Premium
Radiofrequency spectral analysis of EBUS for peripheral pulmonary lesions
Author(s) -
Morikawa Kei,
Kurimoto Noriaki,
Inoue Takeo,
Miyazawa Teruomi,
Mineshita Masamichi
Publication year - 2019
Publication title -
respirology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.857
H-Index - 85
eISSN - 1440-1843
pISSN - 1323-7799
DOI - 10.1111/resp.13443
Subject(s) - medicine , radiology , peripheral , lung , ultrasound , pathology , pathological , lesion , nuclear medicine
Background and objective We previously reported that histogram‐based quantitative evaluation for endobronchial ultrasonography (EBUS) B‐mode images could differentiate between benign and malignant lesions. However, these images were generated from reconstructed raw radiofrequency (RF) signals and had some limitations. Currently, there are no reports on raw RF signal data to quantitatively differentiate ultrasound information for peripheral pulmonary lesions. Methods We prospectively hypothesized that RF spectral analysis from EBUS images could reveal sonographic features of peripheral pulmonary diseases. RF data were imported into a frequency spectral analysis software programme, comparing four parameters: mean frequency (MHz); slope; mid‐band fit (dB); and y‐intercept (dB), to differentiate between benign and malignant lesions. Furthermore, we compared subgroup analysis within benign and malignant lesions. Results RF data from EBUS images were obtained in 146 cases, of which, 106 lung cancers and 40 inflammatory diseases were present. Significant differences were observed for three parameters in benign and malignant lesions (mean frequency: P < 0.05, slope: P < 0.05, y‐intercept: P < 0.01) with diagnostic accuracy of 61%, 57.5%, 63%, respectively. In subgroup analysis, the acute pneumonia group showed higher mean frequency, higher slope and lower y‐intercept patterns compared to mycobacterial and fibrotic diseases ( P < 0.05). In malignant lesions, small cell carcinoma showed higher mean frequency, higher slope and lower y‐intercept pattern compared to other histopathological lung cancers ( P < 0.01). Conclusion RF analysis might be capable of demonstrating aspects of the lesion's pathological heterogeneity rather than precisely differentiating between benign and malignant lesions.