Fluid Segmentation Research Articles

Learning to combine complementary segmentation methods for fetal and 6-month infant brain MRI segmentation

Segmentation of brain structures during the pre-natal and early post-natal periods is the first step for subsequent analysis of brain development. Segmentation techniques can be roughly divided into two families. The first, which we denote as registration-based techniques, rely on initial estimates derived by registration to one (or several) templates. The second family, denoted as learning-based techniques, relate imaging (and spatial) features to their corresponding anatomical labels. Each approach has its own qualities and both are complementary to each other. In this paper, we explore two ensembling strategies, namely, stacking and cascading to combine the strengths of both families. We present experiments on segmentation of 6-month infant brains and a cohort of fetuses with isolated non-severe ventriculomegaly (INSVM). INSVM is diagnosed when ventricles are mildly enlarged and no other anomalies are apparent. Prognosis is difficult based solely on the degree of ventricular enlargement. In order to find markers for a more reliable prognosis, we use the resulting segmentations to find abnormalities in the cortical folding of INSVM fetuses. Segmentation results show that either combination strategy outperform all of the individual methods, thus demonstrating the capability of learning systematic combinations that lead to an overall improvement. In particular, the cascading strategy outperforms the ensembling one, the former one obtaining top 5, 7 and 13 results (out of 21 teams) in the segmentation of white matter, gray matter and cerebro-spinal fluid in the iSeg2017 MICCAI Segmentation Challenge. The resulting segmentations reveal that INSVM fetuses have a less convoluted cortex. This points to cortical folding abnormalities as potential markers of later neurodevelopmental outcomes.

OCT Fluid Segmentation using Graph Shortest Path and Convolutional Neural Network.

Diagnosis and monitoring of retina diseases related to pathologies such as accumulated fluid can be performed using optical coherence tomography (OCT). OCT acquires a series of 2D slices (Bscans). This work presents a fully-automated method based on graph shortest path algorithms and convolutional neural network (CNN) to segment and detect three types of fluid including sub-retinal fluid (SRF), intra-retinal fluid (IRF) and pigment epithelium detachment (PED) in OCT Bscans of subjects with age-related macular degeneration (AMD) and retinal vein occlusion (RVO) or diabetic retinopathy. The proposed method achieves an average dice coefficient of 76.44%, 92.25% and 82.14% in Cirrus, Spectralis and Topcon datasets, respectively. The effectiveness of the proposed methods was also demonstrated in segmenting fluid in OCT images from the 2017 Retouch challenge.

DRINet for Medical Image Segmentation.

Convolutional neural networks (CNNs) have revolutionized medical image analysis over the past few years. The U-Net architecture is one of the most well-known CNN architectures for semantic segmentation and has achieved remarkable successes in many different medical image segmentation applications. The U-Net architecture consists of standard convolution layers, pooling layers, and upsampling layers. These convolution layers learn representative features of input images and construct segmentations based on the features. However, the features learned by standard convolution layers are not distinctive when the differences among different categories are subtle in terms of intensity, location, shape, and size. In this paper, we propose a novel CNN architecture, called Dense-Res-Inception Net (DRINet), which addresses this challenging problem. The proposed DRINet consists of three blocks, namely a convolutional block with dense connections, a deconvolutional block with residual inception modules, and an unpooling block. Our proposed architecture outperforms the U-Net in three different challenging applications, namely multi-class segmentation of cerebrospinal fluid on brain CT images, multi-organ segmentation on abdominal CT images, and multi-class brain tumor segmentation on MR images.

Automatic brain tissue segmentation based on graph filter

BackgroundAccurate segmentation of brain tissues from magnetic resonance imaging (MRI) is of significant importance in clinical applications and neuroscience research. Accurate segmentation is challenging due to the tissue heterogeneity, which is caused by noise, bias filed and partial volume effects.MethodsTo overcome this limitation, this paper presents a novel algorithm for brain tissue segmentation based on supervoxel and graph filter. Firstly, an effective supervoxel method is employed to generate effective supervoxels for the 3D MRI image. Secondly, the supervoxels are classified into different types of tissues based on filtering of graph signals.ResultsThe performance is evaluated on the BrainWeb 18 dataset and the Internet Brain Segmentation Repository (IBSR) 18 dataset. The proposed method achieves mean dice similarity coefficient (DSC) of 0.94, 0.92 and 0.90 for the segmentation of white matter (WM), grey matter (GM) and cerebrospinal fluid (CSF) for BrainWeb 18 dataset, and mean DSC of 0.85, 0.87 and 0.57 for the segmentation of WM, GM and CSF for IBSR18 dataset.ConclusionsThe proposed approach can well discriminate different types of brain tissues from the brain MRI image, which has high potential to be applied for clinical applications.

