Metabolic brain imaging with glucosamine CEST MRI: in vivo characterization and first insights.

The utility of chemical exchange saturation transfer (CEST) MRI for monitoring the uptake of glucosamine (GlcN), a safe dietary supplement, has been previously demonstrated in detecting breast cancer in both murine and human subjects. Here, we studied and characterized the detectability of GlcN uptake and metabolism in the brain. Following intravenous GlcN administration in mice, CEST brain signals calculated by magnetization transfer ratio asymmetry (MTRasym) analysis, were significantly elevated, mainly in the cortex, hippocampus, and thalamus. The in vivo contrast remained stable during 40 min of examination, which can be attributed to GlcN uptake and its metabolic products accumulation as confirmed using 13C NMR spectroscopic studies of brain extracts. A Lorentzian multi-pool fitting analysis revealed an increase in the hydroxyl, amide, and relayed nuclear Overhauser effect (rNOE) signal components after GlcN treatment. With its ability to cross the blood-brain barrier (BBB), the GlcN CEST technique has the potential to serve as a metabolic biomarker for the diagnosis and monitoring various brain disorders.

Collaborative-guided spectral abundance learning with bilinear mixing model for hyperspectral subpixel target detection.

Detecting subpixel targets is a considerably challenging issue in hyperspectral image processing and interpretation. Most of the existing hyperspectral subpixel target detection methods construct detectors based on the linear mixing model which regards a pixel as a linear combination of different spectral signatures. However, due to the multiple scattering, the linear mixing model cannot​ illustrate the multiple materials interactions that are nonlinear and widespread in real-world hyperspectral images, which could result in unsatisfactory performance in detecting subpixel targets. To alleviate this problem, this work presents a novel collaborative-guided spectral abundance learning model (denoted as CGSAL) for subpixel target detection based on the bilinear mixing model in hyperspectral images. The proposed CGSAL detects subpixel targets by learning a spectral abundance of the target signature in each pixel. In CGSAL, virtual endmembers and their abundance help to achieve good accuracy for modeling nonlinear scattering accounts for multiple materials interactions according to the bilinear mixing model. Besides, we impose a collaborative term to the spectral abundance learning model to emphasize the collaborative relationships between different endmembers, which contributes to accurate spectral abundance learning and further help to detect subpixel targets. Plentiful experiments and analyses are conducted on three real-world and one synthetic hyperspectral datasets to evaluate the effectiveness of the CGSAL in subpixel target detection. The experiment results demonstrate that the CGSAL achieves competitive performance in detecting subpixel targets and outperforms other state-of-the-art hyperspectral subpixel target detectors.

Flow imaging using differential Golay encoded ultrasound.

In our research we present a new method of differential compression of the Golay encoded ultrasound (DCGEU) in the standard beamforming mode to visualize the slow (<1cm/s) blood mimicking fluid flow in small diameter tubes. The proposed DCGEU method is based on synthesis of several subsequent B-mode frames acquired with certain time intervals (30 ms in this study) followed by the visualization of differential beamformed radio frequency (RF) echoes, which yielded the images of the scatterers moving slowly in the vessel and suppressing the static echoes outside the vessel. In order to extract small backscattered echoes from the vessel area we took an advantage of improved sensitivity of the complementary Golay coded sequences (CGCS). The validation of the proposed DCGEU method was carried out in two stages. In the first one, we compared the flow images in small tubes with a diameter of 1 mm and 2.5 mm, reconstructed from numerically simulated acoustic data for the standard transmission of short pulses and 16-bits long CGCS signals. In the second stage of the research, the experimental data were acquired in a flow phantom with silicone tubes with an internal diameter of 1.5 mm and 4.5 mm and a fluid flow velocity of 0.9 cm/s. The experiments were carried out using preprogrammed Verasonics Vantage™ research ultrasound system equipped with ALT L12-5/50 mm MHz linear array transducer with 7.8 MHz center frequency. It was evidenced both in simulations and experiments that the DCGEU provided a good flow image along the entire length of tubing with virtually angle independent detection in comparison with the conventional short pulse interrogation.

Saturation-pulse prepared heart-rate independent inversion-recovery (SAPPHIRE) biventricular T1 mapping: inter-field strength, head-to-head comparison of diastolic, systolic and dark-blood measurements.

