Interstitial lung diseases (ILDs) represent a major cause of morbidity and mortality.[1] High-resolution computed tomography (HRCT) has become critical to characterize the imaging patterns of ILD,[2, 3] but this approach remains vulnerable to inter- and intra-observer variation. To overcome human variation, automated techniques have been applied for differentiating a variety of obstructive lung diseases based on the features of a density histogram and texture analyses.[4-11] Quantitative assessment of lung parenchymal texture is important to analyze and differentiate regional diseased patterns of ILD, which would lead to differential diagnosis. Using machine learning technique of SVM (support vector machine) and CNN (convolutional neural net), characterization of local texture of lung parenchyma at HRCT is potentially useful for diagnosis and understanding various lung diseases.To address these unmet clinical needs, we have developed Aview ILD texture platform, a computerized quantitative imaging analysis (QIA) tool for differentiation and quantitative lung parenchymal analysis; thus to provide an efficient and reliable quantification for the assessment of disease subtyping and progression for ILD patients. To address these unmet clinical needs, we have developed Aview ILD texture platform, a computerized quantitative imaging analysis (QIA) tool for differentiation and quantitative lung parenchymal analysis with use of HRCT images; thus to provide an efficient and reliable quantification for the assessment of disease subtyping and progression for ILD patients.
METHODOLOGY/APPLICATIONUsing HRCT images acquired at full inspiration, two experienced radiologists marked a total of 1200 regions of interest (ROIs), including 600 ROIs that were each acquired using a GE or Siemens scanner and consisted of 100 ROIs for sub-regions that included normal and five regional pulmonary disease patterns (ground-glass opacity, consolidation, reticular opacity, emphysema, and honeycombing). We employed the CNN network with six learnable layers that consisted of four convolution layers and two fully-connected layers. The classification results were compared with the data that were calculated using a SVM, which is a representative classifier with a shallow architecture. In this way, we developed a classification tool to identify each parenchymal sub-region in chest HRCT images, which included normal and five regional pulmonary disease patterns can be classified in terms of the texture, shape, histogram analysis.Our Aview ILD texture platform includes three QIA tools:• an automated segmentation tool for segmentation of lung parenchyma with interstitial lung diseases,• an automated classification tool for parenchymal sub-regions in chest HRCT images• an automated post-processing tool for quantifying and analyzing an extent and distribution of parenchymal sub-regions.Aview lung texture platform quantifies the extent and distribution and provides structural report with minimum user interaction. Aview lung texture platform, which runs on the Aview thin client-server platform (Coreline Soft Inc.), is developed jointly by Coreline Soft Inc. and the Department of Radiology at the Asan Medical Center. CNN classification is developed by Vuno Korea Inc. The Advanced Preprocessing Server (APS) in Aview clouding platform provides computing facility for post-processing of HRCT images and delivers real-time interactive 2D/3D visualization to networked PCs running the Aview thin client application. The application shall be available commercially in late 2016.
DEMONSTRATION STRATEGYThe purpose of this demonstration is to showcase an Aview ILD texture platform for automatic differentiation of sub-regional diseased patterns of diffuse infiltrative diseases and quantitative analysis on HRCT. The educational demonstration of Aview ILD texture platform will use computer-based hands-on demonstration at RSNA. We will set up a cloud platform of Aview ILD texture platform with use of multiple computers, one for the thin client server and the other for thin-client and mobile interface. Demonstration will cover the entire workflow ranging from image acquisition protocol, automated post-processing, interactive reviewing, automated measurements, advanced analysis and structured reporting, and will select patient cases from our clinical study approved by institutional review board of Asan Medical Center, which have been anonymized in accordance with the HIPAA Privacy Rule.
REFERENCES AND PUBLICATIONS1. Eur Radiol. 2015 Aug 9
2. Medical Physics (3.761), 2013 May;40(5):051912
3. Journal of Digital Imaging(1.421), 2011 Dec;24(6):1133-40, 21311944
4. Korean J Radiol(1.766), 2009. 10(5): p. 455-63. 19721830
5. Eur Radiol(3.548). 2012 Aug 24.
6. Korean Journal of Radiology(1.766), 2011 May-Jun;12(3), 297-307. doi: 10.3348/kjr.2011.12.3.297, 21603289
7. Comput Meth Prog Bio(1.238), 2009. 93(2): p. 206-15. 19070933
8. Invest Radiol(4.670), 2008. 43(6): p. 395-402. 18496044
Sunday 12:30pm - 1:30pm
Monday 12:15pm - 1:15pm
Tuesday 12:15pm - 1:15pm
Wednesday 12:15pm - 1:15pm
Thursday 12:15pm - 1:15pm