Edge illumination (EI) is an x-ray phase-contrast imaging technique, exploiting sensitivity to x-ray refraction to visualize features, which are often not detected by conventional absorption-based radiography. The method does not require a high degree of spatial coherence and is achromatic and, therefore, can be implemented with both synchrotron radiation and commercial x-ray tubes. Using different retrieval algorithms, information about an object’s attenuation, refraction, and scattering properties can be obtained. In recent years, a theoretical framework has been developed that enables EI computed tomography (CT) and, hence, three-dimensional imaging. This review provides a summary of these advances, covering the development of different image acquisition schemes, retrieval approaches, and applications. These developments constitute an integral part in the transformation of EI CT into a widely spread imaging tool for use in a range of fields.

This guest editorial introduces and summarizes the JMI Special Section on Radiomics and Deep Learning.

Accurate automatic segmentation of the prostate in magnetic resonance images (MRI) is a challenging task due to the high variability of prostate anatomic structure. Artifacts such as noise and similar signal intensity of tissues around the prostate boundary inhibit traditional segmentation methods from achieving high accuracy. We investigate both patch-based and holistic (image-to-image) deep-learning methods for segmentation of the prostate. First, we introduce a patch-based convolutional network that aims to refine the prostate contour which provides an initialization. Second, we propose a method for end-to-end prostate segmentation by integrating holistically nested edge detection with fully convolutional networks. Holistically nested networks (HNN) automatically learn a hierarchical representation that can improve prostate boundary detection. Quantitative evaluation is performed on the MRI scans of 250 patients in fivefold cross-validation. The proposed enhanced HNN model achieves a mean ± standard deviation. A Dice similarity coefficient (DSC) of 89.77%±3.29% and a mean Jaccard similarity coefficient (IoU) of 81.59%±5.18% are used to calculate without trimming any end slices. The proposed holistic model significantly (p<0.001) outperforms a patch-based AlexNet model by 9% in DSC and 13% in IoU. Overall, the method achieves state-of-the-art performance as compared with other MRI prostate segmentation methods in the literature.

We explore noninvasive biomarkers of microvascular invasion (mVI) in patients with hepatocellular carcinoma (HCC) using quantitative and semantic image features extracted from contrast-enhanced, triphasic computed tomography (CT). Under institutional review board approval, we selected 28 treatment-naive HCC patients who underwent surgical resection. Four radiologists independently selected and delineated tumor margins on three axial CT images and extracted computational features capturing tumor shape, image intensities, and texture. We also computed two types of “delta features,” defined as the absolute difference and the ratio computed from all pairs of imaging phases for each feature. 717 arterial, portal-venous, delayed single-phase, and delta-phase features were robust against interreader variability (concordance correlation≥0.8). An enhanced cross-validation analysis showed that combining robust single-phase and delta features in the arterial and venous phases identified mVI (AUC 0.76±0.18). Compared to a previously reported semantic feature signature (AUC 0.47 to 0.58), these features in our cohort showed only slight to moderate agreement (Cohen’s kappa range: 0.03 to 0.59). Though preliminary, quantitative analysis of image features in arterial and venous phases may be potential surrogate biomarkers for mVI in HCC. Further study in a larger cohort is warranted.

To evaluate deep learning in the assessment of breast cancer risk in which convolutional neural networks (CNNs) with transfer learning are used to extract parenchymal characteristics directly from full-field digital mammographic (FFDM) images instead of using computerized radiographic texture analysis (RTA), 456 clinical FFDM cases were included: a “high-risk” BRCA1/2 gene-mutation carriers dataset (53 cases), a “high-risk” unilateral cancer patients dataset (75 cases), and a “low-risk dataset” (328 cases). Deep learning was compared to the use of features from RTA, as well as to a combination of both in the task of distinguishing between high- and low-risk subjects. Similar classification performances were obtained using CNN [area under the curve (AUC)=0.83; standard error (SE)=0.03] and RTA (AUC=0.82; SE=0.03) in distinguishing BRCA1/2 carriers and low-risk women. However, in distinguishing unilateral cancer patients and low-risk women, performance was significantly greater with CNN (AUC=0.82; SE=0.03) compared to RTA (AUC=0.73; SE=0.03). Fusion classifiers performed significantly better than the RTA-alone classifiers with AUC values of 0.86 and 0.84 in differentiating BRCA1/2 carriers from low-risk women and unilateral cancer patients from low-risk women, respectively. In conclusion, deep learning extracted parenchymal characteristics from FFDMs performed as well as, or better than, conventional texture analysis in the task of distinguishing between cancer risk populations.

