Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a unified UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to centralized, federated, and test-time UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/XuzheZ/MAPSeg/.

Automatically identifying the structural substrates underlying cardiac abnormalities can potentially provide real-time guidance for interventional procedures. With the knowledge of cardiac tissue substrates, the treatment of complex arrhythmias such as atrial fibrillation and ventricular tachycardia can be further optimized by detecting arrhythmia substrates to target for treatment (i.e., adipose) and identifying critical structures to avoid. Optical coherence tomography (OCT) is a real-time imaging modality that aids in addressing this need. Existing approaches for cardiac image analysis mainly rely on fully supervised learning techniques, which suffer from the drawback of workload on labor-intensive annotation process of pixel-wise labeling. To lessen the need for pixel-wise labeling, we develop a two-stage deep learning framework for cardiac adipose tissue segmentation using image-level annotations on OCT images of human cardiac substrates. In particular, we integrate class activation mapping with superpixel segmentation to solve the sparse tissue seed challenge raised in cardiac tissue segmentation. Our study bridges the gap between the demand on automatic tissue analysis and the lack of high-quality pixel-wise annotations. To the best of our knowledge, this is the first study that attempts to address cardiac tissue segmentation on OCT images via weakly supervised learning techniques. Within an in-vitro human cardiac OCT dataset, we demonstrate that our weakly supervised approach on image-level annotations achieves comparable performance as fully supervised methods trained on pixel-wise annotations.

An increased interest in longitudinal neurodevelopment during the first few years after birth has emerged in recent years. Noninvasive magnetic resonance imaging (MRI) can provide crucial information about the development of brain structures in the early months of life. Despite the success of MRI collections and analysis for adults, it remains a challenge for researchers to collect high-quality multimodal MRIs from developing infant brains because of their irregular sleep pattern, limited attention, inability to follow instructions to stay still during scanning. In addition, there are limited analytic approaches available. These challenges often lead to a significant reduction of usable MRI scans and pose a problem for modeling neurodevelopmental trajectories. Researchers have explored solving this problem by synthesizing realistic MRIs to replace corrupted ones. Among synthesis methods, the convolutional neural network-based (CNN-based) generative adversarial networks (GANs) have demonstrated promising performance. In this study, we introduced a novel 3D MRI synthesis framework- pyramid transformer network (PTNet3D)- which relies on attention mechanisms through transformer and performer layers. We conducted extensive experiments on high-resolution Developing Human Connectome Project (dHCP) and longitudinal Baby Connectome Project (BCP) datasets. Compared with CNN-based GANs, PTNet3D consistently shows superior synthesis accuracy and superior generalization on two independent, large-scale infant brain MRI datasets. Notably, we demonstrate that PTNet3D synthesized more realistic scans than CNN-based models when the input is from multi-age subjects. Potential applications of PTNet3D include synthesizing corrupted or missing images. By replacing corrupted scans with synthesized ones, we observed significant improvement in infant whole brain segmentation.

Pulmonary emphysema overlaps considerably with chronic obstructive pulmonary disease (COPD), and is traditionally subcategorized into three subtypes previously identified on autopsy. Unsupervised learning of emphysema subtypes on computed tomography (CT) opens the way to new definitions of emphysema subtypes and eliminates the need of thorough manual labeling. However, CT-based emphysema subtypes have been limited to texture-based patterns without considering spatial location. In this work, we introduce a standardized spatial mapping of the lung for quantitative study of lung texture location and propose a novel framework for combining spatial and texture information to discover spatially-informed lung texture patterns (sLTPs) that represent novel emphysema subtype candidates. Exploiting two cohorts of full-lung CT scans from the MESA COPD (n = 317) and EMCAP (n = 22) studies, we first show that our spatial mapping enables population-wide study of emphysema spatial location. We then evaluate the characteristics of the sLTPs discovered on MESA COPD, and show that they are reproducible, able to encode standard emphysema subtypes, and associated with physiological symptoms.

Neuroimaging and genetic biomarkers have been widely studied from discriminative perspectives towards Alzheimer's disease (AD) classification, since neuroanatomical patterns and genetic variants are jointly critical indicators for AD diagnosis. Generative methods, designed to model common occurring patterns, could potentially advance the understanding of this disease, but have not been fully explored for AD characterization. Moreover, the introduction of a supervised component into the generative process can constrain the model for more discriminative characterization. In this study, we propose an original method based on supervised topic modeling to characterize AD from a generative perspective, yet maintaining discriminative power at differentiating disease populations. Our topic modeling jointly exploits discretized image features and categorical genetic features. Diagnostic information - cognitively normal (CN), mild cognitive impairment (MCI) and AD - is introduced as a supervision variable. Experimental results on the ADNI cohort demonstrate that our model, while achieving competitive discriminative performance, can discover topics revealing both well-known and novel neuroanatomical patterns including temporal, parietal and frontal regions; as well as associations between genetic factors and neuroanatomical patterns.

