a written verdict, the buyer conditions that favor each vendor, and a graded side by side. Pairs that cross a category boundary are grouped separately, and their verdicts state where the boundary sits rather than manufacturing a head to head.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Not published.
|
Products are sold through a demo; image analysis as a service is pay per use for small projects or by subscription. No unit rate is published. | Not published. No HIPAA or business associate position is stated. | Not published. AIRADHI comes with hands on help with installation, operational and performance qualification. | Vendor Published |
AIRA Matrix sells its products to pharmaceutical companies, CROs and pathology labs through a demo, and sells image analysis as a service either pay per use, for one or two features on a limited image set, or by subscription for broader analysis on its image management system. No price, unit rate or subscription tier is published.