Segmentation of image ensembles via latent atlases

Spatial priors, such as probabilistic atlases, play an important role in MRI segmentation. However, the availability of comprehensive, reliable and suitable manual segmentations for atlas construction is limited. We therefore propose a method for joint segmentation of corresponding regions of interest in a collection of aligned images that does not require labeled training data. Instead, a latent atlas, initialized by at most a single manual segmentation, is inferred from the evolving segmentations of the ensemble. The algorithm is based on probabilistic principles but is solved using partial differential equations (PDEs) and energy minimization criteria. We evaluate the method on two datasets, segmenting subcortical and cortical structures in a multi-subject study and extracting brain tumors in a single-subject multi-modal longitudinal experiment. We compare the segmentation results to manual segmentations, when those exist, and to the results of a state-of-the-art atlas-based segmentation method. The quality of the results supports the latent atlas as a promising alternative when existing atlases are not compatible with the images to be segmented.     

Neonatal brain image segmentation in longitudinal MRI studies

In the study of early brain development, tissue segmentation of neonatal brain MR images remains challenging because of the insufficient image quality due to the properties of developing tissues. Among various brain tissue segmentation algorithms, atlas-based brain image segmentation can potentially achieve good segmentation results on neonatal brain images. However, their performances rely on both the quality of the atlas and the spatial correspondence between the atlas and the to-be-segmented image. Moreover, it is difficult to build a population atlas for neonates due to the requirement of a large set of tissue-segmented neonatal brain images. To combat these obstacles, we present a longitudinal neonatal brain image segmentation framework by taking advantage of the longitudinal data acquired at late time-point to build a subject-specific tissue probabilistic atlas. Specifically, tissue segmentation of the neonatal brain is formulated as two iterative steps of bias correction and probabilistic-atlas-based tissue segmentation, along with the longitudinal atlas reconstructed by the late time image of the same subject. The proposed method has been evaluated qualitatively through visual inspection and quantitatively by comparing with manual delineations and two population-atlas-based segmentation methods. Experimental results show that the utilization of a subject-specific probabilistic atlas can substantially improve tissue segmentation of neonatal brain images.     

DCAN: Deep contour-aware networks for object instance segmentation from histology images

In histopathological image analysis, the morphology of histological structures, such as glands and nuclei, has been routinely adopted by pathologists to assess the malignancy degree of adenocarcinomas. Accurate detection and segmentation of these objects of interest from histology images is an essential prerequisite to obtain reliable morphological statistics for quantitative diagnosis. While manual annotation is error-prone, time-consuming and operator-dependant, automated detection and segmentation of objects of interest from histology images can be very challenging due to the large appearance variation, existence of strong mimics, and serious degeneration of histological structures. In order to meet these challenges, we propose a novel deep contour-aware network (DCAN) under a unified multi-task learning framework for more accurate detection and segmentation. In the proposed network, multi-level contextual features are explored based on an end-to-end fully convolutional network (FCN) to deal with the large appearance variation. We further propose to employ an auxiliary supervision mechanism to overcome the problem of vanishing gradients when training such a deep network. More importantly, our network can not only output accurate probability maps of histological objects, but also depict clear contours simultaneously for separating clustered object instances, which further boosts the segmentation performance. Our method ranked the first in two histological object segmentation challenges, including 2015 MICCAI Gland Segmentation Challenge and 2015 MICCAI Nuclei Segmentation Challenge. Extensive experiments on these two challenging datasets demonstrate the superior performance of our method, surpassing all the other methods by a significant margin.     

SRBF: Speckle Reducing Bilateral Filtering

Speckle noise negatively affects medical ultrasound image shape interpretation and boundary detection. Speckle removal filters are widely used to selectively remove speckle noise without destroying important image features to enhance object boundaries. In this article, a fully automatic bilateral filter tailored to ultrasound images is proposed. The edge preservation property is obtained by embedding noise statistics in the filter framework. Consequently, the filter is able to tackle the multiplicative behavior modulating the smoothing strength with respect to local statistics. The in silico experiments clearly showed that the speckle reducing bilateral filter (SRBF) has superior performances to most of the state of the art filtering methods. The filter is tested on 50 in vivo US images and its influence on a segmentation task is quantified. The results using SRBF filtered data sets show a superior performance to using oriented anisotropic diffusion filtered images. This improvement is due to the adaptive support of SRBF and the embedded noise statistics, yielding a more homogeneous smoothing. SRBF results in a fully automatic, fast and flexible algorithm potentially suitable in wide ranges of speckle noise sizes, for different medical applications (IVUS, B-mode, 3-D matrix array US). (E-mail: balocco.simone@gmail.com)     