Full quantification of regional cerebral metabolic rate of glucose (rCMRglu) with [18F]fluorodeoxyglucose ([18F]FDG) positron emission tomography (PET) imaging requires measurement of an arterial input function (AIF) curve, which is obtained with an invasive arterial blood sampling procedure during the scan. We previously proposed a non-invasive simultaneous estimation (nSIME) method that quantifies binding of a PET radioligand by combining individual electronic health records information and a pharmacokinetic AIF (PK-AIF) model. Initially applied only to [11C]DASB data, in this study we validate nSIME for a different radioligand, [18F]FDG, adapting the algorithm to the specific distribution and metabolism of this radioligand. We evaluate the impact of the PK-AIF model, the number of [18F]FDG-specific soft constraints, and the type of predictive strategy. The accuracy of nSIME is then compared to a population-based approach. All analyses are conducted on 67 [18F]FDG PET scans with arterial blood data available for comparison. nSIME performance is optimal for [18F]FDG when using the PK-AIF model, two soft constraints, and an aggregate model to predict the soft constraint values. Higher correlation and lower Bland-Altman spread against gold standard rCMRglu values based on arterial blood measurements are observed for nSIME (r = 0.83, spread = 1.55) compared to the population-based approach (r = 0.77, spread = 2.12). nSIME provides a data-driven estimation of both amplitude and shape of the AIF curve at the individual level and potentially enables non-invasive quantification of PET data across radioligands, avoiding the need for arterial blood sampling.

The papers in this special section were presented at three well-known conferences organized in 2014: EAI Mobihealth, IEEE HealthCom, and IEEE Biomedical and Health Informatics. EAI Mobihealth is an annually organized conference, which started in 2010, to address the demands of the rapidly evolving disciplines of wireless communications, mobile computing, and sensing technologies in healthcare. The IEEE-Healthcom is held every year since 1999 in different countries in Asia, Europe, and in America. It aims at bringing together interested parties working in the field of healthcare to exchange ideas, discuss innovative and emerging solutions, and develop collaborations. The IEEE Biomedical Health Informatics Conference started in 2013 and is organized every year providing the forum to showcase enabling technologies of computing, devices, imaging, sensors, and systems that optimize the acquisition, transmission, processing, storage, retrieval, visualization, and analysis of medical data. The aim of this special section is to present an overview of recent advances in sensing technologies, monitoring of patients, security and privacy of data transfer, provision of collaborative environments, data gathering and analysis from various sources, and predictive models, which all finally target the best strategy for patient monitoring and treatment.

Intravascular imaging using ultrasound or optical coherence tomography (OCT) is predominantly used to adjunct clinical information in interventional cardiology. OCT provides high-resolution images for detailed investigation of atherosclerosis-induced thickening of the lumen wall resulting in arterial blockage and triggering acute coronary events. However, the stochastic uncertainty of speckles limits effective visual investigation over large volume of pullback data, and clinicians are challenged by their inability to investigate subtle variations in the lumen topology associated with plaque vulnerability and onset of necrosis. This paper presents a lumen segmentation method using OCT imaging physics-based graph representation of signals and random walks image segmentation approaches. The edge weights in the graph are assigned incorporating OCT signal attenuation physics models. Optical backscattering maxima is tracked along each A-scan of OCT and is subsequently refined using global graylevel statistics and used for initializing seeds for the random walks image segmentation. Accuracy of lumen versus tunica segmentation has been measured on 15 in vitro and 6 in vivo pullbacks, each with 150-200 frames using 1) Cohen's kappa coefficient (0.9786 ±0.0061) measured with respect to cardiologist's annotation and 2) divergence of histogram of the segments computed with Kullback-Leibler (5.17 ±2.39) and Bhattacharya measures (0.56 ±0.28). High segmentation accuracy and consistency substantiates the characteristics of this method to reliably segment lumen across pullbacks in the presence of vulnerability cues and necrotic pool and has a deterministic finite time-complexity. This paper in general also illustrates the development of methods and framework for tissue classification and segmentation incorporating cues of tissue-energy interaction physics in imaging.

Quantitative analysis of positron emission tomography (PET) brain imaging data requires a metabolite-corrected arterial input function (AIF) for estimation of distribution volume and related outcome measures. Collecting arterial blood samples adds risk, cost, measurement error, and patient discomfort to PET studies. Minimally invasive AIF estimation is possible with simultaneous estimation (SIME), but at least one arterial blood sample is necessary. In this study, we describe a noninvasive SIME (nSIME) approach that utilizes a pharmacokinetic input function model and constraints derived from machine learning applied to an electronic health record database consisting of "long tail" data (digital records, paper charts, and handwritten notes) that were collected ancillary to the PET studies. We evaluated the performance of nSIME on 95 [(11)C]DASB PET scans that had measured AIFs. The results indicate that nSIME is a promising alternative to invasive AIF measurement. The general framework presented here may be expanded to other metabolized radioligands, potentially enabling quantitative analysis of PET studies without blood sampling. A glossary of technical abbreviations is provided at the end of this paper.