While lung cancer is the second most diagnosed form of cancer in men and women, a sufficiently early diagnosis can be pivotal in patient survival rates. Imaging-based, or radiomics-driven, detection methods have been developed to aid diagnosticians, but largely rely on hand-crafted features that may not fully encapsulate the differences between cancerous and healthy tissue. Recently, the concept of discovery radiomics was introduced, where custom abstract features are discovered from readily available imaging data. We propose an evolutionary deep radiomic sequencer discovery approach based on evolutionary deep intelligence. Motivated by patient privacy concerns and the idea of operational artificial intelligence, the evolutionary deep radiomic sequencer discovery approach organically evolves increasingly more efficient deep radiomic sequencers that produce significantly more compact yet similarly descriptive radiomic sequences over multiple generations. As a result, this framework improves operational efficiency and enables diagnosis to be run locally at the radiologist’s computer while maintaining detection accuracy. We evaluated the evolved deep radiomic sequencer (EDRS) discovered via the proposed evolutionary deep radiomic sequencer discovery framework against state-of-the-art radiomics-driven and discovery radiomics methods using clinical lung CT data with pathologically proven diagnostic data from the LIDC-IDRI dataset. The EDRS shows improved sensitivity (93.42%), specificity (82.39%), and diagnostic accuracy (88.78%) relative to previous radiomics approaches.

Optical coherence tomography (OCT) yields microscopic volumetric images representing tissue structures based on the contrast provided by elastic light scattering. Multipatient studies using OCT for detection of tissue abnormalities can lead to large datasets making quantitative and unbiased assessment of classification algorithms performance difficult without the availability of automated analytical schemes. We present a mathematical descriptor reducing the dimensionality of a classifier’s input data, while preserving essential volumetric features from reconstructed three-dimensional optical volumes. This descriptor is used as the input of classification algorithms allowing a detailed exploration of the features space leading to optimal and reliable classification models based on support vector machine techniques. Using imaging dataset of paraffin-embedded tissue samples from 38 ovarian cancer patients, we report accuracies for cancer detection <90% for binary classification between healthy fallopian tube and ovarian samples containing cancer cells. Furthermore, multiples classes of statistical models are presented demonstrating <70% accuracy for the detection of high-grade serous, endometroid, and clear cells cancers. The classification approach reduces the computational complexity and needed resources to achieve highly accurate classification, making it possible to contemplate other applications, including intraoperative surgical guidance, as well as other depth sectioning techniques for fresh tissue imaging.

Prostate cancer is a leading cause of cancer-related death among men. Multiparametric magnetic resonance imaging has become an essential part of the diagnostic evaluation of prostate cancer. The internationally accepted interpretation scheme (Pi-Rads v2) has different algorithms for scoring of the transition zone (TZ) and peripheral zone (PZ) of the prostate as tumors can appear different in these zones. Computer-aided detection tools have shown different performances in TZ and PZ and separating these zones for training and detection is essential. The TZ-PZ segmentation which requires the segmentation of prostate whole gland and TZ is typically done manually. We present a fully automatic algorithm for delineation of the prostate gland and TZ in diffusion-weighted imaging (DWI) via a stack of fully convolutional neural networks. The proposed algorithm first detects the slices that contain a portion of prostate gland within the three-dimensional DWI volume and then it segments the prostate gland and TZ automatically. The segmentation stage of the algorithm was applied to DWI images of 104 patients and median Dice similarity coefficients of 0.93 and 0.88 were achieved for the prostate gland and TZ, respectively. The detection of image slices with and without prostate gland had an average accuracy of 0.97.