Three-dimensional segmentation of the left ventricle in late gadolinium enhanced MR images of chronic infarction combining long- and short-axis information

Automatic segmentation of the left ventricle (LV) in late gadolinium enhanced (LGE) cardiac MR (CMR) images is difficult due to the intensity heterogeneity arising from accumulation of contrast agent in infarcted myocardium. In this paper, we present a comprehensive framework for automatic 3D segmentation of the LV in LGE CMR images. Given myocardial contours in cine images as a priori knowledge, the framework initially propagates the a priori segmentation from cine to LGE images via 2D translational registration. Two meshes representing respectively endocardial and epicardial surfaces are then constructed with the propagated contours. After construction, the two meshes are deformed towards the myocardial edge points detected in both short-axis and long-axis LGE images in a unified 3D coordinate system. Taking into account the intensity characteristics of the LV in LGE images, we propose a novel parametric model of the LV for consistent myocardial edge points detection regardless of pathological status of the myocardium (infarcted or healthy) and of the type of the LGE images (short-axis or long-axis). We have evaluated the proposed framework with 21 sets of real patient and four sets of simulated phantom data. Both distance- and region-based performance metrics confirm the observation that the framework can generate accurate and reliable results for myocardial segmentation of LGE images. We have also tested the robustness of the framework with respect to varied a priori segmentation in both practical and simulated settings. Experimental results show that the proposed framework can greatly compensate variations in the given a priori knowledge and consistently produce accurate segmentations.     

Unsupervised quality control of segmentations based on a smoothness and intensity probabilistic model

Monitoring the quality of image segmentation is key to many clinical applications. This quality assessment can be carried out by a human expert when the number of cases is limited. However, it becomes onerous when dealing with large image databases, so partial automation of this process is preferable. Previous works have proposed both supervised and unsupervised methods for the automated control of image segmentations. The former assume the availability of a subset of trusted segmented images on which supervised learning is performed, while the latter does not. In this paper, we introduce a novel unsupervised approach for quality assessment of segmented images based on a generic probabilistic model. Quality estimates are produced by comparing each segmentation with the output of a probabilistic segmentation model that relies on intensity and smoothness assumptions. Ranking cases with respect to these two assumptions allows the most challenging cases in a dataset to be detected. Furthermore, unlike prior work, our approach enables possible segmentation errors to be localized within an image. The proposed generic probabilistic segmentation method combines intensity mixture distributions with spatial regularization prior models whose parameters are estimated with variational Bayesian techniques. We introduce a novel smoothness prior based on the penalization of the derivatives of label maps which allows an automatic estimation of its hyperparameter in a fully data-driven way. Extensive evaluation of quality control on medical and COCO datasets is conducted, showing the ability to isolate atypical segmentations automatically and to predict, in some cases, the performance of segmentation algorithms.     

Bone labelling on micro-magnetic resonance images

The study of trabecular bone is receiving increasing interest within the medical community working in the field of skeletal diseases, such as osteoporosis. Quantification of trabecular bone structure usually requires as a starting point a correct segmentation of the trabecular network. We have developed a probabilistic relaxation labelling technique, which uses local features of the trabecular bone images to improve segmentation. Tests on synthetic images show that bone labelling performs a more accurate segmentation than conventional techniques such as thresholding, especially by preserving the connectivity of the trabecular network. Tests on acquired data show that porosity values obtained after segmentation are in good agreement with experimental values obtained by weighing the bone samples.     