The extent of pulmonary emphysema is commonly estimated from CT scans by computing the proportional area of voxels below a predefined attenuation threshold. However, the reliability of this approach is limited by several factors that affect the CT intensity distributions in the lung. This work presents a novel method for emphysema quantification, based on parametric modeling of intensity distributions and a hidden Markov measure field model to segment emphysematous regions. The framework adapts to the characteristics of an image to ensure a robust quantification of emphysema under varying CT imaging protocols, and differences in parenchymal intensity distributions due to factors such as inspiration level. Compared to standard approaches, the presented model involves a larger number of parameters, most of which can be estimated from data, to handle the variability encountered in lung CT scans. The method was applied on a longitudinal data set with 87 subjects and a total of 365 scans acquired with varying imaging protocols. The resulting emphysema estimates had very high intra-subject correlation values. By reducing sensitivity to changes in imaging protocol, the method provides a more robust estimate than standard approaches. The generated emphysema delineations promise advantages for regional analysis of emphysema extent and progression.

This paper proposes a novel temporal knowledge representation and learning framework to perform large-scale temporal signature mining of longitudinal heterogeneous event data. The framework enables the representation, extraction, and mining of high-order latent event structure and relationships within single and multiple event sequences. The proposed knowledge representation maps the heterogeneous event sequences to a geometric image by encoding events as a structured spatial-temporal shape process. We present a doubly constrained convolutional sparse coding framework that learns interpretable and shift-invariant latent temporal event signatures. We show how to cope with the sparsity in the data as well as in the latent factor model by inducing a double sparsity constraint on the β-divergence to learn an overcomplete sparse latent factor model. A novel stochastic optimization scheme performs large-scale incremental learning of group-specific temporal event signatures. We validate the framework on synthetic data and on an electronic health record dataset.

Over the past two decades, intravascular ultrasound (IVUS) image segmentation has remained a challenge for researchers while the use of this imaging modality is rapidly growing in catheterization procedures and in research studies. IVUS provides cross-sectional grayscale images of the arterial wall and the extent of atherosclerotic plaques with high spatial resolution in real time. In this paper, we review recently developed image processing methods for the detection of media-adventitia and luminal borders in IVUS images acquired with different transducers operating at frequencies ranging from 20 to 45 MHz. We discuss methodological challenges, lack of diversity in reported datasets, and weaknesses of quantification metrics that make IVUS segmentation still an open problem despite all efforts. In conclusion, we call for a common reference database, validation metrics, and ground-truth definition with which new and existing algorithms could be benchmarked.

Computing the morphological similarity of diffusion tensors (DTs) at neighboring voxels within a DT image, or at corresponding locations across different DT images, is a fundamental and ubiquitous operation in the postprocessing of DT images. The morphological similarity of DTs typically has been computed using either the principal directions (PDs) of DTs (i.e., the direction along which water molecules diffuse preferentially) or their tensor elements. Although comparing PDs allows the similarity of one morphological feature of DTs to be visualized directly in eigenspace, this method takes into account only a single eigenvector, and it is therefore sensitive to the presence of noise in the images that can introduce error intothe estimation of that vector. Although comparing tensor elements, rather than PDs, is comparatively more robust to the effects of noise, the individual elements of a given tensor do not directly reflect the diffusion properties of water molecules. We propose a measure for computing the morphological similarity of DTs that uses both their eigenvalues and eigenvectors, and that also accounts for the noise levels present in DT images. Our measure presupposes that DTs in a homogeneous region within or across DT images are random perturbations of one another in the presence of noise. The similarity values that are computed using our method are smooth (in the sense that small changes in eigenvalues and eigenvectors cause only small changes in similarity), and they are symmetric when differences in eigenvalues and eigenvectors are also symmetric. In addition, our method does not presuppose that the corresponding eigenvectors across two DTs have been identified accurately, an assumption that is problematic in the presence of noise. Because we compute the similarity between DTs using their eigenspace components, our similarity measure relates directly to both the magnitude and the direction of the diffusion of water molecules. The favorable performance characteristics of our measure offer the prospect of substantially improving additional postprocessing operations that are commonly performed on DTI datasets, such as image segmentation, fiber tracking, noise filtering, and spatial normalization.

In tomographic medical devices such as single photon emission computed tomography or positron emission tomography cameras, image reconstruction is an unstable inverse problem, due to the presence of additive noise. A new family of regularization methods for reconstruction, based on a thresholding procedure in wavelet and wavelet packet (WP) decompositions, is studied. This approach is based on the fact that the decompositions provide a near-diagonalization of the inverse Radon transform and of prior information in medical images. A WP decomposition is adaptively chosen for the specific image to be restored. Corresponding algorithms have been developed for both two-dimensional and full three-dimensional reconstruction. These procedures are fast, noniterative, and flexible. Numerical results suggest that they outperform filtered back-projection and iterative procedures such as ordered-subset-expectation-maximization.

