This PDF file contains the editorial “Pioneers in Medical Imaging Series” for JMI Vol. 1 Issue 03

This PDF file contains the editorial “Special Section Guest Editorial: Pioneers in Medical Imaging: Honoring the Memory of Robert F. Wagner” for JMI Vol. 1 Issue 03

In an effort to generalize task-based assessment beyond traditional signal detection, there is a growing interest in performance evaluation for combined detection and estimation tasks, in which signal parameters, such as size, orientation, and contrast are unknown and must be estimated. One motivation for studying such tasks is their rich complexity, which offers potential advantages for imaging system optimization. To evaluate observer performance on combined detection and estimation tasks, Clarkson introduced the estimation receiver operating characteristic (EROC) curve and the area under the EROC curve as a summary figure of merit. This work provides practical tools for EROC analysis of experimental data. In particular, we propose nonparametric estimators for the EROC curve, the area under the EROC curve, and for the variance/covariance matrix of a vector of correlated EROC area estimates. In addition, we show that reliable confidence intervals can be obtained for EROC area, and we validate these intervals with Monte Carlo simulation. Application of our methodology is illustrated with an example comparing magnetic resonance imaging k -space sampling trajectories. MATLAB® software implementing the EROC analysis estimators described in this work is publicly available at http://code.google.com/p/iqmodelo/.

In x-ray imaging, contrast information content varies with photon energy. It is, therefore, possible to improve image quality by weighting photons according to energy. We have implemented and evaluated so-called energy weighting on a commercially available spectral photon-counting mammography system. The technique was evaluated using computer simulations, phantom experiments, and analysis of screening mammograms. The CNR benefit of energy weighting for a number of relevant target-background combinations measured by the three methods fell in the range of 2.2 to 5.2% when using optimal weight factors. This translates to a potential dose reduction at constant CNR in the range of 4.5 to 11%. We expect the choice of weight factor in practical implementations to be straightforward because (1) the CNR improvement was not very sensitive to weight, (2) the optimal weight was similar for all investigated target-background combinations, (3) aluminum/PMMA phantoms were found to represent clinically relevant tasks well, and (4) the optimal weight could be calculated directly from pixel values in phantom images. Reasonable agreement was found between the simulations and phantom measurements. Manual measurements on microcalcifications and automatic image analysis confirmed that the CNR improvement was detectable in energy-weighted screening mammograms.

The evaluation of medical imaging devices often involves studies that measure the ability of observers to perform a signal detection task on images obtained from those devices. Data from such studies are frequently regressed ordinally using two-sample receiver operating characteristic (ROC) models. We applied some of these models to a number of randomly chosen data sets from medical imaging and evaluated how well they fit using the Akaike and Bayesian information criteria and cross-validation. We find that for many observer data sets, a single-parameter model is sufficient and that only some studies exhibit evidence for the use of models with more than a single parameter. In particular, the single-parameter power-law model frequently well describes observer data. The power-law model has an asymmetric ROC curve and a constant mean-to-sigma ratio seen in studies analyzed with the bi-normal model. It is identical or very similar to special cases of other two-parameter models.

Early diagnoses of Alzheimer’s disease (AD) would confer many benefits. Several biomarkers have been proposed to achieve such a task, where features extracted from magnetic resonance imaging (MRI) have played an important role. However, studies have focused exclusively on morphological characteristics. This study aims to determine whether features relating to the signal and texture of the image could predict mild cognitive impairment (MCI) to AD progression. Clinical, biological, and positron emission tomography information and MRI images of 62 subjects from the AD neuroimaging initiative were used in this study, extracting 4150 features from each MRI. Within this multimodal database, a feature selection algorithm was used to obtain an accurate and small logistic regression model, generated by a methodology that yielded a mean blind test accuracy of 0.79. This model included six features, five of them obtained from the MRI images, and one obtained from genotyping. A risk analysis divided the subjects into low-risk and high-risk groups according to a prognostic index. The groups were statistically different (p-value=2.04e−11). These results demonstrated that MRI features related to both signal and texture add MCI to AD predictive power, and supported the ongoing notion that multimodal biomarkers outperform single-modality ones.

Modeling and simulation are often used to understand and investigate random quantities and estimators. In 1997, Roe and Metz introduced a simulation model to validate analysis methods for the popular endpoint in reader studies to evaluate medical imaging devices, the reader-averaged area under the receiver operating characteristic (ROC) curve. Here, we generalize the notation of the model to allow more flexibility in recognition that variances of ROC ratings depend on modality and truth state. We also derive and validate equations for computing population variances and covariances for reader-averaged empirical AUC estimates under the generalized model. The equations are one-dimensional integrals that can be calculated using standard numerical integration techniques. This work provides the theoretical foundation and validation for a Java application called iRoeMetz that can simulate multireader multicase ROC studies and numerically calculate the corresponding variances and covariances of the empirical AUC. The iRoeMetz application and source code can be found at the “iMRMC” project on the google code project hosting site. These results and the application can be used by investigators to investigate ROC endpoints, validate analysis methods, and plan future studies.

