Purpose
Lung nodules are a common finding on thoracic CT scans. The prevalence of lung nodules is 33% in screening trials and is similar in routine diagnostic CT [1]. The prevalence of lung cancer in incidentally detected nodules is 1.5% and 1.4% for screen-detected nodules [2].
Distinguishing benign from malignant nodules is currently based on a visual assessment by the reporting radiologist. In the UK, this is in accordance with guidelines from the British Thoracic Society (BTS) which use the Brock University/PanCan risk prediction model [2]....
Methods and materials
All scans used in the study were performed in the course of routine clinical practice and were analysed retrospectively. The project was approved by the UK NHS Health Research Authority. Application for full Research Ethics Committee approval was not required.
Patients with malignant nodules were identified by a search of locally held cancer records for stage T1 lung cancer and patients with lung metastases from other primary sites of cancer.
Patients with benign nodules were identified by a search of electronic patient records for the...
Results
200 nodules were included in the study, 100 proven to be benign and 100 proven to be malignant.
Patient demographic data shown in table 1.
[Table 1]
28 patients had a prior diagnosis of cancer within 5 years, 5 in the benign cohort and 23 in the malignant cohort.
73/200 (37%) of nodules were determined to be ‘low-risk’ for malignancy following BTS guidelines, where surveillance CT is recommended. 65 (89%) of these were proven benign on follow up and 8 (11%) were malignant.
Of these...
Conclusion
We describe the use of a radiomics software tool to determine the risk of malignancy in lung nodules used in combination with an existing strategy for risk stratification (BTS guidelines). 14/200 patients could have potentially changed management based on the outcome of the tool combined with current practice.
For patients with an existing cancer diagnosis, the management of indeterminate lung nodules is often surveillance CT, with increasing nodule size indicating that the nodule is a metastasis. The radiomics tool showed high accuracy in discriminating lung...
Personal information and conflict of interest
L. Duerden:
Nothing to disclose
M. Calhoun:
Other: CTO at RevealDx
C. K. Khine:
Nothing to disclose
Y. Bouhbib:
Nothing to disclose
J. Rodrigues:
Shareholder: Heart and Lung Imaging LTD
Consultant: NHSX
Other: Physician services, Heartflow
Speaker: Sanofi
References
1. Gould, M.K., et al., Recent Trends in the Identification of Incidental Pulmonary Nodules. Am J Respir Crit Care Med, 2015. 192(10): p. 1208-14.
2. Callister, M.E.J., et al., British Thoracic Society guidelines for the investigation and management of pulmonary nodules: accredited by NICE. Thorax, 2015. 70(Suppl 2): p. ii1.
3. Nair, A., et al., Variable radiological lung nodule evaluation leads to divergent management recommendations. 2018.
4. Lambin, P., et al., Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer,...