Cyst-based measurements for assessing lymphangioleiomyomatosis in computed tomography P. Lo, a) M. S. Brown, H. Kim, and H. Kim Center for Computer Vision and Imaging Biomarkers, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, California 90024 R. Argula and C. Strange Division of Pulmonary and Critical Care Medicine, Medical University of South Carolina, Charleston, South Carolina 29425 J. G. Goldin Center for Computer Vision and Imaging Biomarkers, Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, California 90024 (Received 30 July 2014; revised 2 February 2015; accepted for publication 26 February 2015; published 15 April 2015) Purpose: To investigate the efficacy of a new family of measurements made on individual pulmonary cysts extracted from computed tomography (CT) for assessing the severity of lymphangioleiomy- omatosis (LAM). Methods: CT images were analyzed using thresholding to identify a cystic region of interest from chest CT of LAM patients. Individual cysts were then extracted from the cystic region by the watershed algorithm, which separates individual cysts based on subtle edges within the cystic regions. A family of measurements were then computed, which quantify the amount, distribution, and boundary appearance of the cysts. Sequential floating feature selection was used to select a small subset of features for quantification of the severity of LAM. Adjusted R 2 from multiple linear regression and R 2 from linear regression against measurements from spirometry were used to compare the performance of our proposed measurements with currently used density based CT measurements in the literature, namely, the relative area measure and the D measure. Results: Volumetric CT data, performed at total lung capacity and residual volume, from a total of 49 subjects enrolled in the MILES trial were used in our study. Our proposed measures had adjusted R 2 ranging from 0.42 to 0.59 when regressing against the spirometry measures, with p < 0.05. For previously used density based CT measurements in the literature, the best R 2 was 0.46 (for only one instance), with the majority being lower than 0.3 or p > 0.05. Conclusions: The proposed family of CT-based cyst measurements have better correlation with spirometric measures than previously used density based CT measurements. They show potential as a sensitive tool for quantitatively assessing the severity of LAM. C 2015 American Association of Physicists in Medicine. [http://dx.doi.org/10.1118/1.4916655] Key words: lymphangioleiomyomatosis, cyst segmentation, cystic lung disease, quantitative CT, chest CT 1. INTRODUCTION Lymphangioleiomyomatosis (LAM) is a rare lung disease that predominantly affects women and often progresses to respiratory failure. 1 A hallmark of patients with LAM is the occurrence of thin-walled parenchymal cysts in the lungs visible on computed tomography (CT), as shown in the examples in Fig. 1. Current practice in the assessment of LAM on CT scans is based on visual nonquantitative assessment. Avila et al. 2 investigated the use of relative area under -950 HU (RA950) to quantitatively assess LAM severity in a CT scan. The RA950 is a measure that was originally developed and applied in the setting of emphysema and is commercially available on image analysis workstations. In LAM patients, Avila et al. 2 showed that RA950 is both more reproducible and better at predicting the percent predicted forced expiration in 1 s (FEV1%) than qualitative assessment. Another such measure that may be suitable for LAM is the D measure presented by Mishima et al., 3 which is the slope of the cumulative frequency distributions of continuous low attenuating areas obtained by simple linear regression in the log–log domain. They showed that this measure only correlates with the ratio of diffusing capacity to alveolar ventilation (DLCO/VA) and not with the ratio of forced expiratory volume in 1 s to forced vital capacity (FEV1/FVC). This may be useful for the detection of terminal airspace enlargement at an early stage of emphysema, where FEV1/FVC may still be within normal limits. Several more advanced cyst segmentation techniques have also been investigated. Schmithorst et al. 4 proposed the use of an expectation maximization algorithm to determine the threshold for quantifying cyst volume in contrast to the use of fixed threshold at -910 or -950 HU as reported in the literature. Yao et al. 5 proposed a machine learning approach 2287 Med. Phys. 42 (5), May 2015 0094-2405/2015/42(5)/2287/9/$30.00 © 2015 Am. Assoc. Phys. Med. 2287