Segmentation of gray matter, white matter, and CSF with fluid and white matter suppression using MP2RAGE

MP2RAGE can generate uniform T1 -weighted images, which have been used for brain segmentation. However, there remain concerns about carrying out fast brain segmentation. To propose an acquisition-based method for fast segmentation of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) of healthy human brains with fluid and white matter suppression (FLAWS) using MP2RAGE. Prospective. 12 volunteers (23-28 years, seven subjects; 51-62 years, five subjects). 3T/3D MPRAGE and FLAWS. The proposed method was evaluated by calculating tissue volumes and the spatial overlap with the segmentation results from FSL and SPM12. The processing time was recorded. A paired t-test was used to compare the tissue volumes of the proposed method with those from other segmentation methods RESULTS: For the 12 subjects, the tissue volume difference between the proposed and SPM12 were 3.2 ± 2.8%, 4.2 ± 2.5%, 18.2 ± 13.1% for GM, WM, and CSF, respectively. The relative difference between the proposed and FSL was over 14% for all tissue classes. The spatial overlap between the proposed and other methods were 87-94% for GM and WM and less than 80% for CSF. The GM and WM volumes of the proposed method were not significantly different from those of SPM12 using MPRAGE as the input (P = 0.5540 and P = 0.3115, respectively). The rest of the comparisons all showed significant differences between the proposed and other methods. Statistical analysis of the two subgroups yielded similar results. The mean processing time of one subject was 6.5, 185, and 165 seconds for the proposed method, FSL, and SPM12, respectively. Our method may be accurate for the segmentation of most brain structures using FLAWS. In addition, the proposed method is fast and applicable to the two distinct age ranges. 2 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2018;48:1540-1550.

Deep learning approach for the detection and quantification of intraretinal cystoid fluid in multivendor optical coherence tomography.

We developed a deep learning algorithm for the automatic segmentation and quantification of intraretinal cystoid fluid (IRC) in spectral domain optical coherence tomography (SD-OCT) volumes independent of the device used for acquisition. A cascade of neural networks was introduced to include prior information on the retinal anatomy, boosting performance significantly. The proposed algorithm approached human performance reaching an overall Dice coefficient of 0.754 ± 0.136 and an intraclass correlation coefficient of 0.936, for the task of IRC segmentation and quantification, respectively. The proposed method allows for fast quantitative IRC volume measurements that can be used to improve patient care, reduce costs, and allow fast and reliable analysis in large population studies.

Automatic Retinal Layer Segmentation of OCT Images With Central Serous Retinopathy

In this paper, an automatic method is reported for simultaneously segmenting layers and fluid in 3-D OCT retinal images of subjects suffering from central serous retinopathy. To enhance contrast between adjacent layers, multiscale bright and dark layer detection filters are proposed. Due to appearance of serous fluid or pigment epithelial detachment caused fluid, contrast between adjacent layers is often reduced, and also large morphological changes are caused. In addition, 24 features are designed for random forest classifiers. Then, 8 coarse surfaces are obtained based on the trained random forest classifiers. Finally, a hypergraph is constructed based on the smoothed image and the layer structure detection responses. A modified live wire algorithm is proposed to accurately detect surfaces between retinal layers, even though OCT images with fluids are of low contrast and layers are largely deformed. The proposed method was evaluated on 48 spectral domain OCT images with central serous retinopathy. The experimental results showed that the proposed method outperformed the state-of-art methods with regard to layers and fluid segmentation.

Solid and fluid segments within the same molecule of stratum corneum ceramide lipid.

The outer layer of the skin, stratum corneum (SC) is an efficient transport barrier and it tolerates mechanical deformation. At physiological conditions, the majority of SC lipids are solid, while the presence of a small amount of fluid lipids is considered crucial for SC barrier and material properties. Here we use solid-state and diffusion nuclear magnetic resonance to characterize the composition and molecular dynamics of the fluid lipid fraction in SC model lipids, focusing on the role of the essential SC lipid CER EOS, which is a ceramide esterified omega-hydroxy sphingosine linoleate with very long chain. We show that both rigid and mobile structures are present within the same CER EOS molecule, and that the linoleate segments undergo fast isotropic reorientation while exhibiting extraordinarily slow self-diffusion. The characterization of this unusual self-assembly in SC lipids provides deepened insight into the molecular arrangement in the SC extracellular lipid matrix and the role of CER EOS linoleate in the healthy and diseased skin.