BACKGROUND: To assess the feasibility of biventricular SAPPHIRE T1 mapping in vivo across field strengths using diastolic, systolic and dark-blood (DB) approaches. METHODS: 10 healthy volunteers underwent same-day non-contrast cardiovascular magnetic resonance at 1.5 Tesla (T) and 3 T. Left and right ventricular (LV, RV) T1 mapping was performed in the basal, mid and apical short axis using 4-variants of SAPPHIRE: diastolic, systolic, 0th and 2nd order motion-sensitized DB and conventional modified Look-Locker inversion recovery (MOLLI). RESULTS: LV global myocardial T1 times (1.5 T then 3 T results) were significantly longer by diastolic SAPPHIRE (1283 ± 11|1600 ± 17 ms) than any of the other SAPPHIRE variants: systolic (1239 ± 9|1595 ± 13 ms), 0th order DB (1241 ± 10|1596 ± 12) and 2nd order DB (1251 ± 11|1560 ± 20 ms, all p < 0.05). In the mid septum MOLLI and diastolic SAPPHIRE exhibited significant T1 signal contamination (longer T1) at the blood-myocardial interface not seen with the other 3 SAPPHIRE variants (all p < 0.025). Additionally, systolic, 0th order and 2nd order DB SAPPHIRE showed narrower dispersion of myocardial T1 times across the mid septum when compared to diastolic SAPPHIRE (interquartile ranges respectively: 25 ms, 71 ms, 73 ms vs 143 ms, all p < 0.05). RV T1 mapping was achievable using systolic, 0th and 2nd order DB SAPPHIRE but not with MOLLI or diastolic SAPPHIRE. All 4 SAPPHIRE variants showed excellent re-read reproducibility (intraclass correlation coefficients 0.953 to 0.996). CONCLUSION: These small-scale preliminary healthy volunteer data suggest that DB SAPPHIRE has the potential to reduce partial volume effects at the blood-myocardial interface, and that systolic SAPPHIRE could be a feasible solution for right ventricular T1 mapping. Further work is needed to understand the robustness of these sequences and their potential clinical utility.

Liver iron quantification in children and young adults: comparison of a volumetric multi-echo 3-D Dixon sequence with conventional 2-D T2* relaxometry.

BACKGROUND: Magnetic resonance imaging (MRI)-based liver iron quantification is the standard of care to guide chelation therapy in children at risk of hemochromatosis. T2* relaxometry is the most widely used technique but requires third-party software for post-processing. Vendor-provided three-dimensional (3-D) multi-echo Dixon techniques are now available that allow inline/automated post-processing. OBJECTIVE: The purpose of our study was to evaluate the diagnostic accuracy of a volumetric multi-echo Dixon technique using conventional T2* relaxometry as the reference standard in a pediatric and young adult population. MATERIALS AND METHODS: In this retrospective study, we queried the radiology information system to identify all MRIs performed for liver iron quantification from July 2015 to January 2020. All patients had undergone T2* relaxometry on a 1.5-tesla (T) scanner for liver iron concentration (LIC) estimation. In addition, a 3-D multi-echo Dixon was performed using Siemens Healthineers LiverLab (Erlangen, Germany). Two readers independently estimated liver R2* and T2* on the multi-echo Dixon by drawing free-hand regions of interest on the scanner-generated R2* and T2* maps. Conventional T2*-relaxometry-based LIC was the reference standard. We estimated interobserver agreement by concordance correlation coefficient (CCC). We used Bland-Altman analysis and Pearson correlation coefficient (r) to compare LIC by the two methods. RESULTS: Fifty-four MRIs on 38 patients (22 females) were available for analysis. Mean patient age was 11.8 years (standard deviation [SD] 5.3 years). Reference standard LIC ranged 1.1-21.1 (median 6.8) mg/g dry weight of liver. The concordance between readers for T2* estimation using 3-D multi-echo Dixon was substantial (CCC 0.99, confidence interval 0.99-1.00). Bland-Altman plot showed that all observations were clustered around the zero bias line if the LIC average was ≤8 mg/g, and r was very strong (reader 1 r=0.93, reader 2 r=0.92, both P-values <0.001). With increasing LIC, there was a pattern of poor agreement on the Bland-Altman plot, with observations crossing the lower limits of agreement, and r was very weak (reader 1 r=0.05, P-value 0.84; reader 2 r=0.17, P-value 0.44). CONCLUSION: Vendor-based 3-D multi-echo Dixon allows for excellent interobserver correlation in liver T2* estimation. LIC estimated by this method has a very strong correlation with conventional T2* relaxometry if liver iron overload is mild-moderate (LIC ≤8 mg/g).

