At Bradford Teaching Hospitals NHS Foundation Trust (BTHFT), radiology colleagues have a “second set of eyes” in the form of Annalise Enterprise CXR. Team members are finding that the AI solution complements their expertise, helping to enhance diagnostic accuracy and improve patient care.
Providing outstanding care local communities
Bradford Teaching Hospitals provides inpatient and outpatient services for approximately 550,000 people living in Bradford and communities across Yorkshire, and specialist services for about 1.1 million people. Its staff of 6,500+ work across Bradford Royal Infirmary, St Luke’s Hospital, and five community sites.
With its vision to be an “outstanding provider of healthcare, research and education”, the Trust aims to be recognised for innovating and delivering exceptional patient care.
The Trust’s radiology services form part of the Yorkshire Imaging Collaborative (YIC), a collection of six NHS radiology departments that work together across West Yorkshire and Harrogate.
Diagnosing lung cancer earlier for better patient outcomes
Respiratory conditions are a leading cause of illness and premature death in the Bradford district, and lung cancer is among the key drivers of mortality—with 500 people on average dying from respiratory-related diseases in the district each year.[1]
Chest x-rays play a crucial role in the diagnostic work-up for respiratory issues. Reflecting its commitment to innovation in patient care BTHFT (as part of YIC) successfully bid to procure an AI Diagnostic Fund (AIDF) grant for an AI tool that could help accelerate detection of lung issues.
It specifically wanted a solution that could support staff, enhance accuracy and prioritise patients with pathology, thereby improving patient care.
Expanding work with a trusted AI partner
Dr Mark Kon, Clinical Radiology Lead at BTHFT explains it’s rare for reporting staff to overlook anything on a chest x-ray, but Annalise.ai’s tool acts like a “second pair of eyes”, complementing their expertise. “The AI can highlight things that it has seen and prompt the reporter to [take] a second look to make sure we’re not missing things,” he says.
“We were one of the first groups to take on Annalise.ai’s chest x-ray first as a trial, and now as an established platform for reporting x-rays,” says Dr Kon.
Anna Jowett, Advanced Radiographic Practitioner at BTHFT, says reporting staff typically look at and report on the x-ray themselves first. “And then we look at the AI tool and make sure that everything we’ve reported is the same as the AI tool.”
Supporting clinical decision-making for high-priority cases
The solution is helping them reach their goal of detecting abnormal cases faster, reading chest x-rays within minutes of them being taken.
“AI is available as soon as the x-ray is taken, so clinicians can see that image alongside the chest x-ray when they’re in the clinical setting. It can be a real-time decision-making tool,” says Ms Jowett. “The main benefit is that the AI can prioritise cases; so, cases where there are urgent findings can be prioritised to the top of the list.”
Dr Kon explains “this means patients can see a chest physician earlier, can get their CT scans earlier and can be discussed at lung cancer meetings earlier. The overall result is a quicker time to surgery or treatment.”
Facilitating accuracy and supporting staff in high-pressure situations
Preliminary studies by BTHFT back up staff observations. In a retrospective study[2] done with 300 real-life cases of chest radiographs (including normal and abnormal cases), each case was analysed using the Annalise Enterprise CXR AI algorithm. The results were then validated against subsequent CT scans and an independent reference standard review by two experienced consultant radiologists, blinded to the AI outputs.
The results showed that the solution correctly identified all 19 cases of multiple lung lesions (e.g. multiple lung metastases, miliary TB, other multi-focal infection), meaning it did not miss pathologies which have more than one lesion. Clinically, this suggests strong potential for helping detect cancer or early-onset infections. For single lung lesions, the AI correctly identified 81.5% of cases where a lesion was present and accurately confirmed the absence of a lesion in 98.8% of the scans where no lesion was found in a total of 54 cases. It also accurately detected all cases of pneumothorax, indicating it could be a reliable tool in emergency or triage settings where quick and accurate decisions are critical.
Ms Jowett notes having access to Annalise Enterprise CXR is especially helpful under stressful circumstances.
“All our practitioners have to have a high accuracy level anyway,” she explains. “But where a human can suffer from fatigue or interruptions, the AI tool is a constant. So, pairing a human and the AI together can provide better accuracy.”
Dr Kon agrees: “Not only is it useful for day-to-day reporting, but overnight in the on-call situation, acute radiographers who have just taken a chest x-ray or the junior doctors on the ward can get AI to help them diagnose important things, and they can act on those chest x-rays in the middle of the night without having to wait for a formal report.”
AI integral to radiology’s future
Dr Kon thinks AI will be increasingly integrated into imaging. “AI will help make faster diagnosis and more accurate diagnosis,” he says. “I’m sure in the long term this will help improve survival rates.”
Since deploying Annalise Enterprise CXR, BTHFT has also started using Annalise Enterprise CTB – an AI solution that can detect up to 130 findings on non-contrast CT brain studies.
[1] City of Bradford Metropolitan District Council. Respiratory Health in Bradford District. Available at https://bradford.moderngov.co.uk/documents/s40913/2023%2002%20HSCOSC_Respiratory%20health_020223_Final.pdf. Accessed 31.03.2025.
[2] Data from the Bradford Teaching Hospitals NHS Foundation Trust