A three-dimensional (3-D) convolutional neural network (CNN) trained from scratch is presented for the classification of pulmonary nodule malignancy from low-dose chest CT scans. Recent approval of lung cancer screening in the United States provides motivation for determining the likelihood of malignancy of pulmonary nodules from the initial CT scan finding to minimize the number of follow-up actions. Classifier ensembles of different combinations of the 3-D CNN and traditional machine learning models based on handcrafted 3-D image features are also explored. The dataset consisting of 326 nodules is constructed with balanced size and class distribution with the malignancy status pathologically confirmed. The results show that both the 3-D CNN single model and the ensemble models with 3-D CNN outperform the respective counterparts constructed using only traditional models. Moreover, complementary information can be learned by the 3-D CNN and the conventional models, which together are combined to construct an ensemble model with statistically superior performance compared with the single traditional model. The performance of the 3-D CNN model demonstrates the potential for improving the lung cancer screening follow-up protocol, which currently mainly depends on the nodule size.

Convolutional neural networks (CNNs), the state of the art in image classification, have proven to be as effective as an ophthalmologist, when detecting referable diabetic retinopathy. Having a size of &lt;1% of the total image, microaneurysms are early lesions in diabetic retinopathy that are difficult to classify. A model that includes two CNNs with different input image sizes, 60×60 and 420×420  pixels, was developed. These models were trained using the Kaggle and Messidor datasets and tested independently against the Kaggle dataset, showing a sensitivity &gt;91%, a specificity &gt;93%, and an area under the receiver operating characteristics curve &gt;93%. Furthermore, by combining these trained models, there was a reduction of false positives for complete images by about 50% and a sensitivity of 96% when tested against the DiaRetDB1 dataset. In addition, a powerful image preprocessing procedure was implemented, improving not only images for annotations, but also decreasing the number of epochs during training. Finally, a feedback method was developed increasing the accuracy of the CNN 420×420  pixel input model.

Duchenne muscular dystrophy (DMD) is a childhood-onset neuromuscular disease that results in the degeneration of muscle, starting in the extremities, before progressing to more vital areas, such as the lungs. Respiratory failure and pneumonia due to respiratory muscle weakness lead to hospitalization and early mortality. However, tracking the disease in this region can be difficult, as current methods are based on breathing tests and are incapable of distinguishing between muscle involvements. Cine MRI scans give insight into respiratory muscle movements, but the images suffer due to low spatial resolution and poor signal-to-noise ratio. Thus, a robust lung segmentation method is required for accurate analysis of the lung and respiratory muscle movement. We deployed a deep learning approach that utilizes sequence-specific prior information to assist the segmentation of lung in cine MRI. More specifically, we adopt a holistically nested network to conduct image-to-image holistic training and prediction. One frame of the cine MRI is used in the training and applied to the remainder of the sequence ( &gt;200 frames). We applied this method to cine MRIs of the lung in the axial, sagittal, and coronal planes. Characteristic lung motion patterns during the breathing cycle were then derived from the segmentations and used for diagnosis. Our data set consisted of 31 young boys, age 9.6±2.3 years, 15 of whom suffered from DMD. The remaining 16 subjects were age-matched healthy volunteers. For validation, slices from inspiratory and expiratory cycles were manually segmented and compared with results obtained from our method. The Dice similarity coefficient for the deep learning-based method was 97.2±1.3 for the sagittal view, 96.1±3.8 for the axial view, and 96.6±1.7 for the coronal view. The holistic neural network approach was compared with an approach using Demon’s registration and showed superior performance. These results suggest that the deep learning-based method reliably and accurately segments the lung across the breathing cycle.

The work explores the use of denoising autoencoders (DAEs) for brain lesion detection, segmentation, and false-positive reduction. Stacked denoising autoencoders (SDAEs) were pretrained using a large number of unlabeled patient volumes and fine-tuned with patches drawn from a limited number of patients ( n=20, 40, 65). The results show negligible loss in performance even when SDAE was fine-tuned using 20 labeled patients. Low grade glioma (LGG) segmentation was achieved using a transfer learning approach in which a network pretrained with high grade glioma data was fine-tuned using LGG image patches. The networks were also shown to generalize well and provide good segmentation on unseen BraTS 2013 and BraTS 2015 test data. The manuscript also includes the use of a single layer DAE, referred to as novelty detector (ND). ND was trained to accurately reconstruct nonlesion patches. The reconstruction error maps of test data were used to localize lesions. The error maps were shown to assign unique error distributions to various constituents of the glioma, enabling localization. The ND learns the nonlesion brain accurately as it was also shown to provide good segmentation performance on ischemic brain lesions in images from a different database.