Contribution to Intra-Retinal Fluid Segmentation in Optical Coherence Tomography by Using Automatic Personalized Thresholding

Contribution to Intra-Retinal Fluid Segmentation in Optical Coherence Tomography by Using Automatic Personalized Thresholding

Fully-automated segmentation of fluid regions in exudative age-related macular degeneration subjects: Kernel graph cut in neutrosophic domain.

A fully-automated method based on graph shortest path, graph cut and neutrosophic (NS) sets is presented for fluid segmentation in OCT volumes for exudative age related macular degeneration (EAMD) subjects. The proposed method includes three main steps: 1) The inner limiting membrane (ILM) and the retinal pigment epithelium (RPE) layers are segmented using proposed methods based on graph shortest path in NS domain. A flattened RPE boundary is calculated such that all three types of fluid regions, intra-retinal, sub-retinal and sub-RPE, are located above it. 2) Seed points for fluid (object) and tissue (background) are initialized for graph cut by the proposed automated method. 3) A new cost function is proposed in kernel space, and is minimized with max-flow/min-cut algorithms, leading to a binary segmentation. Important properties of the proposed steps are proven and quantitative performance of each step is analyzed separately. The proposed method is evaluated using a publicly available dataset referred as Optima and a local dataset from the UMN clinic. For fluid segmentation in 2D individual slices, the proposed method outperforms the previously proposed methods by 18%, 21% with respect to the dice coefficient and sensitivity, respectively, on the Optima dataset, and by 16%, 11% and 12% with respect to the dice coefficient, sensitivity and precision, respectively, on the local UMN dataset. Finally, for 3D fluid volume segmentation, the proposed method achieves true positive rate (TPR) and false positive rate (FPR) of 90% and 0.74%, respectively, with a correlation of 95% between automated and expert manual segmentations using linear regression analysis.

Automatic fetal body and amniotic fluid segmentation from fetal ultrasound images by encoder-decoder network with inner layers.

This paper explores the effectiveness of applying a deep learning based method to segment the amniotic fluid and fetal tissues in fetal ultrasound (US) images. The deeply learned model firstly encodes the input image into down scaled feature maps by convolution and pooling structures, then up-scale the feature maps to confidence maps by corresponded un-pooling and convolution layers. Additional convolution layers with 1×1 sized kernels are adopted to enhance the feature representations, which could be used to further improve the discriminative learning of our model. We effectively update the weights of the network by fine-tuning on part of the layers from a pre-trained model. By conducting experiments using clinical data, the feasibility of our proposed approach is compared and discussed. The result proves that this work achieves satisfied results for segmentation of specific anatomical structures from US images.

Computational image analysis for prognosis determination in DME

In this pilot study, we evaluated the potential of computational image analysis of optical coherence tomography (OCT) data to determine the prognosis of patients with diabetic macular edema (DME). Spectral-domain OCT scans with fully automated retinal layer segmentation and segmentation of intraretinal cystoid fluid (IRC) and subretinal fluid of 629 patients receiving anti-vascular endothelial growth factor therapy for DME in a randomized prospective clinical trial were analyzed. The results were used to define 312 potentially predictive features at three timepoints (baseline, weeks 12 and 24) for best-corrected visual acuity (BCVA) at baseline and after one year used in a random forest prediction path. Preliminarily, IRC in the outer nuclear layer in the 3-mm area around the fovea seemed to have the greatest predictive value for BCVA at baseline, and IRC and the total retinal thickness in the 3-mm area at weeks 12 and 24 for BCVA after one year. The overall model accuracy was R2=0.21/0.23 (p<0.001). The outcomes of this pilot analysis highlight the great potential of the proposed machine-learning approach for large-scale image data analysis in DME and other retinal diseases.

Supervised learning and dimension reduction techniques for quantification of retinal fluid in optical coherence tomography images.

PurposeThe purpose of the present study is to develop fast automated quantification of retinal fluid in optical coherence tomography (OCT) image sets.MethodsWe developed an image analysis pipeline tailored towards OCT images that consists of five steps for binary retinal fluid segmentation. The method is based on feature extraction, pre-segmention, dimension reduction procedures, and supervised learning tools.ResultsFluid identification using our pipeline was tested on two separate patient groups: one associated to neovascular age-related macular degeneration, the other showing diabetic macular edema. For training and evaluation purposes, retinal fluid was annotated manually in each cross-section by human expert graders of the Vienna Reading Center. Compared with the manual annotations, our pipeline yields good quantification, visually and in numbers.ConclusionsBy demonstrating good automated retinal fluid quantification, our pipeline appears useful to expert graders within their current grading processes. Owing to dimension reduction, the actual learning part is fast and requires only few training samples. Hence, it is well-suited for integration into actual manufacturer's devices, further improving segmentation by its use in daily clinical life.

