Texture Analysis in the Assessment of Rectal Cancer: Comparison of T2WI and Diffusion-Weighted Imaging.

Texture analysis (TA) techniques derived from T2-weighted imaging (T2WI) and apparent diffusion coefficient (ADC) maps of rectal cancer can both achieve good diagnosis performance. This study was to compare TA from T2WI and ADC maps between different pathological T and N stages to confirm which TA analysis is better in diagnosis performance. 146 patients were enrolled in this study. Tumor TA was performed on every patient's T2WI and ADC maps, respectively; then, skewness, kurtosis, uniformity, entropy, energy, inertia, and correlation were calculated. Our results demonstrated that those significant different parameters derived from T2WI had better diagnostic performance than those from ADC maps in differentiating pT3b-4 and pN1-2 stage tumors. In particular, the energy derived from T2WI was an optimal parameter for diagnostic efficiency. High-resolution T2WI plays a key point in the local stage of rectal cancer; thus, TA derived from T2WI may be a more useful tool to aid radiologists and surgeons in selecting treatment.

Musiré: multimodal simulation and reconstruction framework for the radiological imaging sciences.

A software-based workflow is proposed for managing the execution of simulation and image reconstruction for SPECT, PET, CBCT, MRI, BLI and FMI packages in single and multimodal biomedical imaging applications. The workflow is composed of a Bash script, the purpose of which is to provide an interface to the user, and to organize data flow between dedicated programs for simulation and reconstruction. The currently incorporated simulation programs comprise GATE for Monte Carlo simulation of SPECT, PET and CBCT, SpinScenario for simulating MRI, and Lipros for Monte Carlo simulation of BLI and FMI. Currently incorporated image reconstruction programs include CASToR for SPECT and PET as well as RTK for CBCT. MetaImage (mhd) standard is used for voxelized phantom and image data format. Meshlab project (mlp) containers incorporating polygon meshes and point clouds defined by the Stanford triangle format (ply) are employed to represent anatomical structures for optical simulation, and to represent tumour cell inserts. A number of auxiliary programs have been developed for data transformation and adaptive parameter assignment. The software workflow uses fully automatic distribution to, and consolidation from, any number of Linux workstations and CPU cores. Example data are presented for clinical SPECT, PET and MRI systems using the Mida head phantom and for preclinical X-ray, PET and BLI systems employing the Digimouse phantom. The presented method unifies and simplifies multimodal simulation setup and image reconstruction management and might be of value for synergistic image research. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 2'.

Automatic Segmentation of Left Ventricle in Echocardiography Based on YOLOv3 Model to Achieve Constraint and Positioning.

Cardiovascular disease (CVD) is the most common type of disease and has a high fatality rate in humans. Early diagnosis is critical for the prognosis of CVD. Before using myocardial tissue strain, strain rate, and other indicators to evaluate and analyze cardiac function, accurate segmentation of the left ventricle (LV) endocardium is vital for ensuring the accuracy of subsequent diagnosis. For accurate segmentation of the LV endocardium, this paper proposes the extraction of the LV region features based on the YOLOv3 model to locate the positions of the apex and bottom of the LV, as well as that of the LV region; thereafter, the subimages of the LV can be obtained, and based on the Markov random field (MRF) model, preliminary identification and binarization of the myocardium of the LV subimages can be realized. Finally, under the constraints of the three aforementioned positions of the LV, precise segmentation and extraction of the LV endocardium can be achieved using nonlinear least-squares curve fitting and edge approximation. The experiments show that the proposed segmentation evaluation indices of the method, including computation speed (fps), Dice, mean absolute distance (MAD), and Hausdorff distance (HD), can reach 2.1-2.25 fps, 93.57 ± 1.97%, 2.57 ± 0.89 mm, and 6.68 ± 1.78 mm, respectively. This indicates that the suggested method has better segmentation accuracy and robustness than existing techniques.

(An overview of) Synergistic reconstruction for multimodality/multichannel imaging methods.

Imaging is omnipresent in modern society with imaging devices based on a zoo of physical principles, probing a specimen across different wavelengths, energies and time. Recent years have seen a change in the imaging landscape with more and more imaging devices combining that which previously was used separately. Motivated by these hardware developments, an ever increasing set of mathematical ideas is appearing regarding how data from different imaging modalities or channels can be synergistically combined in the image reconstruction process, exploiting structural and/or functional correlations between the multiple images. Here we review these developments, give pointers to important challenges and provide an outlook as to how the field may develop in the forthcoming years. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 1'.

Prospective assessment of breast cancer risk from multimodal multiview ultrasound images via clinically applicable deep learning.

The clinical application of breast ultrasound for the assessment of cancer risk and of deep learning for the classification of breast-ultrasound images has been hindered by inter-grader variability and high false positive rates and by deep-learning models that do not follow Breast Imaging Reporting and Data System (BI-RADS) standards, lack explainability features and have not been tested prospectively. Here, we show that an explainable deep-learning system trained on 10,815 multimodal breast-ultrasound images of 721 biopsy-confirmed lesions from 634 patients across two hospitals and prospectively tested on 912 additional images of 152 lesions from 141 patients predicts BI-RADS scores for breast cancer as accurately as experienced radiologists, with areas under the receiver operating curve of 0.922 (95% confidence interval (CI) = 0.868-0.959) for bimodal images and 0.955 (95% CI = 0.909-0.982) for multimodal images. Multimodal multiview breast-ultrasound images augmented with heatmaps for malignancy risk predicted via deep learning may facilitate the adoption of ultrasound imaging in screening mammography workflows.

