Visual versus automatic ultrasound scoring of lung B-lines: reliability and consistency between systems

Abstract

Aims. To evaluate the agreement between a visual and an automatic counting system of lung B-lines by ultrasound (US) as well as to test the inter- and intra-observer reliability of both systems in patients with lung diseases.

Material and methods. We included four patients with different lung conditions. Four ultrasonographers expert in lung US blindly, independently and consecutively performed, in two rounds, a US B-mode assessment of 8 lung intercostal spaces of each patient. Each US assessment consisted of a visual and an automatic counting of B-lines.

Results. Agreement between visual and automatic counting of B-lines was good to excellent [intraclass correlation coefficient (ICC) 0.79-0.84, p<0.001]. Intra-observer reliability was good to excellent [ICC 0.62-0.99, p<0.001] except for one investigator in whom it was close to moderate for the automatic system [ICC 0.49, p<0.05]. Inter-observer reliability was excellent for both systems in both rounds [ICC 0.86-0.90, p<0.001].

Conclusions. US automatic counting was consistent with US visual counting of lung B-lines, as performed by experts in the field. Both systems showed a high intra- and interobserver reliability.