Skip to main content
Fig. 6 | Cancer Imaging

Fig. 6

From: Deep learning for semi-automated unidirectional measurement of lung tumor size in CT

Fig. 6

DL algorithm’s measurement error by tumor invasion type. The measurement error was calculated by using the following formula: \(\frac{2\bullet \left(measuremen{t}_{DL} - measuremen{t}_{human}\right)}{(measuremen{t}_{DL} + measuremen{t}_{human})}\). The invasion types were classified as follows: a Parietal pleura/chest wall invasion. b Mediastinal pleural invasion. c Endobronchial invasion less than 2 cm distal to the carina. d Invasion associated with collapse (atelectasis). e Peripheral invasion surrounded by lung. f Diaphragm invasion. Box plots show the distribution of percent measurement errors stratified by invasion type. Bonferroni multiple comparison was performed, with statistical significance defined as *P < 0.05

Back to article page