Segmentation of elastographic images using a coarse-to-fine active contour model

Breast-lesion Segmentation Combining B-Mode and Elastography Ultrasound

Breast ultrasound (BUS) imaging has become a crucial modality, especially for providing a complementary view when other modalities (i.e., mammography) are not conclusive in the task of assessing lesions. The specificity in cancer detection using BUS imaging is low. These false-positive findings often lead to an increase of unnecessary biopsies. In addition, increasing sensitivity is also challenging given that the presence of artifacts in the B-mode ultrasound (US) images can interfere with lesion detection. To deal with these problems and improve diagnosis accuracy, ultrasound elastography was introduced. This paper validates a novel lesion segmentation framework that takes intensity (B-mode) and strain information into account using a Markov Random Field (MRF) and a Maximum a Posteriori (MAP) approach, by applying it to clinical data. A total of 33 images from two different hospitals are used, composed of 14 cancerous and 19 benign lesions. Results show that combining both the B-mode and strain data in a unique framework improves segmentation results for cancerous lesions (Dice Similarity Coefficient of 0.49 using B-mode, while including strain data reaches 0.70), which are difficult images where the lesions appear with blurred and not well-defined boundaries.

3-D visualization and non-linear tissue classification of breast tumors using ultrasound elastography in vivo

The goal of the study described here was to introduce new methods for the classification and visualization of human breast tumors using 3-D ultrasound elastography. A tumor's type, shape and size are key features that can help the physician to decide the sort and extent of necessary treatment. In this work, tumor type, being either benign or malignant, was classified non-invasively for nine volunteer patients. The classification was based on estimating four parameters that reflect the tumor's non-linear biomechanical behavior, under multi-compression levels. Tumor prognosis using non-linear elastography was confirmed with biopsy as a gold standard. Three tissue classification parameters were found to be statistically significant with a p-value < 0.05, whereas the fourth non-linear parameter was highly significant, having a p-value < 0.001. Furthermore, each breast tumor's shape and size were estimated in vivo using 3-D elastography, and were enhanced using interactive segmentation. Segmentation with level sets was used to isolate the stiff tumor from the surrounding soft tissue. Segmentation also provided a reliable means to estimate tumors volumes. Four volumetric strains were investigated: the traditional normal axial strain, the first principal strain, von Mises strain and maximum shear strain. It was noted that these strains can provide varying degrees of boundary enhancement to the stiff tumor in the constructed elastograms. The enhanced boundary improved the performance of the segmentation process. In summary, the proposed methods can be employed as a 3-D non-invasive tool for characterization of breast tumors, and may provide early prognosis with minimal pain, as well as diminish the risk of late-stage breast cancer.

Nonlinear characterization of breast cancer using multi-compression 3D ultrasound elastography in vivo

The main objective of this article is to introduce a new nonlinear elastography based classification method for human breast masses. Multi-compression elastography imaging is elucidated in this study to differentiate malignant from benign lesions, based on their nonlinear mechanical behavior under compression. Three classification parameters were used and compared in this work: a new nonlinear parameter based on a power-law behavior of the strain difference between breast masses and healthy tissues, mass-soft tissue strain ratio and the mass relative volume between B-mode and elastography imaging. Using 3D elastography, these parameters were tested in vivo. A pilot study on 10 patients was performed, and results were compared with biopsy diagnosis as a gold standard. Initial elastography results showed a good agreement with biopsy outcomes. The new estimated nonlinear parameter had an average value of 0.163±0.063 and 1.642±0.261 for benign and malignant masses, respectively. Strain ratio values for the benign and malignant masses had an average value of 2.135±0.707 and 4.21±2.108, respectively. Relative mass volume was 0.848±0.237 and 2.18±0.522 for benign and malignant masses. In addition to the traditional normal axial strain, new strain types were used for elastography and constructed in 3D, including the first principal, maximum shear and Von Mises strains. The new strains provided an enhanced distinction of the stiff lesion from the soft tissue. In summary, the proposed elastographic techniques can be used as a noninvasive quantitative characterization tool for breast cancer, with the capability of visualizing and separating the masses in a three dimensional space. This may reduce the number of unnecessary painful breast biopsies.