Radiologists' Increasing Role in Population Health Management: AJR Expert Panel Narrative Review.

Population health management (PHM) is the holistic process of improving health outcomes of groups of individuals through the support of appropriate financial and care models. Radiologists' presence at the intersection of many aspects of health care, including screening, diagnostic imaging, and image-guided therapies, provides the opportunity for increased radiologist engagement in PHM. Furthermore, innovations in artificial intelligence and imaging informatics will serve as critical tools to improve value in health care through evidence-based and equitable approaches. Given radiologists' limited engagement in PHM to date, it is imperative to define the PHM priorities of the specialty so that radiologists' full value in improving population health is realized. The purpose of this expert review is to explore programs and future directions for radiologists in PHM.

Multitask Interactive Attention Learning Model Based on Hand Images for Assisting Chinese Medicine in Predicting Myocardial Infarction.

Acute myocardial infarction (AMI) is one of the most serious and dangerous cardiovascular diseases. In recent years, the number of patients around the world has been increasing significantly, among which people under the age of 45 have become the high-risk group for sudden death of AMI. AMI occurs quickly and does not show obvious symptoms before onset. In addition, postonset clinical testing is also a complex and invasive test, which may cause some postoperative complications. Therefore, it is necessary to propose a noninvasive and convenient auxiliary diagnostic method. In traditional Chinese medicine (TCM), it is an effective auxiliary diagnostic strategy to complete the disease diagnosis through some body surface features. It is helpful to observe whether the palmar thenar undergoes hypertrophy and whether the metacarpophalangeal joint is swelling in detecting acute myocardial infarction. Combined with deep learning, we propose a depth model based on traditional palm image (MTIALM), which can help doctors of traditional Chinese medicine to predict myocardial infarction. By building the shared network, the model learns information that covers all the tasks. In addition, task-specific attention branch networks are built to simultaneously detect the symptoms of different parts of the palm. The information interaction module (IIM) is proposed to further integrate the information between task branches to ensure that the model learns as many features as possible. Experimental results show that the accuracy of our model in the detection of metacarpophalangeal joints and palmar thenar is 83.16% and 84.15%, respectively, which are significantly improved compared with the traditional classification methods.

Novel Diffusion-Weighted Imaging Score Showed Good Prognostic Value for Acute Basilar Artery Occlusion Following Endovascular Treatment: The Pons-Midbrain and Thalamus Score.

BACKGROUND AND PURPOSE: Prognostic factors for outcome of endovascular treatment remains to be investigated in patients with acute basilar artery occlusion. We aimed to assess the prognostic value of a novel pretreatment diffusion-weighted imaging score: The Pons-Midbrain and Thalamus (PMT) score. METHODS: Eligible patients who underwent endovascular treatment due to acute basilar artery occlusion were reviewed. The PMT score was a diffusion-weighted imaging-based semiquantitative scale in which the infarctions of pons, midbrain, and thalamus were fully considered. The PMT score was assessed as well as the posterior circulation Acute Stroke Prognosis Early Computed Tomography Score and Brain Stem Score. Good outcomes were defined as a modified Rankin Scale score of ≤3 at 90-day and successful reperfusion as Thrombolysis in Cerebral Infarction grades 2b/3. The associations between baseline clinical parameters and good outcomes were evaluated with logistic regression. RESULTS: A total of 107 patients with pretreatment magnetic resonance imaging were included in this cohort. The baseline PMT score (median [interquartile range], 3 [1-5] versus 7 [5-9]; P<0.001) and Brain Stem Score (median [interquartile range], 2 [1-4] versus 3 [2-5]; P=0.001) were significantly lower in good outcome group; the posterior circulation Acute Stroke Prognosis Early Computed Tomography Score was higher in good outcome group without statistical significance. As a result of receiver operating characteristic curve analyses, the posterior circulation Acute Stroke Prognosis Early Computed Tomography Score showed poor prognostic accuracy for good outcome (area under the curve, 0.60 [95% CI, 0.49-0.71]; P=0.081); The baseline PMT score showed significantly better prognostic accuracy for 90-day good outcome than the Brain Stem Score and National Institutes of Health Stroke Scale (area under the curve, 0.80 versus 0.68 versus 0.78, P=0.003). In addition, favorable PMT score <7 (odds ratio, 22.0 [95% CI, 6.0-80.8], P<0.001), Brain Stem Score <3 (odds ratio, 4.65 [95% CI, 2.05-10.55], P<0.001) and baseline National Institutes of Health Stroke Scale <23 (odds ratio, 8.0 [95% CI, 2.5-25.6], P<0.001) were associated with improved good outcome. CONCLUSIONS: In patients with acute basilar artery occlusion following endovascular treatment, the pretreatment diffusion-weighted imaging based PMT score showed good prognostic value for clinical outcome.