Modeling envelope statistics of blood and myocardium for segmentation of echocardiographic images

The objective of this study was to investigate the use of speckle statistics as a preprocessing step for segmentation of the myocardium in echocardiographic images. Three-dimensional (3D) and biplane image sequences of the left ventricle of two healthy children and one dog (beagle) were acquired. Pixel-based speckle statistics of manually segmented blood and myocardial regions were investigated by fitting various probability density functions (pdf). The statistics of heart muscle and blood could both be optimally modeled by a K-pdf or Gamma-pdf (Kolmogorov-Smirnov goodness-of-fit test). Scale and shape parameters of both distributions could differentiate between blood and myocardium. Local estimation of these parameters was used to obtain parametric images, where window size was related to speckle size (5 x 2 speckles). Moment-based and maximum-likelihood estimators were used. Scale parameters were still able to differentiate blood from myocardium; however, smoothing of edges of anatomical structures occurred. Estimation of the shape parameter required a larger window size, leading to unacceptable blurring. Using these parameters as an input for segmentation resulted in unreliable segmentation. Adaptive mean squares filtering was then introduced using the moment-based scale parameter (sigma(2)/mu) of the Gamma-pdf to automatically steer the two-dimensional (2D) local filtering process. This method adequately preserved sharpness of the edges. In conclusion, a trade-off between preservation of sharpness of edges and goodness-of-fit when estimating local shape and scale parameters is evident for parametric images. For this reason, adaptive filtering outperforms parametric imaging for the segmentation of echocardiographic images.     

Unifying framework for multimodal brain MRI segmentation based on Hidden Markov Chains

In the frame of 3D medical imaging, accurate segmentation of multimodal brain MR images is of interest for many brain disorders. However, due to several factors such as noise, imaging artifacts, intrinsic tissue variation and partial volume effects, tissue classification remains a challenging task. In this paper, we present a unifying framework for unsupervised segmentation of multimodal brain MR images including partial volume effect, bias field correction, and information given by a probabilistic atlas. Here-proposed method takes into account neighborhood information using a Hidden Markov Chain (HMC) model. Due to the limited resolution of imaging devices, voxels may be composed of a mixture of different tissue types, this partial volume effect is included to achieve an accurate segmentation of brain tissues. Instead of assigning each voxel to a single tissue class (i.e., hard classification), we compute the relative amount of each pure tissue class in each voxel (mixture estimation). Further, a bias field estimation step is added to the proposed algorithm to correct intensity inhomogeneities. Furthermore, atlas priors were incorporated using probabilistic brain atlas containing prior expectations about the spatial localization of different tissue classes. This atlas is considered as a complementary sensor and the proposed method is extended to multimodal brain MRI without any user-tunable parameter (unsupervised algorithm). To validate this new unifying framework, we present experimental results on both synthetic and real brain images, for which the ground truth is available. Comparison with other often used techniques demonstrates the accuracy and the robustness of this new Markovian segmentation scheme.     

FUN-SIS: A Fully UNsupervised approach for Surgical Instrument Segmentation

Automatic surgical instrument segmentation of endoscopic images is a crucial building block of many computer-assistance applications for minimally invasive surgery. So far, state-of-the-art approaches completely rely on the availability of a ground-truth supervision signal, obtained via manual annotation, thus expensive to collect at large scale. In this paper, we present FUN-SIS, a Fully-UNsupervised approach for binary Surgical Instrument Segmentation. FUN-SIS trains a per-frame segmentation model on completely unlabelled endoscopic videos, by solely relying on implicit motion information and instrument shape-priors. We define shape-priors as realistic segmentation masks of the instruments, not necessarily coming from the same dataset/domain as the videos. The shape-priors can be collected in various and convenient ways, such as recycling existing annotations from other datasets. We leverage them as part of a novel generative-adversarial approach, allowing to perform unsupervised instrument segmentation of optical-flow images during training. We then use the obtained instrument masks as pseudo-labels in order to train a per-frame segmentation model; to this aim, we develop a learning-from-noisy-labels architecture, designed to extract a clean supervision signal from these pseudo-labels, leveraging their peculiar noise properties. We validate the proposed contributions on three surgical datasets, including the MICCAI 2017 EndoVis Robotic Instrument Segmentation Challenge dataset. The obtained fully-unsupervised results for surgical instrument segmentation are almost on par with the ones of fully-supervised state-of-the-art approaches. This suggests the tremendous potential of the proposed method to leverage the great amount of unlabelled data produced in the context of minimally invasive surgery.     