One of the challenges for iterative image reconstruction (IIR) is that such algorithms solve an imaging model implicitly, requiring a complete representation of the scanned subject within the viewing domain of the scanner. This requirement can place a prohibitively high computational burden for IIR applied to x-ray computed tomography (CT), especially when high-resolution tomographic volumes are required. In this work, we aim to develop an IIR algorithm for direct region-of-interest (ROI) image reconstruction. The proposed class of IIR algorithms is based on an optimization problem that incorporates a data fidelity term, which compares a derivative of the estimated data with the available projection data. In order to characterize this optimization problem, we apply it to computer-simulated two-dimensional fan-beam CT data, using both ideal noiseless data and realistic data containing a level of noise comparable to that of the breast CT application. The proposed method is demonstrated for both complete field-of-view and ROI imaging. To demonstrate the potential utility of the proposed ROI imaging method, it is applied to actual CT scanner data.

In cancer treatment, it is highly desirable to classify single cancer cells in real time. The standard method is polymerase chain reaction requiring a substantial amount of resources and time. Here, we present an innovative approach for rapidly classifying different cell types: we measure the diffraction pattern of a single cell illuminated with coherent extreme ultraviolet (XUV) laser-generated radiation. These patterns allow distinguishing different breast cancer cell types in a subsequent step. Moreover, the morphology of the object can be retrieved from the diffraction pattern with submicron resolution. In a proof-of-principle experiment, we prepared single MCF7 and SKBR3 breast cancer cells on gold-coated silica slides. The output of a laser-driven XUV light source is focused onto a single unstained and unlabeled cancer cell. With the resulting diffraction pattern, we could clearly identify the different cell types. With an improved setup, it will not only be feasible to classify circulating tumor cells with a high throughput, but also to identify smaller objects such as bacteria or even viruses.

We statistically compare the contributions of parenchymal phenotypes to mammographic density in distinguishing between high-risk cases and low-risk controls. The age-matched evaluation included computerized mammographic assessment of breast percent density (PD) and parenchymal patterns (phenotypes of coarseness and contrast) from radiographic texture analysis (RTA) of the full-field digital mammograms from 456 cases: 53 women with BRCA1/2 gene mutations, 75 with unilateral cancer, and 328 at low risk of developing breast cancer. Image-based phenotypes of parenchymal pattern coarseness and contrast were each found to significantly discriminate between the groups; however, PD did not. From ROC analysis, PD alone yielded area under the fitted ROC curve (AUC) values of 0.53 (SE=0.05) and 0.57 (SE=0.04) in the classification task between BRCA1/2 gene-mutation carriers and low-risk women, and between unilateral cancer and low-risk women, respectively. In a round-robin evaluation with Bayesian artificial neural network (BANN) analysis, RTA yielded AUC values of 0.81 (95% confidence interval [0.71, 0.89]) and 0.70 (95% confidence interval [0.63, 0.77]) between the BRCA1/2 gene-mutation carriers and low-risk women, and between unilateral cancer and low-risk women, respectively. These results show that high-risk and low-risk women have different mammographic parenchymal patterns with significantly higher discrimination resulting from characteristics of the parenchymal patterns than just the breast PD.

We compare several approaches to estimation of Hotelling observer (HO) performance in x-ray computed tomography (CT). We consider the case where the signal of interest is small so that the reconstructed image can be restricted to a small region of interest (ROI) surrounding the signal. This reduces the dimensionality of the image covariance matrix so that direct computation of HO metrics within the ROI is feasible. We propose that this approach is directly applicable to systems optimization in CT; however, many alternative approaches exist, which make computation of HO performance tractable through a range of approximations, assumptions, or estimation strategies. Here, we compare several of these methods, including the use of Laguerre-Gauss channels, discrete Fourier domain computation of the HO (which assumes noise stationarity), and two approaches to HO estimation through samples of noisy images. Since our method computes HO performance exactly within an ROI, this allows us to investigate the validity of the assumptions inherent in various common approaches to HO estimation, such as the stationarity assumption in the case of the discrete Fourier transform domain method.

We treat multireader multicase (MRMC) reader studies for which a reader’s diagnostic assessment is converted to binary agreement (1: agree with the truth state, 0: disagree with the truth state). We present a mathematical model for simulating binary MRMC data with a desired correlation structure across readers, cases, and two modalities, assuming the expected probability of agreement is equal for the two modalities (P₁=P₂). This model can be used to validate the coverage probabilities of 95% confidence intervals (of P₁, P₂, or P₁−P₂ when P₁−P₂=0), validate the type I error of a superiority hypothesis test, and size a noninferiority hypothesis test (which assumes P₁=P₂). To illustrate the utility of our simulation model, we adapt the Obuchowski–Rockette–Hillis (ORH) method for the analysis of MRMC binary agreement data. Moreover, we use our simulation model to validate the ORH method for binary data and to illustrate sizing in a noninferiority setting. Our software package is publicly available on the Google code project hosting site for use in simulation, analysis, validation, and sizing of MRMC reader studies with binary agreement data.