Image Integration Using Intracardiac Echography and Three-dimensional Reconstruction for Mapping and Ablation of Atrial and Ventricular Arrhythmias.

This article reviews the basis for image integration of intracardiac echocardiography (ICE) with three-dimensional electroanatomic mapping systems and preprocedural cardiac imaging modalities to enhance anatomic understanding and improve guidance for atrial and ventricular ablation procedures. It discusses the technical aspects of ICE-based integration and the clinical evidence for its use. In addition, it presents the current technical limitations and future directions for this technology. This article also includes figures and videos of clinical representative arrhythmia cases where the use of ICE is key to a safe and successful outcome.

Automatic Construction of Chinese Herbal Prescriptions From Tongue Images Using CNNs and Auxiliary Latent Therapy Topics.

The tongue image provides important physical information of humans. It is of great importance for diagnoses and treatments in clinical medicine. Herbal prescriptions are simple, noninvasive, and have low side effects. Thus, they are widely applied in China. Studies on the automatic construction technology of herbal prescriptions based on tongue images have great significance for deep learning to explore the relevance of tongue images for herbal prescriptions, it can be applied to healthcare services in mobile medical systems. In order to adapt to the tongue image in a variety of photographic environments and construct herbal prescriptions, a neural network framework for prescription construction is designed. It includes single/double convolution channels and fully connected layers. Furthermore, it proposes the auxiliary therapy topic loss mechanism to model the therapy of Chinese doctors and alleviate the interference of sparse output labels on the diversity of results. The experiment use the real-world tongue images and the corresponding prescriptions and the results can generate prescriptions that are close to the real samples, which verifies the feasibility of the proposed method for the automatic construction of herbal prescriptions from tongue images. Also, it provides a reference for automatic herbal prescription construction from more physical information.

The Preventive Effect of Computed Tomography Image-Guided Electroacupuncture Combined with Continuous Femoral Nerve Block on Deep Vein Thrombosis After Total Knee Arthroplasty Based on an Adaptive Algorithm.

Based on an adaptive algorithm model, this study proposed 2 special model structures of randomized fusion and an optimized convolution kernel and use it for image recognition. The adaptive algorithm model combined image-guided electroacupuncture with a continuous femoral nerve block to prevent deep vein thrombosis after total knee arthroplasty. A total of 200 patients after total knee arthroplasty were randomly divided into 4 groups. We assessed the incidence of postoperative lower limb deep vein thrombosis and platelet count before and after surgery. Electroacupuncture combined with continuous femoral nerve block can reduce the incidence of deep vein thrombosis and has obvious advantages in multimode prevention. The effective analgesia provided by electroacupuncture combined with continuous femoral nerve block relieved postoperative pain. It also enabled patients to participate in joint movement and lower limb muscle strength training as soon as possible, which not only is conducive to postoperative functional recovery, but also reduces the body stress response triggered by pain and the hypercoagulable state. Moreover, electroacupuncture stimulation of electroacupuncture points can reduce the inflammatory edema associated with surgery, improve blood circulation at the surgical site, and activate the body's anticoagulation mechanism. This study provides new ideas and references for formulating multimode prevention and control strategies.

TISNet-Enhanced Fully Convolutional Network with Encoder-Decoder Structure for Tongue Image Segmentation in Traditional Chinese Medicine.