In the H-scan analysis and display, visualization of different scattering sizes and types is enabled by a matched filter approach involving different orders of Gaussian weighted Hermite functions. An important question with respect to clinical applications involves the change in H-scan outputs with respect to small changes in scatterer sizes. The sensitivity of H-scan outputs is analyzed using the theory of backscatter from a compressible sphere. Experimental corroboration is established using mono dispersed spherical scatterers in phantoms. With a 6-MHz center frequency broadband transducer, it is possible to visualize changes in scattering size in the order of 10 to 15  μm in phantoms and also changes in ex vivo bovine liver tissue due to edema caused by hypotonic perfusion.

An ultra-high resolution (UHR) mode, with a detector pixel size of 0.25 mm×0.25 mm relative to isocenter, has been implemented on a whole body research photon-counting detector (PCD) computed tomography (CT) system. Twenty synthetic lung nodules were scanned using UHR and conventional resolution (macro) modes and reconstructed with medium and very sharp kernels. Linear regression was used to compare measured nodule volumes from CT images to reference volumes. The full-width-at-half-maximum of the calculated curvature histogram for each nodule was used as a shape index, and receiver operating characteristic analysis was performed to differentiate sphere- and star-shaped nodules. Results showed a strong linear relationship between measured nodule volumes and reference volumes for both modes. The overall volume estimation was more accurate using UHR mode and the very sharp kernel, having 4.8% error compared with 10.5% to 12.6% error in the macro mode. The improvement in volume measurements using the UHR mode was more evident for small nodule sizes or star-shaped nodules. Images from the UHR mode with the very sharp kernel consistently demonstrated the best performance [ AUC=(0.839,0.867)] for separating star- from sphere-shaped nodules, showing advantages of UHR mode on a PCD CT scanner for lung nodule characterization.

X-ray phase-contrast imaging (XPCI) overcomes the problem of low contrast between different soft tissues achieved in conventional x-ray imaging by introducing x-ray phase as an additional contrast mechanism. This work describes a compact x-ray light source (CXLS) and compares, via simulations, the high quality XPCI results that can be produced from this source to those produced using a microfocus x-ray source. The simulation framework is first validated using an image acquired with a microfocus-source, propagation-based XPCI (PB-XPCI) system. The phase contrast for a water sphere simulating a simple cyst submersed in muscle is evaluated and the evolution of PB-XPCI signal as the object to detector distance is increased is demonstrated. The proposed design of a PB-XPCI system using the CXLS is described and simulated images of a coronary artery compared between CXLS and microfocus source PB-XPCI systems. To generate images with similar noise levels, a microfocus source would require a 3000 times longer exposure than would the CXLS. We conclude that CXLS technology has the potential to provide high-quality XPCI in a medical environment using extremely short exposure times relative to microfocus source approaches.

Multicolor fluorescence in situ hybridization (M-FISH) is a multichannel imaging technique for rapid detection of chromosomal abnormalities. It is a critical and challenging step to segment chromosomes from M-FISH images toward better chromosome classification. Recently, several fuzzy C-means (FCM) clustering-based methods have been proposed for M-FISH image segmentation or classification, e.g., adaptive fuzzy C-means (AFCM) and improved AFCM (IAFCM), but most of these methods used only one channel imaging information with limited accuracy. To improve the segmentation for better accuracy and more robustness, we proposed an FCM clustering-based method, denoted by spatial- and spectral-FCM. Our method has the following advantages: (1) it is able to exploit information from neighboring pixels (spatial information) to reduce the noise and (2) it can incorporate pixel information across different channels simultaneously (spectral information) into the model. We evaluated the performance of our method by comparing with other FCM-based methods in terms of both accuracy and false-positive detection rate on synthetic, hybrid, and real images. The comparisons on 36 M-FISH images have shown that our proposed method results in higher segmentation accuracy (0.9382±0.0250) and a lower false-positive ratio (0.1042±0.1481) than conventional FCM (accuracy: 0.9210±0.0457, and false-positive ratio: 0.1389±0.1899) and the IAFCM (accuracy: 0.8730±0.2438 and false-positive ratio: 0.2438±0.2438) methods by incorporating both spatial and spectral information from M-FISH images.