Segmentation of 4D cardiac MR images using a probabilistic atlas and the EM algorithm

In this paper an automatic atlas-based segmentation algorithm for 4D cardiac MR images is proposed. The algorithm is based on the 4D extension of the expectation maximisation (EM) algorithm. The EM algorithm uses a 4D probabilistic cardiac atlas to estimate the initial model parameters and to integrate a priori information into the classification process. The probabilistic cardiac atlas has been constructed from the manual segmentations of 3D cardiac image sequences of 14 healthy volunteers. It provides space and time-varying probability maps for the left and right ventricles, the myocardium, and background structures such as the liver, stomach, lungs and skin. In addition to using the probabilistic cardiac atlas as a priori information, the segmentation algorithm incorporates spatial and temporal contextual information by using 4D Markov Random Fields. After the classification, the largest connected component of each structure is extracted using a global connectivity filter which improves the results significantly, especially for the myocardium. Validation against manual segmentations and computation of the correlation between manual and automatic segmentation on 249 3D volumes were calculated. We used the 'leave one out' test where the image set to be segmented was not used in the construction of its corresponding atlas. Results show that the procedure can successfully segment the left ventricle (LV) (r = 0.96), myocardium (r = 0.92) and right ventricle (r = 0.92). In addition, 4D images from 10 patients with hypertrophic cardiomyopathy were also manually and automatically segmented yielding a good correlation in the volumes of the LV (r = 0.93) and myocardium (0.94) when the atlas constructed with volunteers is blurred.     

Correlation based 3-D segmentation of the left ventricle in pediatric echocardiographic images using radio-frequency data

Clinical diagnosis of heart disease might be substantially supported by automated segmentation of the endocardial surface in three-dimensional (3-D) echographic images. Because of the poor echogenicity contrast between blood and myocardial tissue in some regions and the inherent speckle noise, automated analysis of these images is challenging. A priori knowledge on the shape of the heart cannot always be relied on, e.g., in children with congenital heart disease, segmentation should be based on the echo features solely. The objective of this study was to investigate the merit of using temporal cross-correlation of radio-frequency (RF) data for automated segmentation of 3-D echocardiographic images. Maximum temporal cross-correlation (MCC) values were determined locally from the RF-data using an iterative 3-D technique. MCC values as well as a combination of MCC values and adaptive filtered, demodulated RF-data were used as an additional, external force in a deformable model approach to segment the endocardial surface and were tested against manually segmented surfaces. Results on 3-D full volume images (Philips, iE33) of 10 healthy children demonstrate that MCC values derived from the RF signal yield a useful parameter to distinguish between blood and myocardium in regions with low echogenicity contrast and incorporation of MCC improves the segmentation results significantly. Further investigation of the MCC over the whole cardiac cycle is required to exploit the full benefit of it for automated segmentation.     

A unified framework for cross-modality multi-atlas segmentation of brain MRI

Multi-atlas label fusion is a powerful image segmentation strategy that is becoming increasingly popular in medical imaging. A standard label fusion algorithm relies on independently computed pairwise registrations between individual atlases and the (target) image to be segmented. These registrations are then used to propagate the atlas labels to the target space and fuse them into a single final segmentation. Such label fusion schemes commonly rely on the similarity between intensity values of the atlases and target scan, which is often problematic in medical imaging – in particular, when the atlases and target images are obtained via different sensor types or imaging protocols.In this paper, we present a generative probabilistic model that yields an algorithm for solving the atlas-to-target registrations and label fusion steps simultaneously. The proposed model does not directly rely on the similarity of image intensities. Instead, it exploits the consistency of voxel intensities within the target scan to drive the registration and label fusion, hence the atlases and target image can be of different modalities. Furthermore, the framework models the joint warp of all the atlases, introducing interdependence between the registrations.We use variational expectation maximization and the Demons registration framework in order to efficiently identify the most probable segmentation and registrations. We use two sets of experiments to illustrate the approach, where proton density (PD) MRI atlases are used to segment T1-weighted brain scans and vice versa. Our results clearly demonstrate the accuracy gain due to exploiting within-target intensity consistency and integrating registration into label fusion.     