Extracting the tongue body accurately from a digital tongue image is a challenge for automated tongue diagnoses, as the blurred edge of the tongue body, interference of pathological details, and the huge difference in the size and shape of the tongue. In this study, an automated tongue image segmentation method using enhanced fully convolutional network with encoder-decoder structure was presented. In the frame of the proposed network, the deep residual network was adopted as an encoder to obtain dense feature maps, and a Receptive Field Block was assembled behind the encoder. Receptive Field Block can capture adequate global contextual prior because of its structure of the multibranch convolution layers with varying kernels. Moreover, the Feature Pyramid Network was used as a decoder to fuse multiscale feature maps for gathering sufficient positional information to recover the clear contour of the tongue body. The quantitative evaluation of the segmentation results of 300 tongue images from the SIPL-tongue dataset showed that the average Hausdorff Distance, average Symmetric Mean Absolute Surface Distance, average Dice Similarity Coefficient, average precision, average sensitivity, and average specificity were 11.2963, 3.4737, 97.26%, 95.66%, 98.97%, and 98.68%, respectively. The proposed method achieved the best performance compared with the other four deep-learning-based segmentation methods (including SegNet, FCN, PSPNet, and DeepLab v3+). There were also similar results on the HIT-tongue dataset. The experimental results demonstrated that the proposed method can achieve accurate tongue image segmentation and meet the practical requirements of automated tongue diagnoses.

Characterizing the Subcortical Structures in Youth with Congenital Heart Disease.

BACKGROUND AND PURPOSE: Congenital heart disease is a leading cause of neurocognitive impairment. Many subcortical structures are known to play a crucial role in higher-order cognitive processing. However, comprehensive anatomic characterization of these structures is currently lacking in the congenital heart disease population. Therefore, this study aimed to compare the morphometry and volume of the globus pallidus, striatum, and thalamus between youth born with congenital heart disease and healthy peers. MATERIALS AND METHODS: We recruited youth between 16 and 24 years of age born with congenital heart disease who underwent cardiopulmonary bypass surgery before 2 years of age (n = 48) and healthy controls of the same age (n = 48). All participants underwent a brain MR imaging to acquire high-resolution 3D T1-weighted images. RESULTS: Smaller surface area and inward bilateral displacement across the lateral surfaces of the globus pallidus were concentrated anteriorly in the congenital heart disease group compared with controls (q < 0.15). On the lateral surfaces of bilateral thalami, we found regions of both larger and smaller surface areas, as well as inward and outward displacement in the congenital heart disease group compared with controls (q < 0.15). We did not find any morphometric differences between groups for the striatum. For the volumetric analyses, only the right globus pallidus showed a significant volume reduction (q < 0.05) in the congenital heart disease group compared with controls. CONCLUSIONS: This study reports morphometric alterations in youth with congenital heart disease in the absence of volume reductions, suggesting that volume alone is not sufficient to detect and explain subtle neuroanatomic differences in this clinical population.

A Case Series of X-Linked Deafness-2 with Sensorineural Hearing Loss, Stapes Fixation, and Perilymphatic Gusher: MR Imaging and Clinical Features of Hypothalamic Malformations.

X-linked deafness-2 (DFNX2) is an X-linked recessive disorder characterized by profound sensorineural hearing loss and a pathognomonic temporal bone deformity. Because hypothalamic malformations associated with DFNX2 have been rarely described, we aimed to further describe these lesions and compare them with features of a nonaffected population. All patients diagnosed with DFNX2 between 2006 and 2019 were included and compared with age-matched patients with normal MR imaging findings and without hypothalamic dysfunction. MR imaging features differing between groups were selected to help identify DFNX2. Sensitivity and specificity were calculated for these features. Agreement among 3 radiologists was quantified using the index κ. Information on the presence or absence of gelastic seizures, precocious puberty, or delayed puberty was also gathered. We selected distinctive MR imaging features of hypothalamic malformations in DFNX2. The feature selected on axial T2 images was the folded appearance of the ventromedial hypothalamus (sensitivity, 100%; specificity, 95.8%) characterized by an abnormal internal/external cleft (sensitivity, 100%; specificity, 95.7%). On coronal T2, the first distinctive feature was a concave morphology of the medial eminence (sensitivity, 100%; specificity, 97.1%), the second feature was at least 1 hypothalamic-septum angle ≥90° (sensitivity, 90%; specificity, 72.5%), and the third feature was a forebrain-hypothalamic craniocaudal length of ≥6 mm (sensitivity, 70%; specificity, 79.7%). Clinical features were also distinctive because 9 patients with DFNX2 did not present with gelastic seizures or precocious puberty. One patient had delayed puberty. The κ index and intraclass correlation coefficient ranged between 0.78 and 0.95. Imaging and clinical features of the hypothalamus suggest that there is a hypothalamic malformation associated with DFNX2. Early assessment for pubertal delay is proposed.