A new technique to reduce clinical magnetic resonance imaging (MRI) scan time by varying acquisition parameters and sharing k-space data between images, is proposed. To improve data utilization, acquisition of multiple images of different contrast is combined into a single scan, with variable acquisition parameters including repetition time (TR), echo time (TE), and echo train length (ETL). This approach is thus referred to as a "combo acquisition." As a proof of concept, simulations of MRI experiments using spin echo (SE) and fast SE (FSE) sequences were performed based on Bloch equations. Predicted scan time reductions of 25%-50% were achieved for 2-contrast and 3-contrast combo acquisitions. Artifacts caused by nonuniform k-space data weighting were suppressed through semi-empirical optimization of parameter variation schemes and the phase encoding order. Optimization was assessed by minimizing three quantitative criteria: energy of the "residue point spread function (PSF)," energy of "residue profiles" across sharp tissue boundaries, and energy of "residue images." In addition, results were further evaluated by quantitatively analyzing the preservation of contrast, the PSF, and the signal-to-noise ratio. Finally, conspicuity of lesions was investigated for combo acquisitions in comparison with standard scans. Implications and challenges for the practical use of combo acquisitions are discussed.

I. Introduction Wavelets are the result of collective efforts that recognized common threads between ideas and concepts that had been independently developed and investigated by distinct research communities. They provide a unifying framework for decomposing images, volumes, and time-series data into their elementary constituents across scale. Although a relatively recent construct, wavelets have become a tool of choice for engineers, physicists, and mathematicians, leading to efficient solutions in time and space frequency analysis problems, as well as a multitude of other applications. One of the consequences is that wavelet methods of analysis and representation are presently having a significant impact on the science of medical imaging and the diagnosis of disease and screening protocols. Because of a powerful underlying mathematical theory, they offer exciting opportunities for the design of new multiresolution image processing algorithms, and novel acquisition methods such as wavelet-encoded magnetic resonance imaging (MRI). This special issue of the IEEE Transactions on Medical Imaging focuses on these recent developments and highlights progress that has been accomplished in the areas related to medical imaging. II. Sizing the Wave: Some Facts and Figures Rarely has a mathematical concept generated so much response and enthusiasm within and between the engineering and mathematical research communities at large. To give a rough idea of the phenomenon, we provide a brief chronology. While wavelets have been traced all the way back to Alfred Haar in 1910 [1], for many, the starting point of their modern history coincides with two publications in the late 1980s by S. Mallat [2] and I. Daubechies [3]. These groundbreaking papers established a solid mathematical footing which would both shape and define the field. In a nutshell, S. Mallat identified the important concept of multiresolution analysis which is the corner stone of modern wavelet theory, while I. Daubechies constructed the first orthogonal wavelet bases that were compactly supported. These two contributions count among the most cited papers in the scientific literature (over 1500 SCI citations each). From that point on, the number of contributions relating to wavelet-applications and theory has increased steadily on the order of 9000 journal papers published to date. This trend is likely to continue as suggested by the strong response to the call for papers for this special issue (over 30 submissions). Wavelets have become so popular that distinct communities continue to have conferences and scientific journals entirely devoted to them. There are already historical anecdotes and folklore associated with them; an entertaining account of which can be found in the book of B. Burke Hubbard [4]. Readers who want to dive deeper into the subject have the daunting task of choosing among over 200 books written on wavelets. Our only advice in this regard is: in case of doubt, stick with the classics. Given the size of the phenomenon, it is no surprise that wavelets have had an impact on a number of disciplines, medical imaging being no exception. A first record of activity in this particular area is the workshop on wavelets in medicine and biology that took place at the annual IEEE-EMBS meeting, Baltimore, MD, 1992. The first journal paper describing a wavelet application in medical imaging—noise reduction in MRI by soft-thresholding in the wavelet domain—also appeared in 1992 [5]. Note that this work, which is often overlooked, provided the earliest description of a wavelet denoising method that has become extremely popular through the impulsion of Donoho et al. A large palette of wavelet applications in medical imaging is provided in [6]. Two complementary review articles are also available; the first gives a complete account of the activity taking place from the beginning to 1996 [7], while the second covers the more recent papers until 2000 [8]. So far, the primary applications of wavelets in medical imaging have been the following: Compression of medical images. CT reconstruction; local tomography. Wavelet denoising (MRI, ultrasound). Wavelet-based feature extraction; texture and statistical descriptors Medical image enhancement (e.g. fluoroscopy, mammography). Analysis of functional images of the brain [positron emission tomography (PET), functional MRI (fMRI)]. Wavelet-encoded MRI: Most of these topics are still active areas of research, as illustrated by the papers that are published in this special issue. III. Scanning Through the Issue Functional imaging is an area were methods of wavelet processing hold great promise. This particular line of research was initiated by U. Ruttimann, a creative researcher and good friend, who sadly passed away shortly before the publication of his paper in this very