Region of interest (RoI) alignment in medical images plays a crucial role in diagnostics, procedure planning, treatment, and follow-up. Frequently, a model is represented as triangulated mesh while the patient data is provided from computed axial tomography scanners as pixel or voxel data. Previously, we presented a 2-D method for curve-to-pixel registration. This paper contributes (i) a general mesh-to-raster framework to register RoIs in multimodal images; (ii) a 3-D surface-to-voxel application, and (iii) a comprehensive quantitative evaluation in 2-D using ground truth (GT) provided by the simultaneous truth and performance level estimation (STAPLE) method. The registration is formulated as a minimization problem, where the objective consists of a data term, which involves the signed distance function of the RoI from the reference image and a higher order elastic regularizer for the deformation. The evaluation is based on quantitative light-induced fluoroscopy (QLF) and digital photography (DP) of decalcified teeth. STAPLE is computed on 150 image pairs from 32 subjects, each showing one corresponding tooth in both modalities. The RoI in each image is manually marked by three experts (900 curves in total). In the QLF-DP setting, our approach significantly outperforms the mutual information-based registration algorithm implemented with the Insight Segmentation and Registration Toolkit and Elastix.

We propose an image analysis method for quality evaluation of human pluripotent stem cells based on biologically interpretable features. It is important to maintain the undifferentiated state of induced pluripotent stem cells (iPSCs) while culturing the cells during propagation. Cell culture experts visually select good quality cells exhibiting the morphological features characteristic of undifferentiated cells. Experts have empirically determined that these features comprise prominent and abundant nucleoli, less intercellular spacing, and fewer differentiating cellular nuclei. We quantified these features based on experts’ visual inspection of phase contrast images of iPSCs and found that these features are effective for evaluating iPSC quality. We then developed an iPSC quality evaluation method using an image analysis technique. The method allowed accurate classification, equivalent to visual inspection by experts, of three iPSC cell lines.

Psoriasis is a chronic skin disease that is assessed visually by dermatologists. The Psoriasis Area and Severity Index (PASI) is the current gold standard used to measure lesion severity by evaluating four parameters, namely, area, erythema, scaliness, and thickness. In this context, psoriasis skin lesion segmentation is required as the basis for PASI scoring. An automatic lesion segmentation method by leveraging multiscale superpixels and K-means clustering is outlined. Specifically, we apply a superpixel segmentation strategy on CIE- L*a*b* color space using different scales. Also, we suppress the superpixels that belong to nonskin areas. Once similar regions on different scales are obtained, the K-means algorithm is used to cluster each superpixel scale separately into normal and lesion skin areas. Features from both a* and b* color bands are used in the clustering process. Furthermore, majority voting is performed to fuse the segmentation results from different scales to obtain the final output. The proposed method is extensively evaluated on a set of 457 psoriasis digital images, acquired from the Royal Melbourne Hospital, Melbourne, Australia. Experimental results have shown evidence that the method is very effective and efficient, even when applied to images containing hairy skin and diverse lesion size, shape, and severity. It has also been ascertained that CIE- L*a*b* outperforms other color spaces for psoriasis lesion analysis and segmentation. In addition, we use three evaluation metrics, namely, Dice coefficient, Jaccard index, and

Medical image registration establishes a correspondence between images of biological structures, and it is at the core of many applications. Commonly used deformable image registration methods depend on a good preregistration initialization. We develop a learning-based method to automatically find a set of robust landmarks in three-dimensional MR image volumes of the head. These landmarks are then used to compute a thin plate spline-based initialization transformation. The process involves two steps: (1) identifying a set of landmarks that can be reliably localized in the images and (2) selecting among them the subset that leads to a good initial transformation. To validate our method, we use it to initialize five well-established deformable registration algorithms that are subsequently used to register an atlas to MR images of the head. We compare our proposed initialization method with a standard approach that involves estimating an affine transformation with an intensity-based approach. We show that for all five registration algorithms the final registration results are statistically better when they are initialized with the method that we propose than when a standard approach is used. The technique that we propose is generic and could be used to initialize nonrigid registration algorithms for other applications.