Automatic Subretinal Fluid Segmentation of Retinal SD-OCT Images With Neurosensory Retinal Detachment Guided by Enface Fundus Imaging.

Accurate segmentation of neurosensory retinal detachment (NRD) associated subretinal fluid in spectral domain optical coherence tomography (SD-OCT) is vital for the assessment of central serous chorioretinopathy (CSC). A novel two-stage segmentation algorithm was proposed, guided by Enface fundus imaging. In the first stage, Enface fundus image was segmented using thickness map prior to detecting the fluid-associated abnormalities with diffuse boundaries. In the second stage, the locations of the abnormalities were used to restrict the spatial extent of the fluid region, and a fuzzy level set method with a spatial smoothness constraint was applied to subretinal fluid segmentation in the SD-OCT scans. Experimental results from 31 retinal SD-OCT volumes with CSC demonstrate that our method can achieve a true positive volume fraction (TPVF), false positive volume fraction (FPVF), and positive predicative value (PPV) of 94.3%, 0.97%, and 93.6%, respectively, for NRD regions. Our approach can also discriminate NRD-associated subretinal fluid from subretinal pigment epithelium fluid associated with pigment epithelial detachment with a TPVF, FPVF, and PPV of 93.8%, 0.40%, and 90.5%, respectively. We report a fully automatic method for the segmentation of subretinal fluid. Our method shows the potential to improve clinical therapy for CSC.

Segmentation of colon and removal of opacified fluid for virtual colonoscopy

Colorectal cancer (CRC) is the third most common type of cancer. The use of techniques such as flexible sigmoidoscopy and capsule endoscopy for the screening of colorectal cancer causes physical pain and hardship to the patients. Hence, to overcome the above disadvantages, computed tomography (CT) can be employed for the identification of polyps or growth, while screening for CRC. This proposed approach was implemented to improve the accuracy and to reduce the computation time of the accurate segmentation of the colon segments from the abdominal CT images which contain anatomical organs such as lungs, small bowels, large bowels (Colon), ribs, opacified fluid and bones. The segmentation is performed in two major steps. The first step segments the air-filled colon portions by placing suitable seed points using modified 3D seeded region growing which identify and match the similar voxels by 6-neighborhood connectivity technique. The segmentation of the opacified fluid portions is done using fuzzy connectedness approach enhanced with interval thresholding. The membership classes are defined and the voxels are categorized based on the class value. Interval thresholding is performed so that the bones and opacified fluid parts may be extracted. The bones are removed by the placement of seed points as the existence of the continuity of the bone region is more in the axial slices. The resultant image containing bones is subtracted from the threshold output to segment the opacified fluid segments in all the axial slices of a dataset. Finally, concatenation of the opacified fluid with the segmented colon is performed for the 3D rendering of the segmented colon. This method was implemented in 15 datasets downloaded from TCIA and in real-time dataset in both supine and prone position and the accuracy achieved was 98.73%.

Joint retinal layer and fluid segmentation in OCT scans of eyes with severe macular edema using unsupervised representation and auto-context.

Modern optical coherence tomography (OCT) devices used in ophthalmology acquire steadily increasing amounts of imaging data. Thus, reliable automated quantitative analysis of OCT images is considered to be of utmost importance. Current automated retinal OCT layer segmentation methods work reliably on healthy or mildly diseased retinas, but struggle with the complex interaction of the layers with fluid accumulations in macular edema. In this work, we present a fully automated 3D method which is able to segment all the retinal layers and fluid-filled regions simultaneously, exploiting their mutual interaction to improve the overall segmentation results. The machine learning based method combines unsupervised feature representation and heterogeneous spatial context with a graph-theoretic surface segmentation. The method was extensively evaluated on manual annotations of 20,000 OCT B-scans from 100 scans of patients and on a publicly available data set consisting of 110 annotated B-scans from 10 patients, all with severe macular edema, yielding an overall mean Dice coefficient of 0.76 and 0.78, respectively.

Capillary transport of two immiscible fluids in presence of electroviscous retardation.

