Lung cancer is the third most common cancer in the UK and a leading cause of mortality.1 Stage at diagnosis is closely associated with treatment outcomes.2 However, close to half of all lung cancer cases are only diagnosed at stage 3 or 43, dramatically reducing survival rates and posing a tremendous socioeconomic burden for the UK.4
UK lung cancer facts1
The standard lung cancer pathway
The NHS introduced the National Optimal Lung Cancer Pathway (NOLCP) in 2017 to facilitate timely diagnosis and earlier treatment. It aims to ensure patients have a follow up CT within 72h of a suspicious CXR report and cancer ruled out or diagnosed within 28 days of referral for diagnostic tests.5 However, this ambition has been challenged by factors such as increasing imaging volumes, staff shortages and overall burden on the healthcare system, further aggravated by the COVID-19 pandemic.6 NHS Trusts are now looking to AI and whether it can facilitate improved turnaround times for the diagnosis and treatment of lung cancer.
NHS Epsom and St Helier Trust at the forefront of
innovation in lung cancer diagnosis
Plain radiography, including chest X-ray, accounts for nearly half of medical imaging across the NHS.7 The team at NHS Epsom and St Helier Trust (NHS ESTH) in Surrey, Southwest London, which serves a population of close to half a million people, conducts more than 65,000 chest X-rays per year. This large and growing volume of medical imaging and radiological reporting is managed by no more than 25 radiologists and 4 reporting radiographers. At NHS ESTH, chest X-rays are usually reported by reporting radiographers.
Taking a problem-led approach
An audit conducted at NHS ESHT in 2021 revealed that only 66% of patients received their CT scans within a period of 5 days from initial chest X-ray. To help support the care team with the increasing demand of medical imaging while maintaining a high standard of care, NHS ESTH explored whether a decision-support AI solution could expedite access to critical diagnostic tests, diagnoses, and radiologist reports.
Building a team for success
A multidisciplinary project team across 5 South West London hospitals, the SW London AI Group, was set up. After thorough market investigation of all relevant CE-marked AI tools Annalise Enterprise CXR was chosen for a six-month evaluation study.
The study aimed to evaluate whether Annalise Enterprise CXR improves the diagnostic lung cancer pathway by speeding time to diagnosis and report turnaround times in a real-world setting.
Tailoring the AI-enhanced lung cancer pathway
Annalise Enterprise CXR was deployed in January 2023. The comprehensive solution detects up to 124 findings, offering great clinical flexibility. NHS ESTH chose to focus on a subset of 14 findings relevant for lung cancer care. Only those findings were activated for AI analysis (see Box).
The aim was to get back on track with the standard lung cancer diagnosis timeline, using decision-support AI to facilitate same day CT examinations. In the case of lung cancer positive AI results, CXR cases were routed instantly to the duty reporter to expedite reporting. In the case of alignment between AI result and human judgement, patients had a CT scan the same day. Cases classified as unremarkable by the AI tool followed the usual diagnostic routine and process.
Joint efforts for optimised healthcare outcome
The six-month evaluation period ran between January and June 2023. Early results, present at NHS ESTH, are promising, indicating a significantly higher proportion of patients receiving either a same day CT or CT within 72 hours. Final results from the evaluation will be presented by the end of 2023.
14 lung cancer associated findings available with Annalise Enterprise CXR
- Cavitating mass
- Cavitating mass internal content
- Diffuse perihilar airspace opacity
- Hilar lymphadenopathy
- Inferior mediastinal mass
- Multiple pulmonary masses
- Pleural mass
- Single pulmonary mass
- Single pulmonary nodule
- Superior mediastinal mass
- Focal airspace opacity
- Lung collapse
- Segmental collapse
- Diffuse pleural thickening
Reducing wait times for patients
A patient came in for a chest X-ray with multiple indications: a persistent cough, chest pain, and a history of smoking and breast cancer. Prior to 2023, the patient would have been sent home to wait several days for a radiologist report and follow up CT chest appointment. With the new AI assisted workflow, the patient remained in the department while a reporting radiographer reviewed the chest X-ray in PACS. Annalise Enterprise had flagged 3 findings indicative of malignancy. Agreeing with the AI findings, the radiographer arranged a same day CT within 25 minutes of the CXR, shortening the patient’s wait time from days to minutes.