Fuzzy-based classification of breast lesions using ultrasound echography and elastography

Common breast lesions have different elasticity properties. Segmentation of contours of breast lesions from elastography and B mode images by incorporating variational level set method is involved in the proposed work. After segmentation, strain and shape features, such as differences in area, perimeter, and contour and width to height difference and solidity, as well as texture features like contrast, entropy, standard deviation, dissimilarity, homogeneity and energy, are estimated. A nonlinear fuzzy inference system is applied for classifying the breast lesions as benign cyst, benign solid mass, or malignant solid mass. Detection of malignant solid masses is our primary objective. A classification accuracy of 83% is obtained. One hundred percent sensitivity is reported. It can be concluded that the proposed fuzzy-based classification technique can be used as an aid for the automated detection of breast lesions.

Automated Delineation of Lung Tumors from CT Images Using a Single Click Ensemble Segmentation Approach

A single click ensemble segmentation (SCES) approach based on an existing "Click&Grow" algorithm is presented. The SCES approach requires only one operator selected seed point as compared with multiple operator inputs, which are typically needed. This facilitates processing large numbers of cases. Evaluation on a set of 129 CT lung tumor images using a similarity index (SI) was done. The average SI is above 93% using 20 different start seeds, showing stability. The average SI for 2 different readers was 79.53%. We then compared the SCES algorithm with the two readers, the level set algorithm and the skeleton graph cut algorithm obtaining an average SI of 78.29%, 77.72%, 63.77% and 63.76% respectively. We can conclude that the newly developed automatic lung lesion segmentation algorithm is stable, accurate and automated.

Three-dimensional elastography for cervical lymph node volume measurements: a study to investigate feasibility, accuracy and reliability

This study investigated the feasibility of using three-dimensional (3-D) elastography in measuring cervical lymph node volume and compared the accuracy and reliability of 3-D elastography and 3-D grayscale ultrasound in measurement of ill-defined cervical nodes. Eighteen porcine lymph nodes from the neck were embedded in tissue-mimicking phantoms and scanned with the two ultrasound techniques. Ultrasound measurements were compared with the volume determined by water-displacement method to evaluate measurement accuracy. Inter-observer reproducibility and intra-observer repeatability of measurements were evaluated. Four patients with enlarged neck nodes were included to evaluate intra-observer repeatability of ultrasound measurements. Results demonstrated that lymph nodes that appeared ill-defined on grayscale ultrasound showed well-defined boundaries on elastography. 3-D elastography has higher measurement accuracy (84.2%), reproducibility (intraclass correlation coefficient, ICC = 0.909) and repeatability (ICC = 0.964-0.988) than does 3-D grayscale ultrasound (62.2%, 0.777 and 0.863-0.906 respectively). As a conclusion, 3-D elastography is accurate and reliable in volume measurement of ill-defined lymph nodes and has potential for accurate assessment of lymph node volume.

Radiomics: the process and the challenges

"Radiomics" refers to the extraction and analysis of large amounts of advanced quantitative imaging features with high throughput from medical images obtained with computed tomography, positron emission tomography or magnetic resonance imaging. Importantly, these data are designed to be extracted from standard-of-care images, leading to a very large potential subject pool. Radiomics data are in a mineable form that can be used to build descriptive and predictive models relating image features to phenotypes or gene-protein signatures. The core hypothesis of radiomics is that these models, which can include biological or medical data, can provide valuable diagnostic, prognostic or predictive information. The radiomics enterprise can be divided into distinct processes, each with its own challenges that need to be overcome: (a) image acquisition and reconstruction, (b) image segmentation and rendering, (c) feature extraction and feature qualification and (d) databases and data sharing for eventual (e) ad hoc informatics analyses. Each of these individual processes poses unique challenges. For example, optimum protocols for image acquisition and reconstruction have to be identified and harmonized. Also, segmentations have to be robust and involve minimal operator input. Features have to be generated that robustly reflect the complexity of the individual volumes, but cannot be overly complex or redundant. Furthermore, informatics databases that allow incorporation of image features and image annotations, along with medical and genetic data, have to be generated. Finally, the statistical approaches to analyze these data have to be optimized, as radiomics is not a mature field of study. Each of these processes will be discussed in turn, as well as some of their unique challenges and proposed approaches to solve them. The focus of this article will be on images of non-small-cell lung cancer.