Multiresolution texture based adaptive clustering algorithm for breast lesion segmentation

Objective: A specific algorithm is presented for the automatic extraction of breast tumors in ultrasonic imaging. Method: The algorithm involves two-dimensional adaptive K-means clustering of the gray scale and textural feature images. The segmentation problem is formulated as a maximum a posteriori (MAP) estimation problem. The MAP estimation is achieved using Besag's iterated conditional modes algorithm for the minimization of an energy function. This function has three components: the first constrains the region to be close to the data; the second imposes spatial continuity; and the third takes into consideration the texture of the various regions. A multiresolution implementation of the algorithm is performed using a wavelets basis. Results: Experiments were carried out on synthetic images and on in vivo breast ultrasound images. Various parameters involved in the algorithm are discussed to evaluate the robustness and accuracy of the segmentation method. Conclusion: Including textural features in the segmentation of ultrasonic data improves the robustness of the algorithm and makes the segmentation result less parameter dependent.     

Strain-Initialized Robust Bone Surface Detection in 3-D Ultrasound

Three-dimensional ultrasound has been increasingly considered as a safe radiation-free alternative to radiation-based fluoroscopic imaging for surgical guidance during computer-assisted orthopedic interventions, but because ultrasound images contain significant artifacts, it is challenging to automatically extract bone surfaces from these images. We propose an effective way to extract 3-D bone surfaces using a surface growing approach that is seeded from 2-D bone contours. The initial 2-D bone contours are estimated from a combination of ultrasound strain images and envelope power images. Novel features of the proposed method include: (i) improvement of a previously reported 2-D strain imaging-based bone segmentation method by incorporation of a depth-dependent cumulative power of the envelope into the elastographic data; (ii) incorporation of an echo decorrelation measure-based weight to fuse the strain and envelope maps; (iii) use of local statistics of the bone surface candidate points to detect the presence of any bone discontinuity; and (iv) an extension of our 2-D bone contour into a 3-D bone surface by use of an effective surface growing approach. Our new method produced average improvements in the mean absolute error of 18% and 23%, respectively, on 2-D and 3-D experimental phantom data, compared with those of two state-of-the-art bone segmentation methods. Validation on 2-D and 3-D clinical in vivo data also reveals, respectively, an average improvement in the mean absolute fitting error of 55% and an 18-fold improvement in the computation time.     

Functional and structural synergy for resolution recovery and partial volume correction in brain PET

Purpose: Positron Emission Tomography (PET) has the unique capability of measuring brain function but its clinical potential is affected by low resolution and lack of morphological detail. Here we propose and evaluate a wavelet synergistic approach that combines functional and structural information from a number of sources (CT, MRI and anatomical probabilistic atlases) for the accurate quantitative recovery of radioactivity concentration in PET images. When the method is combined with anatomical probabilistic atlases, the outcome is a functional volume corrected for partial volume effects.Methods: The proposed method is based on the multiresolution property of the wavelet transform. First, the target PET image and the corresponding anatomical image (CT/MRI/atlas-based segmented MRI) are decomposed into several resolution elements. Secondly, high-resolution components of the PET image are replaced, in part, with those of the anatomical image after appropriate scaling. The amount of structural input is weighted by the relative high frequency signal content of the two modalities. The method was validated on a digital Zubal phantom and clinical data to evaluate its quantitative potential.Results: Simulation studies showed the expected relationship between functional recovery and the amount of correct structural detail provided, with perfect recovery achieved when true images were used as anatomical reference. The use of T1-MRI images brought significant improvements in PET image resolution. However improvements were maximized when atlas-based segmented images as anatomical references were used; these results were replicated in clinical data sets.Conclusion: The synergistic use of functional and structural data, and the incorporation of anatomical probabilistic information in particular, generates morphologically corrected PET images of exquisite quality.     

Diverse data augmentation for learning image segmentation with cross-modality annotations

The dearth of annotated data is a major hurdle in building reliable image segmentation models. Manual annotation of medical images is tedious, time-consuming, and significantly variable across imaging modalities. The need for annotation can be ameliorated by leveraging an annotation-rich source modality in learning a segmentation model for an annotation-poor target modality. In this paper, we introduce a diverse data augmentation generative adversarial network (DDA-GAN) to train a segmentation model for an unannotated target image domain by borrowing information from an annotated source image domain. This is achieved by generating diverse augmented data for the target domain by one-to-many source-to-target translation. The DDA-GAN uses unpaired images from the source and target domains and is an end-to-end convolutional neural network that (i) explicitly disentangles domain-invariant structural features related to segmentation from domain-specific appearance features, (ii) combines structural features from the source domain with appearance features randomly sampled from the target domain for data augmentation, and (iii) train the segmentation model with the augmented data in the target domain and the annotations from the source domain. The effectiveness of our method is demonstrated both qualitatively and quantitatively in comparison with the state of the art for segmentation of craniomaxillofacial bony structures via MRI and cardiac substructures via CT.     