The transport of two immiscible electrolytes through a narrow confinement whose walls bear a finite surface potential is analyzed through a lumped model by considering the influence of a regulatory self-induced axial electric field, termed as streaming potential. The presence of a surface charge on the channel walls culminates in the aqueous solutions carrying a net charge so as to make the overall system (channel and fluid) electrically neutral. The advection due to pressure driven flow or capillarity in the absence of any externally imposed electric field causes a preferential transport of net charged species. Thus, in order to render a net zero current through the system, there is an induced electric field which also retards the flow as a consequence of the force acting on the charged segments of fluid due to the streaming electric field. It is shown through a lumped model that for the situation of two distinct segments of fluids, the rate of front penetration into the capillary is strongly dependent on the relative conductivities of the two fluids. The streaming electric field evolves in accordance to the net conductivity of the channel and is responsible for dynamic changes in the retarding influence on the segments of fluid.

Droplet-Based Multivolume Digital Polymerase Chain Reaction by a Surface-Assisted Multifactor Fluid Segmentation Approach

Here we developed the surface-assisted multifactor fluid segmentation (SAMFS), an automated, fast, and flexible approach for generating a two-dimensional droplet array with tunable droplet volumes, for multivolume digital polymerase chain reaction (PCR). The SAMFS was developed based on the combination of robotic liquid handling and surface-assisted droplet generation techniques, where a continuous aqueous stream that flowed out from a capillary probe was segmented and immobilized on hydrophilic micropillars of a microchip into massive oil-covered droplets with the probe rapidly scanning over the microchip. We studied various factors affecting the droplet generation process, including micropillar top area, distance between adjacent micropillars, aqueous stream flow rate, and microchip moving speed, and demonstrated a high droplet generation throughput up to 50 droplets/s and a largest droplet volume adjusting range from 0.25 to 350 nL. The SAMFS approach was adopted to form an oil-covered array of 994 droplets with four different volumes (1.2, 6, 30, and 150 nL droplets) required for multivolume digital PCR within 8 min. The droplet array system was applied in absolute quantification of plasmid DNA under the multivolume digital PCR mode with a dynamic range spanning 4 orders of magnitude, as well as measurement of HER2 expression levels in different breast cancer cell lines. The results are consistent to those obtained by quantitative real-time PCR method, while the present one has higher precision.

A density functional theory for association of fluid molecules with a functionalized surface: fluid-wall single and double bonding

In this manuscript we extend Wertheim’s two-density formalism beyond its first order to model a system of fluid molecules with a single association site close to a planar hard wall with association sites on its surface in a density functional theory framework. The association sites of the fluid molecules are small enough that they can form only one bond, while the wall association sites are large enough to bond with more than one fluid molecule. The effects of temperature and of bulk fluid and wall site densities on the fluid density profile, extent of association, and competition between single and double bonding of fluid segments at the wall sites versus distance from the wall are presented. The theory predictions are compared with new Monte Carlo simulation results and they are in good agreement. The theory captures the surface coverage over wide ranges of temperature and bulk density by introducing the effect of steric hindrance in fluid association at a wall site.

Intraoperative Interface Fluid Dynamics and Clinical Outcomes for Intraoperative Optical Coherence Tomography–Assisted Descemet Stripping Automated Endothelial Keratoplasty From the PIONEER Study

To correlate intraoperative interface fluid dynamics during Descemet stripping automated endothelial keratoplasty (DSAEK) using intraoperative optical coherence tomography (iOCT) in the Prospective Intraoperative and Perioperative Ophthalmic Imaging with Optical Coherence Tomography (PIONEER) study with postoperative outcomes. Prospective consecutive, interventional, comparative case series. One hundred seventy-eight eyes of 173 patients undergoing DSAEK from the Cole Eye Institute, Cleveland, Ohio. Eyes that underwent DSAEK between October 2011 and March 2014 from the PIONEER intraoperative and perioperative OCT study were included. An automated interface fluid segmentation algorithm evaluated intraoperative dynamics of interface fluid before and after surgical manipulations. iOCT images were also captured at multiple intraoperative time points for 2 different DSAEK techniques, 1 that used an active air infusion system and 1 that did not. Interface fluid metrics, graft nonadherence. iOCT measurements of interface fluid after final surgical manipulations and immediately before leaving the operating room identified that total fluid volume (P= .002), largest fluid volume pocket (P= .002), max fluid area (P= .006), mean fluid thickness (P=.03), and max fluid thickness (P= .01) significantly correlated with graft nonadherence rates within the first postoperative week. After placement and optimization of intraoperative lenticle adherence, iOCT revealed a significant difference between the area, volume, and thickness of maximum fluid pockets between the 2 surgical techniques, but both techniques resulted in significant reduction of interface fluid during the procedure. Larger residual interface fluid volume, area, and thickness at the end of surgery detected with iOCT are associated with early graft nonadherence and can be quantified with an automated algorithm. iOCT imaging can successfully capture technique-dependent differences in fluid dynamics during DSAEK.