journal [9]. Another first rate statistician who was also active in this area at an early stage is J. Raz. By a sad coincidence, he also suffered a sudden death about a week before he was to present his latest results on wavelet analysis of fMRI [10]. Despite the tragic loss of these two pioneers, research in this area is alive and well as exemplified by the first three papers of this issue. Turkheimer et al. [11] consider the problem of the analysis of dynamic PET data; in particular, they advocate the use of a linear (James-Stein) wavelet estimator as an alternative to the more classical wavelet shrinkage or threshold detectors. Hossein-Zadeh et al. propose a wavelet-technique for the detection of activation in fMRI data [12]. Their contribution is twofold: first, the use of a redundant wavelet transform for better translation invariance, and second, a nonparametric detection method based on a randomization procedure. F. Meyer also considers fMRI time series but applies wavelets differently, within the context of a generalized linear model, to detrend the data [13]; that is, to get rid of signal drifts and disturbances that are not related to the stimulus. Two other areas where wavelets have achieved great success is signal denoising (typically, by simple thresholding in the wavelet domain) and tomographic reconstruction, mainly, because the Radon operator is well localized in a wavelet basis. Pizurica et al. [14] propose a novel wavelet denoising method that uses a local statistical model for improved signal estimation and noise suppression. Willett and Nowak [15] introduce a new multiscale image model, using piecewise planar basis functions, and apply their method to the reconstruction of photon-limited data (with Poisson noise). They develop penalized maximum likelihood methods for image denoising, deconvolution, and tomographic reconstruction. Kalifa et al. present a direct method for the efficient reconstruction of PET and single photom emission computed tomography data [16]. Their approach includes a nonlinear noise reduction step that is implemented by thresholding in the transformed domain; the key here is to select a transform (wavelet packet) that is optimized for the problem and data at hand (sparse representation of the signal and near diagonalization of the Radon operator). Bonnet et al. [17] also develop a direct approach for the reconstruction of cone-beam data which is known to be challenging. In essence, their approach is a wavelet adaptation of the Feldkamp algorithm. The special issue also features two contributions relating to ultrasound imaging. Michailovich and Adam consider the problem of the estimation of the spectrum of a ultrasound pulse [18]. Specifically, they develop a modified (outlier-resistant) wavelet estimator that they apply to the log-spectrum of the radio-freqency sequence. Lee et al. present a pattern recognition system that uses an M-band wavelet filterbank to extract fractal and texture features from ultrasonic images of the liver [19]. They report promising classification results, differientiating normal liver, cirrhosis, and hepatoma using a hierarchical classifier. The recent development of commercial digital mammography imaging systems not only provides a significant improvement in image quality for traditional screening, but translates into a wealth of information for the analysis and detection of mammographic features by computer. The papers by Lemaur et al. [20]. and Heinlein et al. [21] focus on the goals of early detection and visual enhancement of microcalcifications, respectively. In the former, the regularity of a wavelet basis is used to identify microcalcification in clusters. The identification of microcalcifaction in clusters as opposed to individual occurrences is of clinical significance as clusters may suggest the likelihood of malignancy. In the later paper, a discretization of the continuous wavelet transform is developed which allows a filterbank to be adapted for the enhancement of mammographic features. This implementation allows for the reconstruction of modified wavelet coefficients at arbitrary scales and orientations without the introduction of artifacts or loss of completeness. The integration of such an interactive enhancement tool into digital mammographic screening systems will be of great importance as the wealth of dynamic range (contrast) provided by digital detectors become generally available to radiologist through the introduction of lower cost softcopy display systems. The paper of Davatzikos et al. [22] offers another illustration of the versatility of wavelets. Their proposal is to represent the contours of a shape in a wavelet bases and to use this sparse representation to derive active shape models. Their results are promising and significant in terms of providing an automated solution to problems in volume quantification. The amount of data generated by modern imaging devices is often very large, and ever increasing. Thus, an important problem is to find efficient ways of compressing and encoding this information

This paper presents a method of four-dimensional (4-D) (3-D + Time) space-frequency analysis for directional denoising and enhancement of real-time three-dimensional (RT3D) ultrasound and quantitative measures in diagnostic cardiac ultrasound. Expansion of echocardiographic volumes is performed with complex exponential wavelet-like basis functions called brushlets. These functions offer good localization in time and frequency and decompose a signal into distinct patterns of oriented harmonics, which are invariant to intensity and contrast range. Deformable-model segmentation is carried out on denoised data after thresholding of transform coefficients. This process attenuates speckle noise while preserving cardiac structure location. The superiority of 4-D over 3-D analysis for decorrelating additive white noise and multiplicative speckle noise on a 4-D phantom volume expanding in time is demonstrated. Quantitative validation, computed for contours and volumes, is performed on in vitro balloon phantoms. Clinical applications of this spaciotemporal analysis tool are reported for six patient cases providing measures of left ventricular volumes and ejection fraction.