Effects of registration regularization and atlas sharpness on segmentation accuracy

In non-rigid registration, the tradeoff between warp regularization and image fidelity is typically determined empirically. In atlas-based segmentation, this leads to a probabilistic atlas of arbitrary sharpness: weak regularization results in well-aligned training images and a sharp atlas; strong regularization yields a "blurry" atlas. In this paper, we employ a generative model for the joint registration and segmentation of images. The atlas construction process arises naturally as estimation of the model parameters. This framework allows the computation of unbiased atlases from manually labeled data at various degrees of "sharpness", as well as the joint registration and segmentation of a novel brain in a consistent manner. We study the effects of the tradeoff of atlas sharpness and warp smoothness in the context of cortical surface parcellation. This is an important question because of the increasingly availability of atlases in public databases, and the development of registration algorithms separate from the atlas construction process. We find that the optimal segmentation (parcellation) corresponds to a unique balance of atlas sharpness and warp regularization, yielding statistically significant improvements over the FreeSurfer parcellation algorithm. Furthermore, we conclude that one can simply use a single atlas computed at an optimal sharpness for the registration-segmentation of a new subject with a pre-determined, fixed, optimal warp constraint. The optimal atlas sharpness and warp smoothness can be determined by probing the segmentation performance on available training data. Our experiments also suggest that segmentation accuracy is tolerant up to a small mismatch between atlas sharpness and warp smoothness.     

Comparative validation of multi-instance instrument segmentation in endoscopy: Results of the ROBUST-MIS 2019 challenge

Intraoperative tracking of laparoscopic instruments is often a prerequisite for computer and robotic-assisted interventions. While numerous methods for detecting, segmenting and tracking of medical instruments based on endoscopic video images have been proposed in the literature, key limitations remain to be addressed: Firstly, robustness, that is, the reliable performance of state-of-the-art methods when run on challenging images (e.g. in the presence of blood, smoke or motion artifacts). Secondly, generalization; algorithms trained for a specific intervention in a specific hospital should generalize to other interventions or institutions.In an effort to promote solutions for these limitations, we organized the Robust Medical Instrument Segmentation (ROBUST-MIS) challenge as an international benchmarking competition with a specific focus on the robustness and generalization capabilities of algorithms. For the first time in the field of endoscopic image processing, our challenge included a task on binary segmentation and also addressed multi-instance detection and segmentation. The challenge was based on a surgical data set comprising 10,040 annotated images acquired from a total of 30 surgical procedures from three different types of surgery. The validation of the competing methods for the three tasks (binary segmentation, multi-instance detection and multi-instance segmentation) was performed in three different stages with an increasing domain gap between the training and the test data. The results confirm the initial hypothesis, namely that algorithm performance degrades with an increasing domain gap. While the average detection and segmentation quality of the best-performing algorithms is high, future research should concentrate on detection and segmentation of small, crossing, moving and transparent instrument(s) (parts).     

HookNet: Multi-resolution convolutional neural networks for semantic segmentation in histopathology whole-slide images

We propose HookNet, a semantic segmentation model for histopathology whole-slide images, which combines context and details via multiple branches of encoder-decoder convolutional neural networks. Concentric patches at multiple resolutions with different fields of view, feed different branches of HookNet, and intermediate representations are combined via a hooking mechanism. We describe a framework to design and train HookNet for achieving high-resolution semantic segmentation and introduce constraints to guarantee pixel-wise alignment in feature maps during hooking. We show the advantages of using HookNet in two histopathology image segmentation tasks where tissue type prediction accuracy strongly depends on contextual information, namely (1) multi-class tissue segmentation in breast cancer and, (2) segmentation of tertiary lymphoid structures and germinal centers in lung cancer. We show the superiority of HookNet when compared with single-resolution U-Net models working at different resolutions as well as with a recently published multi-resolution model for histopathology image segmentation. We have made HookNet publicly available by releasing the source code¹ as well as in the form of web-based applications^2,3 based on the grand-challenge.org platform.