Evaluation of segmentation algorithms usually involves comparisons of segmentations to gold-standard delineations without regard to the ultimate medical decision-making task. We compare two segmentation evaluations methods—a Dice similarity coefficient (DSC) evaluation and a diagnostic classification task–based evaluation method using lesions from breast computed tomography. In our investigation, we use results from two previously developed lesion-segmentation algorithms [a global active contour model (GAC) and a global with local aspects active contour model]. Although similar DSC values were obtained (0.80 versus 0.77), we show that the global + local active contour (GLAC) model, as compared with the GAC model, is able to yield significantly improved classification performance in terms of area under the receivers operating characteristic (ROC) curve in the task of distinguishing malignant from benign lesions. [Area under the ROC curve (AUC)=0.78 compared to 0.63, p≪0.001]. This is mainly because the GLAC model yields better detailed information required in the calculation of morphological features. Based on our findings, we conclude that the DSC metric alone is not sufficient for evaluating segmentation lesions in computer-aided diagnosis tasks.

Robert F. Wagner wrote his first SPIE paper¹ in 1972, within the first year of his joining the Bureau of Radiological Health, the precursor to the FDA’s Center for Devices and Radiological Health. He had been hired to build a laboratory and develop methodologies for assessing the performance of diagnostic x-ray systems, in support of the passage of the Radiation Control for Health and Safety Act. In that first year, Bob met with leading scientists in medical imaging as well as other specialties including vision, communications, and television. He formulated a risk-benefit approach to his work, recognizing that the patient exposure associated with the creation of a medical image needed to be considered in light of the usefulness of that image. Bob’s manuscript, reprinted in this special section of the Journal of Medical Imaging, provided an insightful review of the image quantification field, including modulation transfer functions, Wiener spectra, and the basis for receiver operating characteristic curves, along with a bold statement that laid the foundation for the entire field of medical imaging assessment to follow, that image quality “must be defined in terms of the task that the image is destined to perform.”

The objective of this study was to investigate the improvement in diagnostic quality of an iterative model–based reconstruction (IMBR) algorithm for low-tube-voltage (80-kVp) and low-tube-current in abdominal computed tomography angiography (CTA). A total of 11 patients were imaged on a 256-slice multidetector computed tomography for visualization of the aorta. For all patients, three different reconstructions from the low-tube-voltage data are generated: filtered backprojection (FBP), IMBR, and a mixture of both IMBR+FBP. To determine the diagnostic value of IMBR-based reconstructions, the image quality was assessed. With IMBR-based reconstructions, image noise could be significantly reduced, which was confirmed by a highly improved contrast-to-noise ratio. In the image quality assessment, radiologists were able to reliably detect more third-order and higher aortic branches in the IMBR reconstructions compared to FBP reconstructions. The effective dose level was, on average, 3.0 mSv for 80-kVp acquisitions. Low-tube-voltage CTAs significantly improve vascular contrast as presented by others; however, this effect in combination with IMBR enabled yet another substantial improvement of diagnostic quality. For IMBR, a significant improvement of image quality and a decreased radiation dose at low-tube-voltage can be reported.

Simultaneous positron emission tomography and magnetic resonance imaging (PET-MR) is an innovative and promising imaging modality that is generating substantial interest in the medical imaging community, while offering many challenges and opportunities. In this study, we investigated whether MR surface coils need to be accounted for in PET attenuation correction. Furthermore, we integrated motion correction, attenuation correction, and point spread function modeling into a single PET reconstruction framework. We applied our reconstruction framework to in vivo animal and patient PET-MR studies. We have demonstrated that our approach greatly improved PET image quality.

Diffraction-enhanced imaging (DEI) is an emerging x-ray imaging method that simultaneously yields x-ray attenuation and refraction images and holds great promise for soft-tissue imaging. The DEI has been mainly studied using synchrotron sources, but efforts have been made to transition the technology to more practical implementations using conventional x-ray sources. The main technical challenge of this transition lies in the relatively lower x-ray flux obtained from conventional sources, leading to photon-limited data contaminated by Poisson noise. Several issues that must be understood in order to design and optimize DEI imaging systems with respect to noise performance are addressed. Specifically, we: (a) develop equations describing the noise properties of DEI images, (b) derive the conditions under which the DEI algorithm is statistically optimal, (c) characterize the imaging performance that can be obtained as measured by task-based metrics, and (d) consider image-processing steps that may be employed to mitigate noise effects.