Mammograms depict most of the significant changes in breast disease. The primary radiographic signs of cancer are related to tumor mass, density, size, borders, and shape, and local distribution of calcifications. We show that each of these features can be well described by coherence and orientation measures and provide visual cues for radiologists to identify possible lesions more easily without increasing false positives. In this paper, an artifact-free enhancement algorithm based on overcomplete multiscale representations is presented. First, an image was decomposed using a fast wavelet transform algorithm. At each level of analysis, energy and phase information are computed via a set of separable steerable filters. Then, a measure of coherence within each level was obtained by weighting an energy measure with the ratio of projections of local energy within a specified window. Each projection was computed onto the central point of a window with respect to the total energy within that window. Finally, a nonlinear operation, integrating coherence and orientation information, was applied to modify transform coefficients within distinct levels of analysis. These modified coefficients were then reconstructed, via an inverse fast wavelet transform, resulting in an improved visualization of significant mammographic features. The novelty of this algorithm lies in the detection of directional multiscale features and the removal of aliased perturbations. Compared to existing multiscale enhancement techniques, images processed with this method appeared more familiar to radiologists due to localized enhancement of features.

Mammographic image enhancement methods are typically aimed at either improvement of the overall visibility of features or enhancement of a specific sign of malignancy. Here, the authors present a synthesis of the two paradigms by means of image fusion. After a redundant B-spline wavelet transform decomposition is carried out, the transform coefficients are processed for enhancement of microcalcifications, circumscribed masses, and stellate lesions. The modified coefficients are then fused for reconstruction of an enhanced image with improved visualization of malignancies. Both processing for enhancement of selected features and fusion of the resultant images are accomplished within a single wavelet transform frameword which contributes to the computational efficiency of the described method. The devised algorithm not only allows for efficient combination of specific features of importance in the contrast enhanced images, but also provides a flexible frameword for incorporation of different enhancement methods and their independent optimization.

The lifting scheme was introduced as a flexible tool to construct compactly supported second generation wavelets and the wavelet transform. However because it is not translation invariant, the traditional lifting framework may not be good for multiscale feature analysis where translation-invariant characteristics are highly desirable. In this paper we address the following question: can the lifting scheme be used as a framework for overcomplete wavelet representations with multiscale feature analysis in mind? We address this question by investigating each stage of the multiscale analysis: split, dual lifting and primal lifting. We introduce a smoothing lazy wavelet in the split stage. We then show that only the dual lifting is necessary since the primal lifting required to compensate for the aliasing needs no longer exist. We also demonstrate that the proposed scheme achieves better performance without introducing boundary artifacts that exist in the traditional methods.

This paper presents an algorithm for speckle reduction and contrast enhancement of echocardiographic images. Within a framework of multiscale wavelet analysis, we apply wavelet shrinkage techniques to eliminate noise while preserving the sharpness of salient features. In addition, nonlinear processing of feature energy is carried out to enhance contrast within local structures and along object boundaries. We show that the algorithm is capable of not only reducing speckle, but also enhancing features of diagnostic importance, such as myocardial walls in two-dimensional echocardiograms obtained from the parasternal short-axis view. Shrinkage of wavelet coefficients via soft thresholding within finer levels of scale is carried out on coefficients of logarithmically transformed echocardiograms. Enhancement of echocardiographic features is accomplished via nonlinear stretching followed by hard thresholding of wavelet coefficients within selected (midrange) spatial-frequency levels of analysis. We formulate the denoising and enhancement problem, introduce a class of dyadic wavelets, and describe our implementation of a dyadic wavelet transform. Our approach for speckle reduction and contrast enhancement was shown to be less affected by pseudo-Gibbs phenomena. We show experimentally that this technique produced superior results both qualitatively and quantitatively when compared to results obtained from existing denoising methods alone. A study using a database of clinical echocardiographic images suggests that such denoising and enhancement may improve the overall consistency of expert observers to manually defined borders.

Mammograms can depict most of the significant changes of breast disease. The primary radiographic signs of cancer are masses (its density, size, shape, borders), spicular lesions and calcification content. These features may be extracted according to their coherence and orientation and can provide important visual cues for radiologists to locate suspicious areas without generating false positives. An artifact free enhancement algorithm based on overcomplete multiscale wavelet analysis is presented. The novelty of this algorithm lies in its detection of directional features and removal of unwanted perturbations. Compared to existing multiscale enhancement approaches, images processed with this method appear more familiar to radiologists and naturally close to the original mammogram.