SlicerDMRI: Diffusion MRI and Tractography Research Software for Brain Cancer Surgery Planning and Visualization.

PURPOSE: We present SlicerDMRI, an open-source software suite that enables research using diffusion magnetic resonance imaging (dMRI), the only modality that can map the white matter connections of the living human brain. SlicerDMRI enables analysis and visualization of dMRI data and is aimed at the needs of clinical research users. SlicerDMRI is built upon and deeply integrated with 3D Slicer, a National Institutes of Health-supported open-source platform for medical image informatics, image processing, and three-dimensional visualization. Integration with 3D Slicer provides many features of interest to cancer researchers, such as real-time integration with neuronavigation equipment, intraoperative imaging modalities, and multimodal data fusion. One key application of SlicerDMRI is in neurosurgery research, where brain mapping using dMRI can provide patient-specific maps of critical brain connections as well as insight into the tissue microstructure that surrounds brain tumors. PATIENTS AND METHODS: In this article, we focus on a demonstration of SlicerDMRI as an informatics tool to enable end-to-end dMRI analyses in two retrospective imaging data sets from patients with high-grade glioma. Analyses demonstrated here include conventional diffusion tensor analysis, advanced multifiber tractography, automated identification of critical fiber tracts, and integration of multimodal imagery with dMRI. RESULTS: We illustrate the ability of SlicerDMRI to perform both conventional and advanced dMRI analyses as well as to enable multimodal image analysis and visualization. We provide an overview of the clinical rationale for each analysis along with pointers to the SlicerDMRI tools used in each. CONCLUSION: SlicerDMRI provides open-source and clinician-accessible research software tools for dMRI analysis. SlicerDMRI is available for easy automated installation through the 3D Slicer Extension Manager.

Connectome-wide search for functional connectivity locus associated with pathological rumination as a target for real-time fMRI neurofeedback intervention.

Real-time fMRI neurofeedback (rtfMRI-nf) enables noninvasive targeted intervention in brain activation with high spatial specificity. To achieve this promise of rtfMRI-nf, we introduced and demonstrated a data-driven framework to design a rtfMRI-nf intervention through the discovery of precise target location associated with clinical symptoms and neurofeedback signal optimization. Specifically, we identified the functional connectivity locus associated with rumination symptoms, utilizing a connectome-wide search in resting-state fMRI data from a large cohort of mood and anxiety disorder individuals (N = 223) and healthy controls (N = 45). Then, we performed a rtfMRI simulation analysis to optimize the online functional connectivity neurofeedback signal for the identified functional connectivity. The connectome-wide search was performed in the medial prefrontal cortex and the posterior cingulate cortex/precuneus brain regions to identify the precise location of the functional connectivity associated with rumination severity as measured by the ruminative response style (RRS) scale. The analysis found that the functional connectivity between the loci in the precuneus (-6, -54, 48 mm in MNI) and the right temporo-parietal junction (RTPJ; 49, -49, 23 mm) was positively correlated with RRS scores (depressive, p < 0.001; brooding, p < 0.001; reflective, p = 0.002) in the mood and anxiety disorder group. We then performed a rtfMRI processing simulation to optimize the online computation of the precuneus-RTPJ connectivity. We determined that the two-point method without a control region was appropriate as a functional connectivity neurofeedback signal with less dependence on signal history and its accommodation of head motion. The present study offers a discovery framework for the precise location of functional connectivity targets for rtfMRI-nf intervention, which could help directly translate neuroimaging findings into clinical rtfMRI-nf interventions.

Weakly Supervised Deep Learning for Whole Slide Lung Cancer Image Analysis.