Changes in arterial wall perfusion mark the onset of atherosclerosis. A characteristic change is the increased spatial density of vasa vasorum (VV), the microvessels in the arterial walls. Measuring this increased VV (IVV) density using contrast-enhanced computed tomography (CT) has had limited success due to blooming effects from contrast media. If the system point-spread function (PSF) is known, then the blooming effect can be modeled as a convolution between the true signal and the PSF. We report the application of image deconvolution to improve the CT number accuracy in the arterial wall of a phantom and in a porcine model of IVV density, both scanned using a whole-body research photon-counting CT scanner. A 3D-printed carotid phantom filled with three concentrations of iodinated contrast material was scanned to assess blooming and its effect on wall CT number accuracy. The results showed a reduction in blooming effects following image deconvolution, and, consequently, a better delineation between lumen and wall was achieved. Results from the animal experiment showed improved CT number difference between the carotid with IVV density and the normal carotid artery after deconvolution, enabling the detection of VV proliferation, which may serve as an early indicator of atherosclerosis.

Cochlear implants (CIs) use electrode arrays that are surgically inserted into the cochlea to stimulate frequency-mapped nerve endings to treat patients with hearing loss. CIs are programmed postoperatively by audiologists using behavioral tests without information on electrode–cochlea spatial relationship. We have recently developed techniques to segment the intracochlear anatomy and to localize individual contacts in clinically acquired computed tomography (CT) images. Using this information, we have proposed a programming strategy that we call image-guided CI programming (IGCIP), and we have shown that it significantly improves outcomes for both adult and pediatric recipients. One obstacle to large-scale deployment of this technique is the need for manual intervention in some processing steps. One of these is the rough registration of images prior to the use of automated intensity-based algorithms. Although seemingly simple, the heterogeneity of our image set makes this task challenging. We propose a solution that relies on the automated random forest-based localization of multiple landmarks used to estimate an initial transformation with a point-based registration method. Results show that it produces results that are equivalent to a manual initialization. This work is an important step toward the full automation of IGCIP.

Down syndrome is one of the most common genetic disorders caused by chromosome abnormalities in humans. Among other physical characteristics, certain facial features are typically associated in people with Down syndrome. We investigate the problem of Down syndrome detection from a collection of face images. As the main contribution, a compact geometric descriptor is used to extract facial features from the images. Experiments are conducted on an available dataset to demonstrate the performance of the proposed methodology.

We investigate the addition of symmetry and temporal context information to a deep convolutional neural network (CNN) with the purpose of detecting malignant soft tissue lesions in mammography. We employ a simple linear mapping that takes the location of a mass candidate and maps it to either the contralateral or prior mammogram, and regions of interest (ROIs) are extracted around each location. Two different architectures are subsequently explored: (1) a fusion model employing two datastreams where both ROIs are fed to the network during training and testing and (2) a stagewise approach where a single ROI CNN is trained on the primary image and subsequently used as a feature extractor for both primary and contralateral or prior ROIs. A “shallow” gradient boosted tree classifier is then trained on the concatenation of these features and used to classify the joint representation. The baseline yielded an AUC of 0.87 with confidence interval [0.853, 0.893]. For the analysis of symmetrical differences, the first architecture where both primary and contralateral patches are presented during training obtained an AUC of 0.895 with confidence interval [0.877, 0.913], and the second architecture where a new classifier is retrained on the concatenation an AUC of 0.88 with confidence interval [0.859, 0.9]. We found a significant difference between the first architecture and the baseline at high specificity with p=0.02. When using the same architectures to analyze temporal change, we yielded an AUC of 0.884 with confidence interval [0.865, 0.902] for the first architecture and an AUC of 0.879 with confidence interval [0.858, 0.898] in the second setting. Although improvements for temporal analysis were consistent, they were not found to be significant. The results show our proposed method is promising and we suspect performance can greatly be improved when more temporal data become available.

This paper presents a local intensity structure analysis based on an intensity targeted radial structure tensor (ITRST) and the blob-like structure enhancement filter based on it (ITRST filter) for the mediastinal lymph node detection algorithm from chest computed tomography (CT) volumes. Although the filter based on radial structure tensor analysis (RST filter) based on conventional RST analysis can be utilized to detect lymph nodes, some lymph nodes adjacent to regions with extremely high or low intensities cannot be detected. Therefore, we propose the ITRST filter, which integrates the prior knowledge on detection target intensity range into the RST filter. Our lymph node detection algorithm consists of two steps: (1) obtaining candidate regions using the ITRST filter and (2) removing false positives (FPs) using the support vector machine classifier. We evaluated lymph node detection performance of the ITRST filter on 47 contrast-enhanced chest CT volumes and compared it with the RST and Hessian filters. The detection rate of the ITRST filter was 84.2% with 9.1 FPs/volume for lymph nodes whose short axis was at least 10 mm, which outperformed the RST and Hessian filters.