Proper sizing of interventional devices to match coronary vessel dimensions improves procedural efficiency and therapeutic outcomes. We have developed a method that uses an inverse geometry x-ray fluoroscopy system [scanning beam digital x-ray (SBDX)] to automatically determine vessel dimensions from angiograms without the need for magnification calibration or optimal views. For each frame period (1/15th of a second), SBDX acquires a sequence of narrow beam projections and performs digital tomosynthesis at multiple plane positions. A three-dimensional model of the vessel is reconstructed by localizing the depth of the vessel edges from the tomosynthesis images, and the model is used to calculate the length and diameter in units of millimeters. The invivo algorithm performance was evaluated in a healthy porcine model by comparing end-diastolic length and diameter measurements from SBDX to coronary computed tomography angiography (CCTA) and intravascular ultrasound (IVUS), respectively. The length error was −0.49±1.76  mm (SBDX – CCTA, mean±1 SD). The diameter error was 0.07±0.27  mm (SBDX − minimum IVUS diameter, mean±1 SD). The in vivo agreement between SBDX-based vessel sizing and gold standard techniques supports the feasibility of calibration-free coronary vessel sizing using inverse geometry x-ray fluoroscopy.

Scatter contamination of projection images in cone-beam computed tomography (CT) degrades the image quality. The use of bowtie filters in dedicated breast CT can decrease this scatter contribution. Three bowtie filter designs that compensate for one or more aspects of the beam-modifying effects due to differences in path length in a projection were studied. These designs have been investigated in terms of their ability to reduce the scatter contamination in projection images acquired in a dedicated breast CT geometry. The scatter magnitude was measured as the scatter-to-primary ratio (SPR) using experimental and Monte Carlo techniques for various breast phantom diameters and tube voltages. The results show that a 55% reduction in the center SPR value could be obtained with the bowtie filter designs. On average, the bowtie filters reduced the center SPR by approximately 18% over all breast diameters. The distribution of the scatter was calculated at a range of different locations to produce scatter distribution maps for all three bowtie filter designs. With the inclusion of the bowtie filters, the scatter distribution was more uniform for all breast diameters. The results of this study will be useful in designing scatter correction methods and understanding the benefits of bowtie filters in dedicated breast CT.

The objective is to optimize low-energy (LE) and high-energy (HE) exposure parameters of contrast-enhanced spectral mammography (CESM) examinations in four different clinical applications for which different levels of average glandular dose (AGD) and ratios between LE and total doses are required. The optimization was performed on a Senographe DS with a SenoBright® upgrade. Simulations were performed to find the optima by maximizing the contrast-to-noise ratio (CNR) on the recombined CESM image using different targeted doses and LE image quality. The linearity between iodine concentration and CNR as well as the minimal detectable iodine concentration was assessed. The image quality of the LE image was assessed on the CDMAM contrast-detail phantom. Experiments confirmed the optima found on simulation. The CNR was higher for each clinical indication than for SenoBright®, including the screening indication for which the total AGD was 22% lower. Minimal iodine concentrations detectable in the case of a 3-mm-diameter round tumor were 12.5% lower than those obtained for the same dose in the clinical routine. LE image quality satisfied EUREF acceptable limits for threshold contrast. This newly optimized set of acquisition parameters allows increased contrast detectability compared to parameters currently used without a significant loss in LE image quality.

The bowtie filter is an essential element of computed tomography scanners. Implementation of this filter in a Monte Carlo dosimetry platform can be based on Turner’s method, which describes how to measure the filter thickness and relate the x-ray beam as a function of bowtie angle to the central beam. In that application, the beam hardening is accounted for by means of weighting factors that are associated to the photons according to their position (fan angle) and energy. We assessed an alternative approximation in which the photon spectrum is given a fan angle-dependent scaling factor. The aim of our investigation was to evaluate the effects on dose accuracy estimation when using the gold standard bowtie filter method versus a beam scaling approximation method. In particular, we wanted to assess the percentage dose differences between the two methods for several water thicknesses representative for different patients of different body mass index. The largest percentage differences were found for the thickest part of the bowtie filter and increased with patient size.

Brain tissue segmentation on magnetic resonance (MR) imaging is a difficult task because of significant intensity overlap between the tissue classes. We present a new knowledge-driven decision theory (KDT) approach that incorporates prior information of the relative extents of intensity overlap between tissue class pairs for volumetric MR tissue segmentation. The proposed approach better handles intensity overlap between tissues without explicitly employing methods for removal of MR image corruptions (such as bias field). Adaptive tissue class priors are employed that combine probabilistic atlas maps with spatial contextual information obtained from Markov random fields to guide tissue segmentation. The energy function is minimized using a variational level-set-based framework, which has shown great promise for MR image analysis. We evaluate the proposed method on two well-established real MR datasets with expert ground-truth segmentations and compare our approach against existing segmentation methods. KDT has low-computational complexity and shows better segmentation performance than other segmentation methods evaluated using these MR datasets.

