Interpolation-split: a data-centric deep learning approach with big interpolated data to boost airway segmentation performance.

The morphology and distribution of airway tree abnormalities enable diagnosis and disease characterisation across a variety of chronic respiratory conditions. In this regard, airway segmentation plays a critical role in the production of the outline of the entire airway tree to enable estimation of disease extent and severity. Furthermore, the segmentation of a complete airway tree is challenging as the intensity, scale/size and shape of airway segments and their walls change across generations. The existing classical techniques either provide an undersegmented or oversegmented airway tree, and manual intervention is required for optimal airway tree segmentation. The recent development of deep learning methods provides a fully automatic way of segmenting airway trees; however, these methods usually require high GPU memory usage and are difficult to implement in low computational resource environments. Therefore, in this study, we propose a data-centric deep learning technique with big interpolated data, Interpolation-Split, to boost the segmentation performance of the airway tree. The proposed technique utilises interpolation and image split to improve data usefulness and quality. Then, an ensemble learning strategy is implemented to aggregate the segmented airway segments at different scales. In terms of average segmentation performance (dice similarity coefficient, DSC), our method (A) achieves 90.55%, 89.52%, and 85.80%; (B) outperforms the baseline models by 2.89%, 3.86%, and 3.87% on average; and (C) produces maximum segmentation performance gain by 14.11%, 9.28%, and 12.70% for individual cases when (1) nnU-Net with instant normalisation and leaky ReLU; (2) nnU-Net with batch normalisation and ReLU; and (3) modified dilated U-Net are used respectively. Our proposed method outperformed the state-of-the-art airway segmentation approaches. Furthermore, our proposed technique has low RAM and GPU memory usage, and it is GPU memory-efficient and highly flexible, enabling it to be deployed on any 2D deep learning model.

Automated airway quantification associates with mortality in idiopathic pulmonary fibrosis.

OBJECTIVES: The study examined whether quantified airway metrics associate with mortality in idiopathic pulmonary fibrosis (IPF). METHODS: In an observational cohort study (n = 90) of IPF patients from Ege University Hospital, an airway analysis tool AirQuant calculated median airway intersegmental tapering and segmental tortuosity across the 2nd to 6th airway generations. Intersegmental tapering measures the difference in median diameter between adjacent airway segments. Tortuosity evaluates the ratio of measured segmental length against direct end-to-end segmental length. Univariable linear regression analyses examined relationships between AirQuant variables, clinical variables, and lung function tests. Univariable and multivariable Cox proportional hazards models estimated mortality risk with the latter adjusted for patient age, gender, smoking status, antifibrotic use, CT usual interstitial pneumonia (UIP) pattern, and either forced vital capacity (FVC) or diffusion capacity of carbon monoxide (DLco) if obtained within 3 months of the CT. RESULTS: No significant collinearity existed between AirQuant variables and clinical or functional variables. On univariable Cox analyses, male gender, smoking history, no antifibrotic use, reduced DLco, reduced intersegmental tapering, and increased segmental tortuosity associated with increased risk of death. On multivariable Cox analyses (adjusted using FVC), intersegmental tapering (hazard ratio (HR) = 0.75, 95% CI = 0.66-0.85, p < 0.001) and segmental tortuosity (HR = 1.74, 95% CI = 1.22-2.47, p = 0.002) independently associated with mortality. Results were maintained with adjustment using DLco. CONCLUSIONS: AirQuant generated measures of intersegmental tapering and segmental tortuosity independently associate with mortality in IPF patients. Abnormalities in proximal airway generations, which are not typically considered to be abnormal in IPF, have prognostic value. CLINICAL RELEVANCE STATEMENT: Quantitative measurements of intersegmental tapering and segmental tortuosity, in proximal (second to sixth) generation airway segments, independently associate with mortality in IPF. Automated airway analysis can estimate disease severity, which in IPF is not restricted to the distal airway tree. KEY POINTS: • AirQuant generates measures of intersegmental tapering and segmental tortuosity. • Automated airway quantification associates with mortality in IPF independent of established measures of disease severity. • Automated airway analysis could be used to refine patient selection for therapeutic trials in IPF.

A Computationally Efficient Approach to Segmentation of the Aorta and Coronary Arteries Using Deep Learning.

