Thoracic CT‐MRI coregistration for regional pulmonary structure–function measurements of obstructive lung disease

Abstract

For a diverse group of patients with COPD and asthma, whole lung and segmental VDP was measured using an automated lung image analysis pipeline which provides a way to incorporate lung functional biomarkers into clinical research and patient care.

Accelerated 129Xe MRI morphometry of terminal airspace enlargement: Feasibility in volunteers and those with alpha‐1 antitrypsin deficiency

“…Currently, we are in the process of optimizing a convex‐optimization based deformable registration approach to address this issue. However, a convex optimization‐based registration approach led to changes in background noise that meant that the SEM‐based fitting could not be used. As a result, all ADC and morphometry values generated from the data sets corrected with the convex optimization‐based registration approach were significantly underestimated.…”

A framework for Fourier‐decomposition free‐breathing pulmonary 1H MRI ventilation measurements

“…In addition, we manually identified 3–6 fiducial points (e.g., apex/bottom and bright vessels in the lung) in a fixed image and the corresponding points in the moving images (6–12 images between the end‐inhalation and end‐exhalation phases) for each subject, resulting in a total of 90 fiducials for the 20 subjects. 1 H MRI lung registration accuracy was quantified by calculating the target‐registration‐error (TRE) between these fiducial points under the estimated displacement field . DSC, MAD, and TRE of lung segmentation and registrations were calculated for each subject, and the mean runtime for each individual component was recorded.…”

“…1 H MRI lung registration accuracy was quantified by calculating the target-registration-error (TRE) between these fiducial points under the estimated displacement field. 37 DSC, MAD, and TRE of lung segmentation and registrations were calculated for each subject, and the mean runtime for each individual component was recorded. 3 He MRI whole lung static ventilation images were also segmented using k-means clustering to classify signal intensities into 5 clusters with cluster C1 representing 3 He ventilation defects.…”

Development of a pulmonary imaging biomarker pipeline for phenotyping of chronic lung disease

“…As shown in schematic in Fig. 2, image processing and analysis modules were integrated within the GUI that were previously developed and validated across a wide range of pulmonary abnormalities 19,21,[40][41][42][43][44] by our group using C++/CUDA (CUDA v6.0, NVIDIA Crop., Santa Clara, California) and implemented for Windows operating systems (Microsoft Corporation, Redmond, Washington). The derived imaging measurements are displayed with the GUI, presented to end users, saved to hard drive (DICOM images and/or numerical values in spreadsheet files), and sent back to the PACS server for archiving.…”

“…50 Free-breathing 1 H MRI series were simultaneously segmented using a coupled Potts model 49 by employing the inherent similarity of lung segmentation between adjacent slices. The segmented lung series were registered together using a coarse-to-fine deformable registration framework 42 that employed a modality-independent-neighborhood descriptor 51 and total variation of the displacement field for registration regularization. Pairwise registrations were implemented in parallel and the registered lung volumes were used for Fourier analysis; Fourier spectrum magnitudes at the respiratory and cardiac frequencies (determined using respiratory bellow data) were used to generate ventilation and perfusion maps.…”