Characterizing the compression-dependent viscoelastic properties of human hepatic pathologies using dynamic compression testing

Recent advances in elastography have provided several imaging modalities capable of quantifying the elasticity of tissue, an intrinsic tissue property. This information is useful for determining tumour margins and may also be useful for diagnosing specific tumour types. In this study, we used dynamic compression testing to quantify the viscoelastic properties of 16 human hepatic primary and secondary malignancies and their corresponding background tissue obtained following surgical resection. Two additional backgrounds were also tested. An analysis of the background tissue showed that F4-graded fibrotic liver tissue was significantly stiffer than F0-graded tissue, with a modulus contrast of 4:1. Steatotic liver tissue was slightly stiffer than normal liver tissue, but not significantly so. The tumour-to-background storage modulus contrast of hepatocellular carcinomas, a primary tumour, was approximately 1:1, and the contrast decreased with increasing fibrosis grade of the background tissue. Ramp testing showed that the background stiffness increased faster than the malignant tissue. Conversely, secondary tumours were typically much stiffer than the surrounding background, with a tumour-to-background contrast of 10:1 for colon metastases and 10:1 for cholangiocarcinomas. Ramp testing showed that colon metastases stiffened faster than their corresponding backgrounds. These data have provided insights into the mechanical properties of specific tumour types, which may prove beneficial as the use of quantitative stiffness imaging increases.

Analysis of shear strain imaging for classifying breast masses: finite element and phantom results

PURPOSE

Features extracted from axial-shear strain images of breast masses have been previously utilized to differentiate and classify benign from malignant breast masses. In this paper, we compare shear strain patterns exhibited by both the full-shear (axial and lateral component) versus only the axial-shear strain component for differentiating between bound masses (malignant) when compared to unbound masses (benign).

METHODS

We examine different breast mass characteristics such as mass shape, asymmetric location of masses, stiffness variations, and mass bonding characteristics to background tissue to assess their impact on shear strain patterns generated due to a uniaxial applied deformation. Two-dimensional finite element simulations of both circular and elliptical inclusions embedded within a uniform background were utilized. Different degrees of bonding were characterized using friction coefficient values ranging from 0.01 to 100 denoting loosely bound to firmly bound masses. Single-inclusion tissue-mimicking phantoms mimicking firmly bound and loosely bound ellipsoidal masses oriented at four different angles to the applied deformation were studied to corroborate the mass differentiation performance.

RESULTS

Our results indicate that the normalized axial-shear strain and full-shear strain area features are larger for bound when compared to unbound masses. A higher stiffness ratio or contrast between the inclusion and background also improves differentiation. Larger applied deformations reduce the discrimination performance for masses with friction coefficients lower than 0.4, due to increased mass slippage with applied deformations. Potential errors with the use of these features would occur for unbound inclusions at larger applied deformations and for asymmetric mass positions within the background normal tissue.

CONCLUSIONS

Finite element and tissue-mimicking phantom results demonstrate the feasibility of utilizing both the normalized axial-shear and full-shear strain area features to classify breast masses. Differentiation between bound or unbound masses was not affected by the mass size or shape for masses where the applied deformation is normal to the mass surface. Shear strain patterns vary significantly especially within unbound masses, when the mass surface is not normal to the applied deformation. Discrimination performance for unbound masses was improved by utilizing only the normalized shear strain area patterns located outside the mass as illustrated in this paper.

Contour detection of atherosclerotic plaques in IVUS images using ellipse template matching and particle swarm optimization

It is valuable for diagnosis of atherosclerosis to detect lumen and media-adventitia contours in intravascular ultrasound (IVUS) images of atherosclerotic plaques. In this paper, a method for contour detection of plaques is proposed utilizing the prior knowledge of elliptic geometry of plaques. Contours are initialized as ellipses by using ellipse template matching, where a matching function is maximized by particle swarm optimization. Then the contours are refined by boundary vector field snakes. The method was evaluated via 88 in vivo images from 21 patients. It outperformed a state-of-the-art method by 3.8 pixels and 4.8% in terms of the mean distance error and relative mean distance error, respectively.

