Lung26 gives your radiology department an AI second pair of eyes on every low-dose chest CT: nodule detection, malignancy triage, ACR Lung-RADS categorisation, double reading and signed reports in one secure workspace.
Lung26 fits around the way your department already works. Technicians register and scan, the AI pre-reads, radiologists decide.
Add the patient with smoking history and pack-years. Lung26 checks USPSTF screening eligibility and creates an accession on the worklist.
Drop the DICOM folder into the browser. The model scans every slice, measures each candidate in millimetres and suggests a Lung-RADS category.
Accept or reject each finding, set the lobe, add anything the AI missed. A second radiologist can read independently, with discrepancies flagged.
Finalise a versioned, signed PDF with nodule images and ACR management. Amendments keep every prior version.
Lung-field segmentation and multi-scale blob detection propose candidates on every slice; a neural classifier grades each as normal, benign or malignant.
Real diameters from DICOM pixel spacing, density type from attenuation, and the ACR category with its management recommendation, all editable.
Turn on independent second reads. Lung26 enforces two different radiologists and requires a written resolution when they disagree.
STAT and urgent studies first, filters by date, reader and status, and a dashboard that surfaces high-risk studies still awaiting a report.
Every report is an immutable snapshot with the signer's name and registration number. Amendments create a new version with a reason.
Every sign-in, record view, finding change and sign-off is written to an append-only log your administrators can review and export.
The AI runs inside the radiologist's browser. Lung26 never receives your DICOM files, only the structured findings and small nodule thumbnails needed for the report.
Every plan starts with a 14-day trial: 25 AI studies, 3 seats, double reading included. Prices in INR, excluding 18% GST.
No. Lung26 is clinical decision support. It highlights candidate nodules and suggests a category, and a qualified radiologist reviews, accepts or rejects every finding and signs the report. It is not registered as a medical device with CDSCO, FDA or under EU MDR, and must not be used as a standalone diagnosis.
No installation. Radiologists open Lung26 in Chrome or Edge and load the study's DICOM folder exported from the PACS or scanner console. The analysis engine downloads once (about 30 MB) and is cached.
The classifier reached 0.86 accuracy and 0.95 AUC across normal, benign and malignant classes on a held-out LIDC-IDRI test split. Candidate detection is classical and tuned for sensitivity, so expect some false positives to reject. Performance on your scanners and population should be validated during the trial.
Paid plans continue with a per-study overage charge shown on the plan. The trial stops AI analysis at 25 studies. Re-running AI on the same study is never billed twice.
Invoices are issued monthly in arrears with GST and can be paid by NEFT, RTGS or UPI. Enterprise customers can be invoiced annually.
Tell us about your department. We'll run a live session with your radiologists and set up a pilot on your scanners.
Prefer to try it yourself? Start the free trial, no card needed.