Although a lot of work has been done on optical coherence tomography and color images in order to detect and quantify diseases such as diabetic retinopathy, exudates, or neovascularizations, none of them is able to evaluate the diffusion of the neovascularizations in retinas. Our work has been to develop a tool that is able to quantify a neovascularization and the fluorescein leakage during an angiography. The proposed method has been developed following a clinical trial protocol; images are taken by a Spectralis (Heidelberg Engineering). Detections are done using a supervised classification using specific features. Images and their detected neovascularizations are then spatially matched by an image registration. We compute the expansion speed of the liquid that we call diffusion index. This last one specifies the state of the disease, permits indication of the activity of neovascularizations, and allows a follow-up of patients. The method proposed in this paper has been built to be robust, even with laser impacts, to compute a diffusion index.

Currently, histopathological tissue examination by a pathologist represents the gold standard for breast lesion diagnostics. Automated classification of histopathological whole-slide images (WSIs) is challenging owing to the wide range of appearances of benign lesions and the visual similarity of ductal carcinoma in-situ (DCIS) to invasive lesions at the cellular level. Consequently, analysis of tissue at high resolutions with a large contextual area is necessary. We present context-aware stacked convolutional neural networks (CNN) for classification of breast WSIs into normal/benign, DCIS, and invasive ductal carcinoma (IDC). We first train a CNN using high pixel resolution to capture cellular level information. The feature responses generated by this model are then fed as input to a second CNN, stacked on top of the first. Training of this stacked architecture with large input patches enables learning of fine-grained (cellular) details and global tissue structures. Our system is trained and evaluated on a dataset containing 221 WSIs of hematoxylin and eosin stained breast tissue specimens. The system achieves an AUC of 0.962 for the binary classification of nonmalignant and malignant slides and obtains a three-class accuracy of 81.3% for classification of WSIs into normal/benign, DCIS, and IDC, demonstrating its potential for routine diagnostics.

In B-mode imaging of the dependent or compressed breast, wave incidence at steep angles can change propagation directions and induce areas of signal dropout. To evaluate the image anomalies in reasonable simulation times, we performed full-wave studies for center frequencies of 1 and 4 MHz. Speed of sound and density of skin, typical coupling gel, and adipose tissue were assigned to the test couplant. Compared with commercial gel, skin-like couplant reduced the dropout area at 1 and 4 MHz by 57.1% and 96.7%, respectively, consistent with a decreased average beam deflection in the breast. Conversely, the adipose-like couplant increased the dropout area from that of simulated commercial gel by 26.5% and 36.7% at 1 and 4 MHz, respectively. In addition, the skin-like couplant resulted in the greatest beam deflection inside the breast among all couplants. The findings could aid the use of three-dimensional simulations to design ultrasound couplants for beam passage through tissue boundaries at steep angles to improve corrections of signal dropout and defocusing and in compound imaging.

Cochlear implants (CIs) are surgically implantable neuroprosthetic devices used to treat profound hearing loss. Recent literature indicates that there is a correlation between the final intracochlear positioning of the CI electrode arrays and the ultimate hearing outcome of the patient, indicating that further studies to better understand the relationship between electrode position and outcomes could have significant implications for future surgical techniques, array design, and processor programming methods. Postimplantation high-resolution computed tomography (CT) imaging is the best modality for localizing electrodes and provides the resolution necessary to visually identify electrode position, although with an unknown degree of accuracy depending on image acquisition parameters, like the hounsfield unit (HU) range of reconstruction, orientation, radiation dose, and image resolution. We report on the development of a phantom and on its use to study how four acquisition parameters, including image resolution and HU range of reconstruction, affect how accurately the true position of the electrodes can be found in a dataset of CT scans acquired from multiple helical and cone beam scanners. We also show how the phantom can be used to evaluate the effect of acquisition parameters on automatic electrode localization techniques.