Early detection and diagnosis of coronary artery disease could reduce the risk of developing a heart attack. The coronary arteries are optimally visualised using computed tomography coronary angiography (CTCA) imaging. These images are reviewed by specialist radiologists who evaluate the coronary arteries for potential narrowing. A lack of radiologists in the UK is a constraint to timely diagnosis of coronary artery disease, particularly in the acute accident and emergency department setting. The development of automated methods by which coronary artery narrowing can be identified rapidly and accurately are therefore timely. Such complex computer based tools also need to be sufficiently computationally efficient that they can run on servers typically found in hospital settings, where graphical processing units for example are unavailable. We propose a fully automatic two-dimensional Unet model to segment the aorta and coronary arteries on CTCA images. Two models are trained to segment two regions of interest, (1) the aorta and the coronary arteries or (2) the coronary arteries alone. Our method achieves 91.20% and 88.80% dice similarity coefficient accuracy on regions of interest 1 and 2 respectively. Compared with a semi-automatic segmentation method, our model performs better when segmenting the coronary arteries alone. The performance of the proposed method is comparable to existing published two-dimensional or three-dimensional deep learning models. Importantly, the algorithmic and graphical processing unit memory efficiencies are maintained such that the model can be deployed without requiring graphical processing units, and therefore can be used in a hospital setting.

A Temporal and Spatial Analysis Approach to Automated Segmentation of Microbubble Signals in Contrast-Enhanced Ultrasound Images: Application to Quantification of Active Vascular Density in Human Lower Limbs.

Contrast-enhanced ultrasound (CEUS) using microbubble contrast agents has shown great promise in visualising and quantifying active vascular density. Most existing approaches for vascular density quantification using CEUS are calculated based on image intensity and are susceptible to confounding factors and imaging artefact. Poor reproducibility is a key challenge to clinical translation. In this study, a new automated temporal and spatial signal analysis approach is developed for reproducible microbubble segmentation and quantification of contrast enhancement in human lower limbs. The approach is evaluated in vitro on phantoms and in vivo in lower limbs of healthy volunteers before and after physical exercise. In this approach, vascular density is quantified based on the relative areas microbubbles occupy instead of their image intensity. Temporal features of the CEUS image sequences are used to identify pixels that contain microbubble signals. A microbubble track density (MTD) measure, the ratio of the segmented microbubble area to the whole tissue area, is calculated as a surrogate for active capillary density. In vitro results reveal a good correlation (r2 = 0.89) between the calculated MTD measure and the known bubble concentration. For in vivo results, a significant increase (129% in average) in the MTD measure is found in lower limbs of healthy volunteers after exercise, with excellent repeatability over a series of days (intra-class correlation coefficient = 0.96). This compares to the existing state-of-the-art approach of destruction and replenishment analysis on the same patients (intra-class correlation coefficient ≤0.78). The proposed new approach shows great potential as an accurate and highly reproducible clinical tool for quantification of active vascular density.

Differential Intensity Projection for Visualisation and Quantification of Plaque Neovascularisation in Contrast-Enhanced Ultrasound Images of Carotid Arteries.

Studies have reported that intraplaque neovascularisation (IPN) is closely correlated with plaque vulnerability. In this study, a new image processing approach, differential intensity projection (DIP), was developed to visualise and quantify IPN in contrast-enhanced non-linear ultrasound image sequences of carotid arteries. DIP used the difference between the local temporal maximum and the local temporal average signals to identify bubbles against tissue non-linear artefact and noise. The total absolute and relative areas occupied by bubbles within each plaque were calculated to quantify IPN. In vitro measurements on a laboratory phantom were made, followed by in vivo measurements in which 24 contrast-enhanced non-linear ultrasound image sequences of carotid arteries from 48 patients were selected and motion corrected. The results using DIP were compared with those obtained by maximum intensity projection (MIP) and visual assessment. The results indicated that DIP can significantly reduce non-linear propagation tissue artefacts and is much more specific in detecting bubble signals than MIP, being able to reveal microbubble signals that are buried in tissue artefacts in the corresponding MIP image. A good correlation was found between microvascular area (MVA) (r = 0.83, p < 0.001)/microvascular density (r = 0.77, p < 0.001) obtained using DIP and the corresponding expert visual grades, comparing favourably to r = 0.26 and 0.23 obtained using MIP on the same data. In conclusion, the proposed method exhibits great potential in quantification of IPN in contrast-enhanced ultrasound images of carotid arteries.

Contrast enhancement of carotid adventitial vasa vasorum as a biomarker of radiation-induced atherosclerosis.

