Nakagawa Hiroaki, Nagatani Yukihiro, Takahashi Masashi, Ogawa Emiko, Tho Nguyen Van, Ryujin Yasushi, Nagao Taishi, Nakano Yasutaka
Division of Respiratory Medicine, Department of Internal Medicine, Shiga University of Medical Science, Seta Tsukinowa-cho, Otsu, Shiga 520-2192, Japan.
Department of Radiology, Shiga University of Medical Science, Seta Tsukinowa-cho, Otsu, Shiga 520-2192, Japan.
Eur J Radiol. 2016 Jan;85(1):125-130. doi: 10.1016/j.ejrad.2015.11.011. Epub 2015 Nov 7.
The 2011 official statement of idiopathic pulmonary fibrosis (IPF) mentions that the extent of honeycombing and the worsening of fibrosis on high-resolution computed tomography (HRCT) in IPF are associated with the increased risk of mortality. However, there are few reports about the quantitative computed tomography (CT) analysis of honeycombing area. In this study, we first proposed a computer-aided method for quantitative CT analysis of honeycombing area in patients with IPF. We then evaluated the correlations between honeycombing area measured by the proposed method with that estimated by radiologists or with parameters of PFTs.
Chest HRCTs and pulmonary function tests (PFTs) of 36 IPF patients, who were diagnosed using HRCT alone, were retrospectively evaluated. Two thoracic radiologists independently estimated the honeycombing area as Identified Area (IA) and the percentage of honeycombing area to total lung area as Percent Area (PA) on 3 axial CT slices for each patient. We also developed a computer-aided method to measure the honeycombing area on CT images of those patients. The total honeycombing area as CT honeycombing area (HA) and the percentage of honeycombing area to total lung area as CT %honeycombing area (%HA) were derived from the computer-aided method for each patient.
HA derived from three CT slices was significantly correlated with IA (ρ=0.65 for Radiologist 1 and ρ=0.68 for Radiologist 2). %HA derived from three CT slices was also significantly correlated with PA (ρ=0.68 for Radiologist 1 and ρ=0.70 for Radiologist 2). HA and %HA derived from all CT slices were significantly correlated with FVC (%pred.), DLCO (%pred.), and the composite physiologic index (CPI) (HA: ρ=-0.43, ρ=-0.56, ρ=0.63 and %HA: ρ=-0.60, ρ=-0.49, ρ=0.69, respectively).
The honeycombing area measured by the proposed computer-aided method was correlated with that estimated by expert radiologists and with parameters of PFTs. This quantitative CT analysis of honeycombing area may be useful and reliable in patients with IPF.
2011年特发性肺纤维化（IPF）官方声明提到，IPF患者高分辨率计算机断层扫描（HRCT）上蜂窝状改变的程度及纤维化加重与死亡风险增加相关。然而，关于蜂窝状区域定量计算机断层扫描（CT）分析的报道较少。在本研究中，我们首次提出了一种用于IPF患者蜂窝状区域定量CT分析的计算机辅助方法。然后，我们评估了用该方法测量的蜂窝状区域与放射科医生估计的蜂窝状区域以及肺功能测试（PFT）参数之间的相关性。
回顾性评估36例仅通过HRCT诊断的IPF患者的胸部HRCT和肺功能测试（PFT）。两名胸部放射科医生分别在每位患者的3个轴向CT切片上独立估计蜂窝状区域为识别区域（IA），以及蜂窝状区域占全肺面积的百分比为面积百分比（PA）。我们还开发了一种计算机辅助方法来测量这些患者CT图像上的蜂窝状区域。通过计算机辅助方法得出每位患者的蜂窝状区域总面积即CT蜂窝状区域（HA）以及蜂窝状区域占全肺面积的百分比即CT蜂窝状区域百分比（%HA）。
从三个CT切片得出的HA与IA显著相关（放射科医生1的ρ = 0.65，放射科医生2的ρ = 0.68）。从三个CT切片得出的%HA也与PA显著相关（放射科医生1的ρ = 0.68，放射科医生2的ρ = 0.70）。从所有CT切片得出的HA和%HA与用力肺活量（FVC）（%预计值））、一氧化碳弥散量（DLCO）（%预计值）和综合生理指数（CPI）显著相关（HA分别为：ρ = -0.43，ρ = -0.56，ρ = 0.63；%HA分别为：ρ = -0.60，ρ = -0.49，ρ = 0.69）。
所提出的计算机辅助方法测量的蜂窝状区域与专家放射科医生估计的蜂窝状区域以及PFT参数相关。这种蜂窝状区域的定量CT分析对于IPF患者可能是有用且可靠的。