We present a discrete compactness (DC) index, together with a classification scheme, based both on the size and shape features extracted from brain volumes, to determine different aging stages: healthy controls (HC), mild cognitive impairment (MCI), and Alzheimer’s disease (AD). A set of 30 brain magnetic resonance imaging (MRI) volumes for each group was segmented and two indices were measured for several structures: three-dimensional DC and normalized volumes (NVs). The discrimination power of these indices was determined by means of the area under the curve (AUC) of the receiver operating characteristic, where the proposed compactness index showed an average AUC of 0.7 for HC versus MCI comparison, 0.9 for HC versus AD separation, and 0.75 for MCI versus AD groups. In all cases, this index outperformed the discrimination capability of the NV. Using selected features from the set of DC and NV measures, three support vector machines were optimized and validated for the pairwise separation of the three classes. Our analysis shows classification rates of up to 98.3% between HC and AD, 85% between HC and MCI, and 93.3% for MCI and AD separation. These results outperform those reported in the literature and demonstrate the viability of the proposed morphological indices to classify different aging stages.

Breast cancer (BCa) grading plays an important role in predicting disease aggressiveness and patient outcome. A key component of BCa grade is the mitotic count, which involves quantifying the number of cells in the process of dividing (i.e., undergoing mitosis) at a specific point in time. Currently, mitosis counting is done manually by a pathologist looking at multiple high power fields (HPFs) on a glass slide under a microscope, an extremely laborious and time consuming process. The development of computerized systems for automated detection of mitotic nuclei, while highly desirable, is confounded by the highly variable shape and appearance of mitoses. Existing methods use either handcrafted features that capture certain morphological, statistical, or textural attributes of mitoses or features learned with convolutional neural networks (CNN). Although handcrafted features are inspired by the domain and the particular application, the data-driven CNN models tend to be domain agnostic and attempt to learn additional feature bases that cannot be represented through any of the handcrafted features. On the other hand, CNN is computationally more complex and needs a large number of labeled training instances. Since handcrafted features attempt to model domain pertinent attributes and CNN approaches are largely supervised feature generation methods, there is an appeal in attempting to combine these two distinct classes of feature generation strategies to create an integrated set of attributes that can potentially outperform either class of feature extraction strategies individually. We present a cascaded approach for mitosis detection that intelligently combines a CNN model and handcrafted features (morphology, color, and texture features). By employing a light CNN model, the proposed approach is far less demanding computationally, and the cascaded strategy of combining handcrafted features and CNN-derived features enables the possibility of maximizing the performance by leveraging the disconnected feature sets. Evaluation on the public ICPR12 mitosis dataset that has 226 mitoses annotated on 35 HPFs (400× magnification) by several pathologists and 15 testing HPFs yielded an F-measure of 0.7345. Our approach is accurate, fast, and requires fewer computing resources compared to existent methods, making this feasible for clinical use.

Dynamic contrast-enhanced computed tomography (CT) could provide an accurate and widely available technique for myocardial blood flow (MBF) estimation to aid in the diagnosis and treatment of coronary artery disease. However, one of its primary limitations is the radiation dose imparted to the patient. We are exploring techniques to reduce the patient dose by either reducing the tube current or by reducing the number of temporal frames in the dynamic CT sequence. Both of these dose reduction techniques result in noisy data. In order to extract the MBF information from the noisy acquisitions, we have explored several data-domain smoothing techniques. In this work, we investigate two specific smoothing techniques: the sinogram restoration technique in both the spatial and temporal domains and the use of the Karhunen–Loeve (KL) transform to provide temporal smoothing in the sinogram domain. The KL transform smoothing technique has been previously applied to dynamic image sequences in positron emission tomography. We apply a quantitative two-compartment blood flow model to estimate MBF from the time-attenuation curves and determine which smoothing method provides the most accurate MBF estimates in a series of simulations of different dose levels, dynamic contrast-enhanced cardiac CT acquisitions. As measured by root mean square percentage error (% RMSE) in MBF estimates, sinogram smoothing generally provides the best MBF estimates except for the cases of the lowest simulated dose levels (tube current=25  mAs, 2 or 3 s temporal spacing), where the KL transform method provides the best MBF estimates. The KL transform technique provides improved MBF estimates compared to conventional processing only at very low doses (<7  mSv). Results suggest that the proposed smoothing techniques could provide high fidelity MBF information and allow for substantial radiation dose savings.

