scholarly journals Lung Segmentation on High-Resolution Computerized Tomography Images Using Deep Learning: A Preliminary Step for Radiomics Studies

2020 ◽  
Vol 6 (11) ◽  
pp. 125 ◽  
Author(s):  
Albert Comelli ◽  
Claudia Coronnello ◽  
Navdeep Dahiya ◽  
Viviana Benfante ◽  
Stefano Palmucci ◽  
...  

Background: The aim of this work is to identify an automatic, accurate, and fast deep learning segmentation approach, applied to the parenchyma, using a very small dataset of high-resolution computed tomography images of patients with idiopathic pulmonary fibrosis. In this way, we aim to enhance the methodology performed by healthcare operators in radiomics studies where operator-independent segmentation methods must be used to correctly identify the target and, consequently, the texture-based prediction model. Methods: Two deep learning models were investigated: (i) U-Net, already used in many biomedical image segmentation tasks, and (ii) E-Net, used for image segmentation tasks in self-driving cars, where hardware availability is limited and accurate segmentation is critical for user safety. Our small image dataset is composed of 42 studies of patients with idiopathic pulmonary fibrosis, of which only 32 were used for the training phase. We compared the performance of the two models in terms of the similarity of their segmentation outcome with the gold standard and in terms of their resources’ requirements. Results: E-Net can be used to obtain accurate (dice similarity coefficient = 95.90%), fast (20.32 s), and clinically acceptable segmentation of the lung region. Conclusions: We demonstrated that deep learning models can be efficiently applied to rapidly segment and quantify the parenchyma of patients with pulmonary fibrosis, without any radiologist supervision, in order to produce user-independent results.

2020 ◽  
Vol 34 (10) ◽  
pp. 13979-13980
Author(s):  
Wenxi Yu ◽  
Hua Zhou ◽  
Jonathan G. Goldin ◽  
Grace Hyun J. Kim

Domain knowledge acquired from pilot studies is important for medical diagnosis. This paper leverages the population-level domain knowledge based on the D-optimal design criterion to judiciously select CT slices that are meaningful for the disease diagnosis task. As an illustrative example, the diagnosis of idiopathic pulmonary fibrosis (IPF) among interstitial lung disease (ILD) patients is used for this work. IPF diagnosis is complicated and is subject to inter-observer variability. We aim to construct a time/memory-efficient IPF diagnosis model using high resolution computed tomography (HRCT) with domain knowledge-assisted data dimension reduction methods. Four two-dimensional convolutional neural network (2D-CNN) architectures (MobileNet, VGG16, ResNet, and DenseNet) are implemented for an automatic diagnosis of IPF among ILD patients. Axial lung CT images are acquired from five multi-center clinical trials, which sum up to 330 IPF patients and 650 non-IPF ILD patients. Model performance is evaluated using five-fold cross-validation. Depending on the model setup, MobileNet achieved satisfactory results with overall sensitivity, specificity, and accuracy greater than 90%. Further evaluation of independent datasets is underway. Based on our knowledge, this is the first work that (1) uses population-level domain knowledge with optimal design criterion in selecting CT slices and (2) focuses on patient-level IPF diagnosis.


2020 ◽  
Vol 35 (2) ◽  
pp. 115-122 ◽  
Author(s):  
Stefano Palmucci ◽  
Sebastiano E. Torrisi ◽  
Daniele Falsaperla ◽  
Alessandro Stefano ◽  
Alfredo G. Torcitto ◽  
...  

2005 ◽  
Vol 172 (4) ◽  
pp. 488-493 ◽  
Author(s):  
David A. Lynch ◽  
J. David Godwin ◽  
Sharon Safrin ◽  
Karen M. Starko ◽  
Phil Hormel ◽  
...  

Author(s):  
Gaetano Rea ◽  
Marina De Martino ◽  
Annalisa Capaccio ◽  
Pasquale Dolce ◽  
Tullio Valente ◽  
...  

Abstract Background Volumetric high-resolution computed tomography (HRCT) of the chest has recently replaced incremental CT in the diagnostic workup of idiopathic pulmonary fibrosis (IPF). Concomitantly, visual and quantitative scores have been proposed for disease extent assessment to ameliorate disease management. Purpose To compare the performance of density histograms (mean lung attenuation, skewness, and kurtosis) and visual scores, along with lung function correlations, in IPF patients submitted to incremental or volumetric thorax HRCT. Material and methods Clinical data and CT scans of 89 newly diagnosed and therapy-naive IPF patients were retrospectively evaluated. Results Forty-six incremental and 43 volumetric CT scans were reviewed. No differences of density histograms and visual scores estimates were found by comparing two HRCT techniques, with an optimal inter-operator agreement (concordance correlation coefficient >0.90 in all instances). Single-breath diffusing lung capacity for carbon monoxide (DLCOsb) was inversely related with the Best score (r = −00.416; p = 0.014), the Kazerooni fibrosis extent (r = −0.481; p = 0.004) and the mean lung attenuation (r = −0.382; p = 0.026), while a positive correlation was observed with skewness (r = 0.583; p = 0.001) and kurtosis (r = 0.543; p = 0.001) in the incremental HRCT sub-group. Similarly, in the volumetric CT sub-cohort, DLCOsb was significantly associated with skewness (r = 0.581; p = 0.007) and kurtosis (r = 0.549; p = 0.018). Correlations with visual scores were not confirmed. Forced vital capacity significantly related to all density indices independently on HRCT technique. Conclusions Density histograms and visual scores similarly perform in incremental and volumetric HRCT. Density quantification displays an optimal reproducibility and proves to be superior to visual scoring as more strongly correlated with lung function.


2019 ◽  
Vol 8 (9) ◽  
pp. 1469 ◽  
Author(s):  
Elisabetta Balestro ◽  
Elisabetta Cocconcelli ◽  
Chiara Giraudo ◽  
Roberta Polverosi ◽  
Davide Biondini ◽  
...  

Antifibrotic treatment slows down functional decline and disease progression in idiopathic pulmonary fibrosis (IPF). High-resolution computed tomography (HRCT) is useful to diagnose IPF; however, little is known about whether and to what extent HRCT changes reflect functional changes during antifibrotic therapy. The aim of this study was, therefore, to assess HRCT change over time after 1 year of treatment and to evaluate whether these changes correlate with functional decline over the same period of time. Sixty-eight IPF patients on antifibrotic treatment (i.e., pirfenidone or nintedanib) were functionally categorized as stable or progressors based on whether (or not) they had a decline in forced vital capacity (FVC) >5% predicted/year, and their HRCT were scored blindly and independently by two expert thoracic radiologists at treatment initiation (HRCT1) and after 1 year of treatment (HRCT2). Ground glass opacities (Alveolar Score, AS), reticulations (Interstitial Score, IS) and honeycombing (HC) were quantified and correlated with FVC decline between HRCT1 and HRCT2. At treatment initiation, HRCT scores were similar in both stable patients and progressors. After one year of treatment, in the entire population, AS and HC increased significantly, while IS did not. However, when stratified by the rate of functional decline, in stable patients, HC increased significantly while AS and IS did not. On the other hand, among progressors AS and HC increased significantly whereas IS did not. In the entire population, the combined score of fibrosis (IS + HC) correlated significantly with FVC decline. In conclusion, IPF patients on antifibrotic treatment exhibit different patterns of HRCT change over time based on their rate of functional decline. HRCT data should be integrated to lung function data when assessing response to antifibrotic treatment in patients with IPF.


Sign in / Sign up

Export Citation Format

Share Document