Measurement consistency from magnetic resonance images

RATIONALE AND OBJECTIVES

In quantifying medical images, length-based measurements are still obtained manually. Due to possible human error, a measurement protocol is required to guarantee the consistency of measurements. In this work, we review various statistical techniques that can be used in determining measurement consistency. The focus is on detecting a possible measurement bias and determining the robustness of the procedures to outliers.

MATERIALS AND METHODS

We review correlation analysis, linear regression, Bland-Altman method, paired t-test, and analysis of variance (ANOVA). These techniques were applied to measurements, obtained by two raters, of head and neck structures from magnetic resonance images.

RESULTS

The correlation analysis and the linear regression were shown to be insufficient for detecting measurement inconsistency. They are also very sensitive to outliers. The widely used Bland-Altman method is a visualization technique, so it lacks the numeric quantification. The paired t-test tends to be sensitive to small measurement bias. In contrast, ANOVA performs well even under small measurement bias.

CONCLUSIONS

In almost all cases, using only one method is insufficient and it is recommended that several methods be used simultaneously. In general, ANOVA performs the best.

Automated volume analysis of head and neck lesions on CT scans using 3D level set segmentation

The authors have developed a semiautomatic system for segmentation of a diverse set of lesions in head and neck CT scans. The system takes as input an approximate bounding box, and uses a multistage level set to perform the final segmentation. A data set consisting of 69 lesions marked on 33 scans from 23 patients was used to evaluate the performance of the system. The contours from automatic segmentation were compared to both 2D and 3D gold standard contours manually drawn by three experienced radiologists. Three performance metric measures were used for the comparison. In addition, a radiologist provided quality ratings on a 1 to 10 scale for all of the automatic segmentations. For this pilot study, the authors observed that the differences between the automatic and gold standard contours were larger than the interobserver differences. However, the system performed comparably to the radiologists, achieving an average area intersection ratio of 85.4% compared to an average of 91.2% between two radiologists. The average absolute area error was 21.1% compared to 10.8%, and the average 2D distance was 1.38 mm compared to 0.84 mm between the radiologists. In addition, the quality rating data showed that, despite the very lax assumptions made on the lesion characteristics in designing the system, the automatic contours approximated many of the lesions very well.

Ablation monitoring with elastography: 2D in-vivo and 3D ex-vivo studies

The clinical feasibility of 2D elastography methods is hindered by the requirement that the operator avoid out-of-plane motion of the ultrasound image during palpation, and also by the lack of volumetric elastography measurements. In this paper, we develop and evaluate a 3D elastography method operating on volumetric data acquired from a 3D probe. Our method is based on minimizing a cost function using dynamic programming (DP). The cost function incorporates similarity of echo amplitudes and displacement continuity. We present, to the best of our knowledge, the first in-vivo patient studies of monitoring liver ablation with freehand DP elastography. The thermal lesion was not discernable in the B-mode image but it was clearly visible in the strain image as well as in validation CT. We also present 3D strain images from thermal lesions in ex-vivo ablation. Good agreement was observed between strain images, CT and gross pathology.

A hybrid multi-scale model for thyroid nodule boundary detection on ultrasound images

A hybrid model for thyroid nodule boundary detection on ultrasound images is introduced. The segmentation model combines the advantages of the "á trous" wavelet transform to detect sharp gray-level variations and the efficiency of the Hough transform to discriminate the region of interest within an environment with excessive structural noise. The proposed method comprise three major steps: a wavelet edge detection procedure for speckle reduction and edge map estimation, based on local maxima representation. Subsequently, a multiscale structure model is utilised in order to acquire a contour representation by means of local maxima chaining with similar attributes to form significant structures. Finally, the Hough transform is employed with 'a priori' knowledge related to the nodule's shape in order to distinguish the nodule's contour from adjacent structures. The comparative study between our automatic method and manual delineations demonstrated that the boundaries extracted by the hybrid model are closely correlated with that of the physicians. The proposed hybrid method can be of value to thyroid nodules' shape-based classification and as an educational tool for inexperienced radiologists.