This paper presents an approach for image segmentation using sub-octave wavelet representations and a dynamic resource-allocating neural network. The algorithm is applied to identify regions of masses in mammographic images of varied degrees of perceptual difficulty. Each mammographic image is first decomposed into overcomplete wavelet representations of sub-octave frequency bands. A feature vector for each pixel through the scale space is constructed from fine to coarse scales. The feature vectors are used to drive a neural network classifier of dynamic resource allocation for segmentation. Sub-octave wavelet representations have an improved capability of characterizing subtle (band-limited) features frequently seen in mammographic images. A radial basis network of dynamic resource allocation is shown to have better adaptation and generalization in a redundant feature space. Experimental results along with statistical analysis are partially compared to a traditional classifier.

We present an interactive scheme for processing of digital mammograms relying upon a steerable dyadic wavelet transform. Coefficients of the translation and rotation-invariant transform are interactively processed before an inverse transform is applied. Analysis is carried out at dyadic scales and along arbitrary orientations. Local orientation is computed at each level of scale and spatial position and formulated into criteria for including or excluding specific orientations for contrast enhancement and enhancing locally radiating structures. Transform coefficients that were selected for contrast enhancement are modified by a piecewise linear enhancement function. The presented scheme is flexible enough to enable efficient position, scale, and orientation based interactive processing and analysis.

An algorithm for endocardial and epicardial border identification of the left ventricle in 2-D short-axis echocardiographic images is presented. Our approach relies on shape modeling of endocardial and epicardial boundaries and prominent border information extracted from image sequences. The algorithm consists of four steps; wavelet-based edge detection, border segment extraction, border reconstruction, and boundary smoothing. Wavelet maximum representation of edges, dynamic shape modeling and matched filtering techniques are utilized to determine the center point of the left ventricle, and carry out feature extraction of border segments to better approximate endocardial and epicardial boundaries. The algorithm can reliably estimate the center point of the left ventricle, and also determine both endocardial and epicardial boundaries. Myocardial boundary identification is autonomous requiring no human input for initial estimation of boundary locations. Sample experimental results are shown for endocardial and epicardial border identification in 2-D short-axis echocardiograms.

We introduce a novel method of feature extraction for texture segmentation that relies on multichannel wavelet frames and 2-D envelope detection. We describe and compare two algorithms for envelope detection based on (1) the Hilbert transform and (2) zero crossings. We present criteria for filter selection and discuss quantitatively their effect on feature extraction. The performance of our method is demonstrated experimentally on samples of both natural and synthetic textures.

An image fusion algorithm based on multiscale analysis along arbitrary orientations is presented. After a steerable dyadic wavelet transform decomposition of multi-sensor images is carried out, the maximum local oriented energy is determined at each level of scale and spatial position. Maximum local oriented energy and local dominant orientation are used to combine transform coefficients obtained from the analysis of each input image. Reconstruction is accomplished from the modified coefficients, resulting in a fused image. Examples of multi-sensor fusion and fusion using different settings of a single sensor are demonstrated.

Feature extraction remains an important part of low-level vision. Traditional oriented filters have been effective tools to identify features, such as lines and edges. Steerable filters, which can be adjusted at arbitrary orientation, have made decisions of feature orientations more precise. Combined with a pyramid structure of a multiscale representation, these filters can provide a reliable and efficient tool for image analysis. This paper takes advantage of multiscale steerable filters in the context of de-noising. First a set of novel filters are designed, that decompose the frequency plane into distinct directional bands. Next, we identify the dominant direction and strength at each point of an image from quadrature pairs of steerable filters. A nonlinear threshold function is then applied to the filtered coefficients to suppress noise. The denoised image is restored from coefficients modified at each level of transform space. We demonstrate the benefits of multiscale steerable filters for de-noising and show that it can greatly reduce noise while preserving image features. Two examples are presented to verify the efficacy of the technique.

This paper describes a method of pattern recognition targeted for recognizing complex annotations found in paper documents. Our investigation is motivated by the high reliability required for accomplishing autonomous interpretation of maps and engineering drawings. Our approach includes a strategy based on multiscale representations obtained by hexagonal wavelet analysis. A feasibility study is described in which more than 10,000 patterns were recognized with an error rate of 2.06% by a neural network trained using multiscale representations from a class of 52 distinct patterns. We observed a 21-fold reduction in the amount of information needed to represent each pattern for recognition. These results suggest that high reliability is possible at a reduced cost of representation.>

Mammography's role in the detection of breast cancer at early stages is well known. Although more accurate than other existing techniques, mammography still only finds 80 to 90 percent of breast cancers. It has been suggested that mammograms, as normally viewed, display only about 3% of the total information detected. The general inability to detect small tumors and other salient features within mammograms motivates our investigation of a system we call the Mammogram Display System (MDS). The core technology used for MDS image enhancement is the wavelet transform.>

Robust and reliable edge detection is an important step for accomplishing automatic detection of heart wall boundaries in echocardiograms. We present an edge detection algorithm that makes use of both spatial and temporal information. Our algorithm is comprised of (1) a 3-D discrete dyadic wavelet transform carried out on a sequence of images, (2) edge detection that is carried out by maxima detection and a search strategy for maxima curves within the transform space. Using detected edges and "a priori" knowledge of cardiographic features we demonstrate the performance of fully automatic detection of epicardial and endocardial boundaries along the posterior and anterior walls in 2-D short-axis echocardiographic image sequences.>