PURPOSE: Abnormal proliferation of adventitial vasa vasorum (vv) occurs early at sites of atherosclerosis and is thought to be an early biomarker of vascular damage. Contrast-enhanced ultrasound (CEUS) can detect this process. Its usefulness in irradiated arteries as a measure of accelerated atherosclerosis is unknown. This study investigates contrast intensity in carotid adventitia as an early marker of radiation-induced damage in head and neck cancer (HNC) patients. MATERIALS/METHODS: Patients with HNC treated with a wedged-pair and matched neck technique or hemi-neck radiotherapy (RT) (unirradiated side as control) at least 2years previously were included. Patients had been prescribed a dose of at least 50Gy to the neck. CEUS was performed on both carotid arteries and a region of interest was selected in the adventitia of the far wall of both left and right distal common carotid arteries. Novel quantification software was used to compare the average intensity per pixel between irradiated and unirradiated arteries. RESULTS: 48 patients (34 males) with median age of 59.2years (interquartile range (IQR) 49.2-64.2) were included. The mean maximum point dose to the irradiated artery was 61.2Gy (IQR 52.6-61.8) and 1.1Gy (IQR 1.0-1.8Gy) to the unirradiated side. The median interval from RT was 59.4months (IQR 41-88.7). There was a significant difference in the mean (SD) contrast intensity per pixel on the irradiated side (1.1 (0.4)) versus 0.96 (0.34) on the unirradiated side (p=0.01). After attenuation correction, the difference in mean contrast intensity per pixel was still significant (1.4 (0.58) versus 1.2 (0.47) (p=0.02). Previous surgery or chemotherapy had no effect on the difference in contrast intensity between the 2 sides of the neck. Mean intensity per pixel did not correlate to traditional risk prediction models (carotid intima-medial thickness, QSTROKE score). CONCLUSIONS: Proliferation of vv is demonstrated by increased contrast intensity in irradiated carotid arteries. This may be a useful, independent biomarker of radiation-induced carotid atherosclerosis when used as a tool to quantify neovascularization.

Attenuation Correction and Normalisation for Quantification of Contrast Enhancement in Ultrasound Images of Carotid Arteries.

An automated attenuation correction and normalisation algorithm was developed to improve the quantification of contrast enhancement in ultrasound images of carotid arteries. The algorithm first corrects attenuation artefact and normalises intensity within the contrast agent-filled lumen and then extends the correction and normalisation to regions beyond the lumen. The algorithm was first validated on phantoms consisting of contrast agent-filled vessels embedded in tissue-mimicking materials of known attenuation. It was subsequently applied to in vivo contrast-enhanced ultrasound (CEUS) images of human carotid arteries. Both in vitro and in vivo results indicated significant reduction in the shadowing artefact and improved homogeneity within the carotid lumens after the correction. The error in quantification of microbubble contrast enhancement caused by attenuation on phantoms was reduced from 55% to 5% on average. In conclusion, the proposed method exhibited great potential in reducing attenuation artefact and improving quantification in contrast-enhanced ultrasound of carotid arteries.

Efficacy of magnetic resonance diffusion tensor imaging and three-dimensional fiber tractography in the detection of clinical manifestations of central nervous system lupus.

Systemic lupus erythematosus (SLE) is an autoimmune disease frequently associated with neuropsychiatric manifestations. No follow-up case report has characterized white matter alterations in patients with neuropsychiatric lupus erythematosus (NPSLE) before and after treatment. In this study, a 16-year-old NPSLE patient with severe neuropsychological symptoms was treated with steroid pulse therapy, and was scanned with conventional magnetic resonance (MR) and diffusion tensor imaging (DTI) at onset and 17months after treatment. Conventional MR images showed diffuse brain atrophy and focal vasogenic edema in the putamen, but they did not reveal abnormalities in the corpus callosum. Region-of-interest analysis of DTI images showed that fractional anisotropy and fiber tracts increased significantly, while axial diffusivity, radial, and mean diffusivity decreased significantly in the corpus callosum after treatment. The results indicated that the vasogenic edema was present in the corpus callosum at onset and was significantly reduced after treatment. These changes were generally compatible with the patient's clinical manifestations. Hence, we concluded that MR-DTI and fiber tractography are helpful to reveal the relationship between white matter alterations and neurological dysfunctions in NPSLE patients.

Primary leptomeningeal melanoma.

Primary melanoma of the central nervous system is a rare melanocytic tumor typically located in the leptomeninges. We report a 57-year-old woman with an intracranial leptomeningeal melanoma who presented with myoclonic seizures. Brain CT scan and MRI revealed a hemorrhagic intracranial tumor. The tumor was completely removed and leptomeningeal melanoma was proven pathologically. Follow-up imaging studies up to 19 months showed no recurrence of the disease. Here we present radiological, gross, and pathological images of leptomeningeal melanoma, discuss its characteristics, and review the relevant literature.