Efficient segmentation editing tools are important components in the segmentation process, as no automatic methods exist that always generate sufficient results. Evaluating segmentation editing algorithms is challenging, because their quality depends on the user’s subjective impression. So far, no established methods for an objective, comprehensive evaluation of such tools exist and, particularly, intermediate segmentation results are not taken into account. We discuss the evaluation of editing algorithms in the context of tumor segmentation in computed tomography. We propose a rating scheme to qualitatively measure the accuracy and efficiency of editing tools in user studies. In order to objectively summarize the overall quality, we propose two scores based on the subjective rating and the quantified segmentation quality over time. Finally, a simulation-based evaluation approach is discussed, which allows a more reproducible evaluation without the need for human input. This automated evaluation complements user studies, allowing a more convincing evaluation, particularly during development, where frequent user studies are not possible. The proposed methods have been used to evaluate two dedicated editing algorithms on 131 representative tumor segmentations. We show how the comparison of editing algorithms benefits from the proposed methods. Our results also show the correlation of the suggested quality score with the qualitative ratings.

The optic nerve (ON) plays a critical role in many devastating pathological conditions. Segmentation of the ON has the ability to provide understanding of anatomical development and progression of diseases of the ON. Recently, methods have been proposed to segment the ON but progress toward full automation has been limited. We optimize registration and fusion methods for a new multi-atlas framework for automated segmentation of the ONs, eye globes, and muscles on clinically acquired computed tomography (CT) data. Briefly, the multi-atlas approach consists of determining a region of interest within each scan using affine registration, followed by nonrigid registration on reduced field of view atlases, and performing statistical fusion on the results. We evaluate the robustness of the approach by segmenting the ON structure in 501 clinically acquired CT scan volumes obtained from 183 subjects from a thyroid eye disease patient population. A subset of 30 scan volumes was manually labeled to assess accuracy and guide method choice. Of the 18 compared methods, the ANTS Symmetric Normalization registration and nonlocal spatial simultaneous truth and performance level estimation statistical fusion resulted in the best overall performance, resulting in a median Dice similarity coefficient of 0.77, which is comparable with inter-rater (human) reproducibility at 0.73.

Standard clinical magnetic resonance imaging (MRI) is acquired in two-dimensions where the in-plane resolution is higher than the slice select direction. These acquisitions include axial, coronal, and sagittal planes. To date, there have been few attempts to combine the information of these three orthogonal orientations. This paper aims to take advantage of the different in-plane resolution acquired from each plane orientation and combine them into one volume in order to attain a higher resolution image. This combination of MRI data will allow the detection of smaller areas that would otherwise be missed using only one slice orientation. A comparison of slice thicknesses along with image registration is performed. The mean-squared error and peak signal-to-noise were computed for quantitative assessment. MRI and phantom scans and joint histograms were used for qualitative assessment.

Hepatocellular carcinoma (HCC) is the most common histological type of primary liver cancer. HCC is graded according to the malignancy of the tissues. It is important to diagnose low-grade HCC tumors because these tissues have good prognosis. Image interpretation-based computer-aided diagnosis (CAD) systems have been developed to automate the HCC grading process. Generally, the HCC grade is determined by the characteristics of liver cell nuclei. Therefore, it is preferable that CAD systems utilize only liver cell nuclei for HCC grading. This paper proposes an automated HCC diagnosing method. In particular, it defines a pipeline-path that excludes nonliver cell nuclei in two consequent pipeline-modules and utilizes the liver cell nuclear features for HCC grading. The significance of excluding the nonliver cell nuclei for HCC grading is experimentally evaluated. Four categories of liver cell nuclear features were utilized for classifying the HCC tumors. Results indicated that nuclear texture is the dominant feature for HCC grading and others contribute to increase the classification accuracy. The proposed method was employed to classify a set of regions of interest selected from HCC whole slide images into five classes and resulted in a 95.97% correct classification rate.

Manual segmentation of anatomy in brain MRI data taken to be the closest to the “gold standard” in quality is often used in automated registration-based segmentation paradigms for transfer of template labels onto the unlabeled MRI images. This study presents a library of template data with 16 subcortical structures in the central brain area which were manually labeled for MRI data from 22 children (8 male, mean age=8±0.6  years). The lateral ventricle, thalamus, caudate, putamen, hippocampus, cerebellum, third vevntricle, fourth ventricle, brainstem, and corpuscallosum were segmented by two expert raters. Cross-validation experiments with randomized template subset selection were conducted to test for their ability to accurately segment MRI data under an automated segmentation pipeline. A high value of the dice similarity coefficient (0.86±0.06, min=0.74, max=0.96) and small Hausdorff distance (3.33±4.24, min=0.63, max=25.24) of the automated segmentation against the manual labels was obtained on this template library data. Additionally, comparison with segmentation obtained from adult templates showed significant improvement in accuracy with the use of an age-matched library in this cohort. A manually delineated pediatric template library such as the one described here could provide a useful benchmark for testing segmentation algorithms.