Introduces a novel approach for accomplishing mammographic feature analysis by overcomplete multiresolution representations. The authors show that efficient representations may be identified within a continuum of scale-space and used to enhance features of importance to mammography. Methods of contrast enhancement are described based on three overcomplete multiscale representations: 1) the dyadic wavelet transform (separable), 2) the phi-transform (nonseparable, nonorthogonal), and 3) the hexagonal wavelet transform (nonseparable). Multiscale edges identified within distinct levels of transform space provide local support for image enhancement. Mammograms are reconstructed from wavelet coefficients modified at one or more levels by local and global nonlinear operators. In each case, edges and gain parameters are identified adaptively by a measure of energy within each level of scale-space. The authors show quantitatively that transform coefficients, modified by adaptive nonlinear operators, can make more obvious unseen or barely seen features of mammography without requiring additional radiation. The authors' results are compared with traditional image enhancement techniques by measuring the local contrast of known mammographic features. They demonstrate that features extracted from multiresolution representations can provide an adaptive mechanism for accomplishing local contrast enhancement. By improving the visualization of breast pathology, one can improve chances of early detection while requiring less time to evaluate mammograms for most patients.

This correspondence introduces a new approach to characterize textures at multiple scales. The performance of wavelet packet spaces are measured in terms of sensitivity and selectivity for the classification of twenty-five natural textures. Both energy and entropy metrics were computed for each wavelet packet and incorporated into distinct scale space representations, where each wavelet packet (channel) reflected a specific scale and orientation sensitivity. Wavelet packet representations for twenty-five natural textures were classified without error by a simple two-layer network classifier. An analyzing function of large regularity (D/sub 20/) was shown to be slightly more efficient in representation and discrimination than a similar function with fewer vanishing moments (D/sub 6/) In addition, energy representations computed from the standard wavelet decomposition alone (17 features) provided classification without error for the twenty-five textures included in our study. The reliability exhibited by texture signatures based on wavelet packets analysis suggest that the multiresolution properties of such transforms are beneficial for accomplishing segmentation, classification and subtle discrimination of texture.>

A novel method for character recognition targeted at complex annotations found in engineering documents is presented. A feasibility study is described in which characters extracted from engineering drawings were recognized without error from a class of 36 distinct alphanumeric patterns by a neural network classifier trained with multiscale representations. An incremental strategy is presented for resolution which relies upon the continuity between hierarchical levels of a novel multiscale decomposition. The authors observed a 16-fold reduction in the amount of information needed to represent each character for recognition. These results suggest high reliability at a reduced cost of representation.>

An effort has been made to develop a robust high-speed stereo matcher by exploiting parallel algorithms executing on general-purpose SIMD machines. This approach is based on several existing techniques dealing with the classification and evaluation of matches, the application of ordering constraints, and relaxation-based matching. The techniques have been integrated and reformulated in terms of parallel execution on a theoretical SIMD machine. An ideal machine topology for executing this parallel algorithm is identified through complexity analysis. Feasibility is demonstrated by implementation on a commercially available SIMD machine, and its performance is compared with that of the idealized machine. Sample results are shown for real and synthetic stereo pairs.>

A parallel algorithm for stereo matching that achieves high speed by exploiting the parallel architectures of typical single-instruction multiple-data (SIMD) processors is presented. The approach is based on several existing techniques dealing with the classification and evaluation of matches, the application of ordering constraints, and relaxation-based matching. The techniques have been integrated and reformulated in terms of parallel execution on a theoretical SIMD machine. Feasibility is demonstrated by implementation on a commercially available SIMD machine. An ideal machine, operating at 60 Hz, can accomplish stereo matching in 1.5 s using 88 machine cycles. On the commercial machine, stereo matching was achieved in 13.5 s using 404 cycles.>

A methodology is presented for designing detectors which locate specific features in an image. The method is applied to the detection and segmentation of spherical features. A vital part of the detection and segmentation is the use of the gradient angle transform. An analysis of the gradient angle for ideal spheres is presented, with a discussion of how this may be used to locate the boundaries of the sphere. The algorithms used by a program which detects and segments spherical features are then presented. The results of applying the program to images with man-made spherical features are given.>

The authors are concerned with the development of a novel approach to edge-based stereo matching. In the context of an incremental matching strategy the authors have replaced the traditional hierarchical (coarse-fine) matching by an approach called complexity control based matching. The implementation of this method allows the user to select interactively features which (given the context provided by previous matches) are most likely to be matched successfully. The selection is done at the resolution of the original image and utilizes a rich set of feature properties (e.g. edge strength, orientation, length, texture etc.), either alone or in logical combinations. Both feature-selection and feature-matching algorithms execute at real-time rates, and all interactions are via a stereo workstation.>