AI-supported example case
1. Original CXR (PA) 2. Annalise Viewer indicating three suspected findings including solitary lung nodule finding localised by purple overlay. 3. Follow-up CT chest examination, confirming malignancy.
“This patient received an earlier diagnosis and treatment plan, which is important for both their physical and mental wellbeing.”
Quality Manager and Reporting Radiographer NHS ESTH
“By using an AI tool, we are able to triage urgent chest x-rays, expedite reporting and offer same day CT to provide patients with a faster diagnosis.”
Supporting lung cancer diagnosis with AI
Facilitate early detection of lung cancer to improve patient outcomes.
Detect cancer more accurately
AI support has been shown to, on average, improve the detection of lung masses and nodules across different nodule sizes for AI assisted vs. unassisted radiologists.8
CT follow up faster
X-ray findings suspicious for lung cancer are followed up by a CT scan for confirmation. The sooner a confirmed diagnosis is available, the sooner patients can be treated. As an example, the UK’s NHS has established the aim of 72 hours from referral to follow up CT examination and treatment options should be discussed 28 days after diagnosis.5
Reduced treatment costs
Cancer treatment costs generally increase as cancer stages progress.4 Better survival rates are typically seen where lung cancer is detected sooner. This results in better patient outcomes and reduces the burden on the healthcare system.
Detect cancer earlier
Lung cancer survival rates are reported to be inversely proportional to stage at diagnosis.2 Detecting cancer early is one of the important variables impacting lung cancer survival.1 Annalise Enterprise CXR has been shown to enable faster detection and triage of cases with signs of lung cancer requiring further clinical investigation.9
Annalise Enterprise CXR
An assistive clinical tool for your acute care teams, when every minute counts.
Annalise Enterprise CXR uses deep learning to identify the suspected presence of up to 124 findings on chest X-rays. Identifying findings relevant to different care settings empowers healthcare providers to benefit from one AI algorithm that adds value across different care pathways.
1 Cancer Research UK, https://www.cancerresearchuk.org/, Accessed 07/08/2023.
2 Walters, S et al. Lung cancer survival and stage at diagnosis in Australia, Canada, Denmark, Norway, Sweden, and the UK: a population-based study, 2004–2007. Thorax 2013;68:551-564.
3 Cronin, KA et al. (2018), Annual Report to the Nation on the Status of Cancer, part I: National cancer statistics. Cancer, 124: 2785-2800.
4 Laudicella M et al. Cost of care for cancer patients in England: evidence from population-based patient-level data. Br J Cancer 114, 1286–1292 (2016).
5 NHS England, National Optimal Lung Cancer Pathway. https://www.england.nhs.uk/long-read/implementing-a-timed-lung-cancer-diagnostic-pathway/. Accessed 07/08/23.
6 Written evidence submitted by Professor David Baldwin (chair), on behalf of the UK Lung Cancer and Mesothelioma Clinical Expert Group (CSV0012). https://committees.parliament.uk/writtenevidence/38485/pdf/. Accessed 07/08/23.
7 NHS 75 England. Diagnostic Imaging Dataset 2021-22 Data. https://www.england.nhs.uk/statistics/statistical-work-areas/diagnostic-imaging-dataset/diagnostic-imaging-dataset-2021-22-data/. Accessed 07/08/23.
8 Seah, JCY et al. Effect of a comprehensive deep-learning model on the accuracy of chest x-ray interpretation by radiologists: a retrospective, multireader multicase study. Lancet Digit Health. 2021 Aug;3(8): e496-e506.
9 Shelmerdine, SC. Improving Radiology Turnaround Times: AI enhanced efficiencies in the Lung Cancer Pathway. Oral presentation at Intelligent Health UK, London, UK, 24/05/2023.
Information contained herein is for distribution in the US only. The device is intended for use by healthcare professionals only. The device is not intended to provide direct diagnosis. For detailed device information, including indications for use, contraindications, and warnings, please consult the user guide prior to use.
Annalise-AI Pty Ltd., Level P, 24 Campbell Street, Sydney NSW 2000, Australia