Histopathology image analysis serves as the gold standard for cancer diagnosis. Efficient and precise diagnosis is quite critical for the subsequent therapeutic treatment of patients. So far, computer-aided diagnosis has not been widely applied in pathological field yet as currently well-addressed tasks are only the tip of the iceberg. Whole slide image (WSI) classification is a quite challenging problem. First, the scarcity of annotations heavily impedes the pace of developing effective approaches. Pixelwise delineated annotations on WSIs are time consuming and tedious, which poses difficulties in building a large-scale training dataset. In addition, a variety of heterogeneous patterns of tumor existing in high magnification field are actually the major obstacle. Furthermore, a gigapixel scale WSI cannot be directly analyzed due to the immeasurable computational cost. How to design the weakly supervised learning methods to maximize the use of available WSI-level labels that can be readily obtained in clinical practice is quite appealing. To overcome these challenges, we present a weakly supervised approach in this article for fast and effective classification on the whole slide lung cancer images. Our method first takes advantage of a patch-based fully convolutional network (FCN) to retrieve discriminative blocks and provides representative deep features with high efficiency. Then, different context-aware block selection and feature aggregation strategies are explored to generate globally holistic WSI descriptor which is ultimately fed into a random forest (RF) classifier for the image-level prediction. To the best of our knowledge, this is the first study to exploit the potential of image-level labels along with some coarse annotations for weakly supervised learning. A large-scale lung cancer WSI dataset is constructed in this article for evaluation, which validates the effectiveness and feasibility of the proposed method. Extensive experiments demonstrate the superior performance of our method that surpasses the state-of-the-art approaches by a significant margin with an accuracy of 97.3%. In addition, our method also achieves the best performance on the public lung cancer WSIs dataset from The Cancer Genome Atlas (TCGA). We highlight that a small number of coarse annotations can contribute to further accuracy improvement. We believe that weakly supervised learning methods have great potential to assist pathologists in histology image diagnosis in the near future.

Uncertainty and interpretability in convolutional neural networks for semantic segmentation of colorectal polyps.

Colorectal polyps are known to be potential precursors to colorectal cancer, which is one of the leading causes of cancer-related deaths on a global scale. Early detection and prevention of colorectal cancer is primarily enabled through manual screenings, where the intestines of a patient is visually examined. Such a procedure can be challenging and exhausting for the person performing the screening. This has resulted in numerous studies on designing automatic systems aimed at supporting physicians during the examination. Recently, such automatic systems have seen a significant improvement as a result of an increasing amount of publicly available colorectal imagery and advances in deep learning research for object image recognition. Specifically, decision support systems based on Convolutional Neural Networks (CNNs) have demonstrated state-of-the-art performance on both detection and segmentation of colorectal polyps. However, CNN-based models need to not only be precise in order to be helpful in a medical context. In addition, interpretability and uncertainty in predictions must be well understood. In this paper, we develop and evaluate recent advances in uncertainty estimation and model interpretability in the context of semantic segmentation of polyps from colonoscopy images. Furthermore, we propose a novel method for estimating the uncertainty associated with important features in the input and demonstrate how interpretability and uncertainty can be modeled in DSSs for semantic segmentation of colorectal polyps. Results indicate that deep models are utilizing the shape and edge information of polyps to make their prediction. Moreover, inaccurate predictions show a higher degree of uncertainty compared to precise predictions.

Magnetic Resonance Enema in Rectosigmoid Endometriosis.

Intestinal endometriosis occurs in 4% to 37% of women with deep endometriosis (DE). Noninvasive diagnosis of presence and characteristics of rectosigmoid endometriosis permits the best counseling of patients and ensures best therapeutic planning. Magnetic resonance enema (MR-e) is accurate in diagnosing DE. After colon cleansing, rectal distention and opacification improves the performance of MR-e in diagnosing rectosigmoid endometriosis. MR imaging cannot optimally assess the depth of penetration of endometriosis in the intestinal wall. There is a need for multicentric studies with a larger sample size to evaluate reproducibility of MR-e in diagnosis of rectosigmoid endometriosis for less experienced radiologists.

Evaluation and visualization of facial massage effects by using ultraviolet stereo-image correlation.

BACKGROUND/PURPOSE: This study proposes a technique for visualizing the effect of facial massage using stereo-image correlation with melanin pigment. METHOD: In this method, the melanin pigment of a subject's face is made visible by using an ultraviolet light and utilized as a random pattern for stereo-image correlation. Stereo-pair images of the face with the melanin pigment before and after facial massage are recorded using a desk-sized measurement equipment. Then, the deformation of the face by the massage can be obtained based on the principle of stereovision. The effectiveness of the proposed method is demonstrated by applying it to the massage effect evaluation of eight subjects (females in their 40s). RESULTS: The results show that the massage effect can be visualized from the displacement and strain distributions across the face obtained by the proposed method. In addition, it is observed that the face is displaced significantly by the massage and individual differences between the subjects can be captured. CONCLUSION: The proposed method is effective for evaluating the effect of a facial massage when the painted pattern disappears due to the applied cream during the massage.