Accurate and automatic segmentation of the pectoralis muscle is essential in many breast image processing procedures, for example, in the computation of volumetric breast density from digital mammograms. Its segmentation is a difficult task due to the heterogeneity of the region, neighborhood complexities, and shape variability. The segmentation is achieved by pixel classification through a Markov random field (MRF) image model. Using the image intensity feature as observable data and local spatial information as a priori, the posterior distribution is estimated in a stochastic process. With a variable potential component in the energy function, by the maximum a posteriori (MAP) estimate of the labeling image, given the image intensity feature which is assumed to follow a Gaussian distribution, we achieved convergence properties in an appropriate sense by Metropolis sampling the posterior distribution of the selected energy function. By proposing an adjustable spatial constraint, the MRF-MAP model is able to embody the shape requirement and provide the required flexibility for the model parameter fitting process. We demonstrate that accurate and robust segmentation can be achieved for the curving-triangle-shaped pectoralis muscle in the medio-lateral-oblique (MLO) view, and the semielliptic-shaped muscle in cranio-caudal (CC) view digital mammograms. The applicable mammograms can be either “For Processing” or “For Presentation” image formats. The algorithm was developed using 56 MLO-view and 79 CC-view FFDM “For Processing” images, and quantitatively evaluated against a random selection of 122 MLO-view and 173 CC-view FFDM images of both presentation intent types.

Glaucoma is neurodegenerative disease characterized by distinctive changes in the optic nerve head and visual field. Without treatment, glaucoma can lead to permanent blindness. Therefore, monitoring glaucoma progression is important to detect uncontrolled disease and the possible need for therapy advancement. In this context, three-dimensional (3-D) spectral domain optical coherence tomography (SD-OCT) has been commonly used in the diagnosis and management of glaucoma patients. We present a new framework for detection of glaucoma progression using 3-D SD-OCT images. In contrast to previous works that use the retinal nerve fiber layer thickness measurement provided by commercially available instruments, we consider the whole 3-D volume for change detection. To account for the spatial voxel dependency, we propose the use of the Markov random field (MRF) model as a prior for the change detection map. In order to improve the robustness of the proposed approach, a nonlocal strategy was adopted to define the MRF energy function. To accommodate the presence of false-positive detection, we used a fuzzy logic approach to classify a 3-D SD-OCT image into a “non-progressing” or “progressing” glaucoma class. We compared the diagnostic performance of the proposed framework to the existing methods of progression detection.

Laser interstitial thermotherapy (LITT) is a relatively new focal therapy technique for the ablation of localized prostate cancer. In this study, for the first time, we are integrating ex vivo pathology and magnetic resonance imaging (MRI) to assess the imaging characteristics of prostate cancer and treatment changes following LITT. Via a unique clinical trial, which gave us the availability of ex vivo histology and pre- and post-LITT MRIs, (1) we investigated the imaging characteristics of treatment effects and residual disease, and (2) evaluated treatment-induced feature changes in the ablated area relative to the residual disease. First, a pathologist annotated the ablated area and the residual disease on the ex vivo histology. Subsequently, we transferred the annotations to the post-LITT MRI using a semi-automatic elastic registration. The pre- and post-LITT MRIs were registered and features were extracted. A scoring metric based on the change in median pre- and post-LITT feature values was introduced, which allowed us to identify the most treatment responsive features. Our results show that (1) image characteristics for treatment effects and residual disease are different, and (2) the change of feature values between pre- and post-LITT MRIs can be a quantitative biomarker for treatment response. Finally, using feature change improved discrimination between the residual disease and treatment effects.

Brain tumor margin removal is challenging because diseased tissue is often visually indistinguishable from healthy tissue. Leaving residual tumor leads to decreased survival, and removing normal tissue causes life-long neurological deficits. Thus, a surgical robotics system with a high degree of dexterity, accurate navigation, and highly precise resection is an ideal candidate for image-guided removal of fluorescently labeled brain tumor cells. To image, we developed a scanning fiber endoscope (SFE) which acquires concurrent reflectance and fluorescence wide-field images at a high resolution. This miniature flexible endoscope was affixed to the arm of a RAVEN II surgical robot providing programmable motion with feedback control using stereo-pair surveillance cameras. To verify the accuracy of the three-dimensional (3-D) reconstructed surgical field, a multimodal physical-sized model of debulked brain tumor was used to obtain the 3-D locations of residual tumor for robotic path planning to remove fluorescent cells. Such reconstruction is repeated intraoperatively during margin clean-up so the algorithm efficiency and accuracy are important to the robotically assisted surgery. Experimental results indicate that the time for creating this 3-D surface can be reduced to one-third by using known trajectories of a robot arm, and the error from the reconstructed phantom is within 0.67 mm in average compared to the model design.