As a promising imaging modality, digital breast tomosynthesis (DBT) leads to better diagnostic performance than traditional full-field digital mammograms (FFDM) alone. DBT allows different planes of the breast to be visualized, reducing occlusion from overlapping tissue. Although DBT is gaining popularity, best practices for search strategies in this medium are unclear. Eye tracking allowed us to describe search patterns adopted by radiologists searching DBT and FFDM images. Eleven radiologists examined eight DBT and FFDM cases. Observers marked suspicious masses with mouse clicks. Eye position was recorded at 1000 Hz and was coregistered with slice/depth plane as the radiologist scrolled through the DBT images, allowing a 3-D representation of eye position. Hit rate for masses was higher for tomography cases than 2-D cases and DBT led to lower false positive rates. However, search duration was much longer for DBT cases than FFDM. DBT was associated with longer fixations but similar saccadic amplitude compared with FFDM. When comparing radiologists’ eye movements to a previous study, which tracked eye movements as radiologists read chest CT, we found DBT viewers did not align with previously identified “driller” or “scanner” strategies, although their search strategy most closely aligns with a type of vigorous drilling strategy.

The goal for positron emission tomography (PET)/X is measuring changes in radiotracer uptake for early assessment of response to breast cancer therapy. Upper bounds for detecting such changes were investigated using simulation and two image reconstruction algorithms customized to the PET/X rectangular geometry. Analytical reconstruction was used to study spatial resolution, comparing results with the distance of the closest approach (DCA) resolution surrogate that is independent of the reconstruction method. An iterative reconstruction algorithm was used to characterize contrast recovery in small targets. Resolution averaged &lt;2  mm full width at half maximum when using depth-of-interaction (DOI) information. Without DOI, resolution ranged from 2.1±0.13 to 3.1±0.42  mm for scanner crystal thickness between 5 and 15 mm. The DCA resolution surrogate was highly correlated to image-based FWHM. Receiver-operating characteristic analysis showed specificity and sensitivity over 95% for detecting contrast change from 5:1 to 4:1 (area under curve &gt;99%). For PET/X parameters modeled here, the ability to measure contrast changes benefited from higher photon absorption efficiency of thicker crystals while being largely unaffected by degraded resolution obtained with thicker crystals; DOI provided marginal improvements. These results assumed perfect data corrections and other idealizations, and thus represent an upper bound for detecting changes in small lesion radiotracer uptake of clinical interest using the PET/X system.

Maintaining or even improving image quality while lowering patient dose is always the desire in clinical computed tomography (CT) imaging. Iterative reconstruction (IR) algorithms have been designed to allow for a reduced dose while maintaining or even improving an image. However, we have previously shown that the dose-saving capabilities allowed with IR are different for different clinical tasks. The channelized scanning linear observer (CSLO) was applied to study clinical tasks that combine detection and estimation when assessing CT image data. The purpose of this work is to illustrate the importance of task complexity when assessing dose savings and to move toward more realistic tasks when performing these types of studies. Human-observer validation of these methods will take place in a future publication. Low-contrast objects embedded in body-size phantoms were imaged multiple times and reconstructed by filtered back projection (FBP) and an IR algorithm. The task was to detect, localize, and estimate the size and contrast of low-contrast objects in the phantom. Independent signal-present and signal-absent regions of interest cropped from images were channelized by the dense-difference of Gauss channels for CSLO training and testing. Estimation receiver operating characteristic (EROC) curves and the areas under EROC curves (EAUC) were calculated by CSLO as the figure of merit. The one-shot method was used to compute the variance of the EAUC values. Results suggest that the IR algorithm studied in this work could efficiently reduce the dose by ∼50% while maintaining an image quality comparable to conventional FBP reconstruction warranting further investigation using real patient data.

Cytology, a method of estimating cancer or cellular atypia from microscopic images of scraped specimens, is used according to the pathologist’s experience to diagnose cases based on the degree of structural changes and atypia. Several methods of cell feature quantification, including nuclear size, nuclear shape, cytoplasm size, and chromatin texture, have been studied. We focus on chromatin distribution in the cell nucleus and propose new feature values that indicate the chromatin complexity, spreading, and bias, including convex hull ratio on multiple binary images, intensity distribution from the gravity center, and tangential component intensity and texture biases. The characteristics and cellular classification accuracies of the proposed features were verified through experiments using cervical smear samples, for which clear nuclear morphologic diagnostic criteria are available. In this experiment, we also used a stepwise support vector machine to create a machine learning model and a cross-validation algorithm with which to derive identification accuracy. Our results demonstrate the effectiveness of our proposed feature values.