Purpose
Lung cancer has one of the poorest survival outcomes of all cancers, as most patients are diagnosed at an advanced stage when curative treatment is unavailable (1). Early diagnosis of lung cancer is the most effective way to reduce mortality through appropriate follow up (2).
The retrospective error rate among radiologic examinations is approximately 30%, with real-time errors in radiology practice averaging 3–5% of daily reads. The current gold standard for quality assurance is the re-reading of the study by another radiologist, called a “double...
Methods and materials
From June 2020 and December 2020, all CT examinations that included total or partially imaged lungs were analysed prospectively by Ferrum for the presence of nodules. A NL algorithm was then applied to determine whether a nodule was described on radiologist reporting. Studies were flagged if a potential nodule was identified on CT, but not mentioned in the report. Flagged studies were assessed by consultant radiologists as ‘documented in report’ (NL false-positive) ‘pseudo-lesion’ (MV false-positive), ‘clinically insignificant finding’ or ‘clinical significant finding’. Reports amended and...
Results
6000 CT studies were analysed by Ferrum from May 2020 until December 2020. 18 of the 201 flagged studies had clinically significant nodules, as summarised in Figure 1. Radiologists had recommended follow up in 13 of these patients. The 7 patients with record of follow-up CT examination demonstrated increase (n=3), no change (n=1), decrease (n=1), resection of nodule (n=1) or not explicitly mentioned (n=1) with respect to lung nodule size. The 3 patients with increasing missed nodule size were initially scanned for surveillance of existing...
Conclusion
This preliminary study demonstrated the potential for an automated second-read system to detect clinically significant lung nodules, albeit with a small sample size. Most missed nodules with recorded follow-up were in patients with existing cancer diagnoses, likely due to the frequency of surveillance scanning. Flagging of these clinically significant nodules changed patient care in most of our sample through follow-up monitoring, with one patient receiving surgical intervention. A future study will generate follow-up data for a larger sample size to determine further downstream effects, as...
References
Australian Institute of Health and Welfare. Lung cancer in Australia: an overview. Canberra: AIHW; 2011.
Qian F, Yang W, Chen Q, Zhang X, Han B. Screening for early stage lung cancer and its correlation with lung nodule detection. Journal of thoracic disease. 2018;10(Suppl 7):S846.
Lee CS, Nagy PG, Weaver SJ, Newman-Toker DE. Cognitive and system factors contributing to diagnostic errors in radiology. American Journal of Roentgenology. 2013;201(3):611-7.
McEnery KW, Suitor CT, Hildebrand S, Downs RL. Integration of radiologist peer review into clinical review workstation. Journal...