Loading
Software that reads CT and MRI scans, marks places where a cancer may be and puts the most worrying scans to the top of a radiologist's list. In large studies it has matched or beaten the average radiologist at some tasks, but a radiologist still makes the diagnosis.
This record covers AI that detects suspected cancer on cross-sectional imaging and prioritises the reading list; AI in radiology as a whole, including mammography, is on its own page. Two studies set the reference points. Ardila and colleagues (Google, Nature Medicine 2019) trained a three-dimensional deep learning model on low-dose screening CT; it reached an area under the curve of 94.4 per cent on 6,716 National Lung Screening Trial cases and, where no prior scan was available, outperformed all six radiologists in a reader study with absolute reductions of 11 per cent in false positives and 5 per cent in false negatives. With a prior scan it was on par with them.
The PI-CAI study (Saha and colleagues, Lancet Oncology 2024) trained and externally validated an AI system for clinically significant prostate cancer on 10,207 MRI examinations. On 400 paired test cases its area under the curve was 0.91 against 0.86 for a pool of 62 radiologists using PI-RADS 2.1, and at the same sensitivity it produced 50.4 per cent fewer false positives. Against the radiology readings made in routine multidisciplinary practice it was comparable, and the authors called for prospective validation. Commercial products are cleared device by device.
Deep networks trained on annotated CT or MRI volumes output a per-lesion probability; a threshold flags studies for priority reading or marks findings for the radiologist, who confirms or rejects them.
Query for this technology: (TITLE:"AI detection and triage on CT and MRI for cancer" OR ABSTRACT:"AI detection and triage on CT and MRI for cancer" OR TITLE:"AI radiology triage" OR ABSTRACT:"AI radiology triage" OR TITLE:"AI triage for oncology imaging" OR ABSTRACT:"AI triage for oncology imaging" OR TITLE:"computer-aided detection on CT and MRI" OR ABSTRACT:"computer-aided detection on CT and MRI") AND (cancer OR tumor OR tumour OR oncology OR carcinoma OR lymphoma OR leukemia OR leukaemia OR myeloma OR sarcoma OR melanoma OR glioma). Results are unfiltered search hits about AI detection and triage on CT and MRI for cancer, not a curated reading list.
Shares Aidoc, Lunit, CT (computed tomography), MRI.
Shares Aidoc, Lunit, AI in radiology, CT (computed tomography).
Shares Aidoc, Lunit, AI in radiology, CT (computed tomography).
Shares Quibim, Radiomics, AI in radiology, CT (computed tomography).
Shares Aidoc, AI in radiology, CT (computed tomography).
Shares Low-dose CT lung screening, AI in radiology, CT (computed tomography), Lung cancer (all types).
Shares Radiomics, AI in radiology.
Shares Lunit, AI in radiology.
Every source this page cites, grouped by who published it. Grouping is read from the web address, not assigned by hand.
A source that has moved or gone is reported by the weekly link check. /audit/