Detecting cancerous lesions is a major clinical application in emission tomography. Previously, we developed a method to design a shift-variant quadratic penalty function in penalized maximum-likelihood (PML) image reconstruction to improve the lesion detectability. We used a multiview channelized Hotelling observer (mvCHO) to assess the lesion detectability in three-dimensional images and validated the penalty design using computer simulations. In this study, we evaluate the benefit of the proposed penalty function for lesion detection using real patient data and artificial lesions. A high-count real patient dataset with no identifiable tumor inside the field of view is used as the background data. A Na-22 point source is scanned in air at variable locations and the point source data are superimposed onto the patient data as artificial lesions after being attenuated by the patient body. Independent Poisson noise is introduced to the high-count sinograms to generate 200 pairs of lesion-present and lesion-absent datasets, each mimicking a 5-min scan. Lesion detectability is assessed using a mvCHO and a human observer two-alternative forced choice (2AFC) experiment. The results show improvements in lesion detection by the proposed method compared with the conventional first-order quadratic penalty function and a total variation (TV) edge-preserving penalty function.

Medical imaging is an effective technique used to detect and prevent disease in cancer research. To optimize medical imaging, a calibration medium or phantom with tissue-mimicking properties is required. Although the feasibility of various polymer gel materials has previously been studied, the stability of the gels’ properties has not been investigated. In this study, we fabricated carrageenan-based polymer gel to examine the stability of its properties such as density, conductivity, permittivity, elastic modulus, and T₁ and T₂ relaxation times over six weeks. We fabricated eight samples with different carrageenan and agar concentrations and found that the density, elastic modulus, and compressive strength fluctuated with no specific pattern. The elastic modulus in sample 4 with 3 wt. % carrageenan and 1.5 wt. % agar fluctuated from 0.51 to 0.64 MPa in five weeks. The T₁ and T₂ relaxation times also varied by 23% to 29%. We believe that the fluctuation of these properties is related to the change in water content of the sample due to cycles of water expulsion and absorption in their containers. The fluctuation of the properties should be minimized to achieve accurate calibration over the shelf life of the phantom and to serve as the standard for quality assurance. Furthermore, a full liver phantom with spherical lesion particles was fabricated to demonstrate the potential for phantom production.

Radiology practice is based on the implicit assumption that the preference for a particular presentation mode goes hand in hand with superior performance. The present experiment tests this assumption in what pertains to image size. Forty-three radiologists were asked to identify intracranial hemorrhages on 20 cranial computed tomography scans in two image sizes, 14×14 and 28×28  cm. They were asked to indicate which size they preferred and subsequently rated each size on a continuous scale in terms of how much they liked them. The results show no correlation between the jackknife free-response receiver operating characteristic figure of merit and preference rated on a continuous scale (large image: r=0.14, p=0.38; small images: r=0.14, p=0.39). Similarly, there was no significant correlation between the time a radiologist took to read a case and preference rated on the continuous scale (large image: r=−0.07, p=0.64; small images: r=−0.04, p=0.80). When dividing radiologists into two groups according to their size preference, there was no significant difference in performance between groups with regard to either large or small images. The results suggest that the preference for an image size and performance with regard to it are not related

Obtaining a “correct” view in echocardiography is a subjective process in which an operator attempts to obtain images conforming to consensus standard views. Real-time objective quantification of image alignment may assist less experienced operators, but no reliable index yet exists. We present a fully automated algorithm for detecting incorrect medial/lateral translation of an ultrasound probe by image analysis. The ability of the algorithm to distinguish optimal from sub-optimal four-chamber images was compared to that of specialists—the current “gold-standard.” The orientation assessments produced by the automated algorithm correlated well with consensus visual assessments of the specialists (r=0.87) and compared favourably with the correlation between individual specialists and the consensus, 0.82±0.09. Each individual specialist’s assessments were within the consensus of other specialists, 75±14% of the time, and the algorithm’s assessments were within the consensus of specialists 85% of the time. The mean discrepancy in probe translation values between individual specialists and their consensus was 0.97±0.87  cm, and between the automated algorithm and specialists’ consensus was 0.92±0.70  cm. This technology could be incorporated into hardware to provide real-time guidance for image optimisation—a potentially valuable tool both for training and quality control.

We present a platform for designing and executing studies that compare pathologists interpreting histopathology of whole slide images (WSIs) on a computer display to pathologists interpreting glass slides on an optical microscope. eeDAP is an evaluation environment for digital and analog pathology. The key element in eeDAP is the registration of the WSI to the glass slide. Registration is accomplished through computer control of the microscope stage and a camera mounted on the microscope that acquires real-time images of the microscope field of view (FOV). Registration allows for the evaluation of the same regions of interest (ROIs) in both domains. This can reduce or eliminate disagreements that arise from pathologists interpreting different areas and focuses on the comparison of image quality. We reduced the pathologist interpretation area from an entire glass slide (10 to 30  mm²) to small ROIs (<50  μm²). We also made possible the evaluation of individual cells. We summarize eeDAP’s software and hardware and provide calculations and corresponding images of the microscope FOV and the ROIs extracted from the WSIs. The eeDAP software can be downloaded from the Google code website (project: eeDAP) as a MATLAB source or as a precompiled stand-alone license-free application.