Texture Feature-Based Classification on Transrectal Ultrasound Image for Prostatic Cancer Detection.

Prostate cancer is one of the most common cancers in men. Early detection of prostate cancer is the key to successful treatment. Ultrasound imaging is one of the most suitable methods for the early detection of prostate cancer. Although ultrasound images can show cancer lesions, subjective interpretation is not accurate. Therefore, this paper proposes a transrectal ultrasound image analysis method, aiming at characterizing prostate tissue through image processing to evaluate the possibility of malignant tumours. Firstly, the input image is preprocessed by optical density conversion. Then, local binarization and Gaussian Markov random fields are used to extract texture features, and the linear combination is performed. Finally, the fused texture features are provided to SVM classifier for classification. The method has been applied to data set of 342 transrectal ultrasound images obtained from hospitals with an accuracy of 70.93%, sensitivity of 70.00%, and specificity of 71.74%. The experimental results show that it is possible to distinguish cancerous tissues from noncancerous tissues to some extent.

Deep-learning-based accurate hepatic steatosis quantification for histological assessment of liver biopsies.

Hepatic steatosis droplet quantification with histology biopsies has high clinical significance for risk stratification and management of patients with fatty liver diseases and in the decision to use donor livers for transplantation. However, pathology reviewing processes, when conducted manually, are subject to a high inter- and intra-reader variability, due to the overwhelmingly large number and significantly varying appearance of steatosis instances. This process is challenging as there is a large number of overlapped steatosis droplets with either missing or weak boundaries. In this study, we propose a deep-learning-based region-boundary integrated network for precise steatosis quantification with whole slide liver histopathology images. The proposed model consists of two sequential steps: a region extraction and a boundary prediction module for foreground regions and steatosis boundary prediction, followed by an integrated prediction map generation. Missing steatosis boundaries are next recovered from the predicted map and assembled from adjacent image patches to generate results for the whole slide histopathology image. The resulting steatosis measures both at the pixel level and steatosis object-level present strong correlation with pathologist annotations, radiology readouts and clinical data. In addition, the segregated steatosis object count is shown as a promising alternative measure to the traditional metrics at the pixel level. These results suggest a high potential of artificial intelligence-assisted technology to enhance liver disease decision support using whole slide images.

Assessment of Lumbar Lordosis Distribution with a Novel Mathematical Approach and Its Adaptation for Lumbar Intervertebral Disc Degeneration.

INTRODUCTION: Low back pain and disc degeneration could be linked to global spinal geometry. Our study aimed to develop a reliable new mathematical method to assess the local distribution of total lumbar lordosis with a single numeric parameter and compare it with lumbar intervertebral disc degeneration using routine MRI scans. METHODS: An online, open access, easy-to-use platform for measurements was developed based on a novel mathematical approach using MRIs of 60 patients. Our Spinalyze Software can be used online with uploaded MRIs. Several new parameters were introduced and assessed to describe variation in segmental lordosis distribution with a single numerical value. The Pfirrmann grading system was used for the classification of lumbar intervertebral disc degeneration. Relationships were investigated between the grade categories of L1-S1 lumbar discs and the MRI morphological parameters with correlation analysis. RESULTS: Results confirm that the determination of measurement points and calculated parameters are reliable (ICCs and Pearson r values > 0.90), and these parameters were independent of gender. The digression percentage (K%), one of our new parameters, did not show a statistical relationship with the Cobb-angle. According to our results, the maximum deflection breaking-point of lumbar lordosis and its location can be different with the same Cobb-angle and the distribution of global lordosis is uneven because the shape of the lumbar lordosis is shifted downward and centered around the L4 lumbar vertebra. The interobserver reliability of the Pfirrmann grades reading was in the excellent agreement category (88.33% agreement percentage, 0.84 kappa), and digression percentage (K%) showed a significant negative correlation with all L1-S1 disc grades with increasing r correlation values. This means that the smaller the value of digression percentage (K%), the more the number of worn discs in the lower lumbar sections. CONCLUSIONS: Spinalyze Software based on a novel mathematical approach provides a free, easy-to-use, reliable, and online measurement tool using standard MRIs to approximate the curvature of lumbar lordosis. The new reliable K% (digression percentage) is one single quantitative parameter to assess the local distribution of total lumbar lordosis. The results indicate that digression percentage (K%) may possibly be associated with the development of lumbar intervertebral disc degeneration. Further evaluation is needed to assess its behavior and advantage.

Group-representative functional network estimation from multi-subject fMRI data via MRF-based image segmentation.

BACKGROUND AND OBJECTIVE: There has been growing interest in using functional connectivity patterns, determined from fMRI data to characterize groups of individuals exhibiting common traits. However, the present challenge lies in efficient and accurate identification of distinct patterns observed consistently across multiple subjects. Existing approaches either impose strong assumptions, require aligning images before processing, or require data-intensive machine learning algorithms with manually labeled training datasets. In this paper, we propose a more principled and flexible approach to address this. METHODS: Our approach redefines the problem of estimating the group-representative functional network as an image segmentation problem. After employing an improved clustering-based ICA scheme to pre-process the dataset of individual functional network images, we use a maximum a posteriori-Markov random field (MAP-MRF) framework to solve the image segmentation problem. In this framework, we propose a probabilistic model of the individual pixels of the fMRI data, with the model involving a latent group-representative functional network image. Given an observed dataset, we apply a novel and efficient variational Bayes algorithm to recover the associated latent group image. Our methodology seeks to overcome limitations in more traditional schemes by exploiting spatial relationships underlying the connectivity maps and accounting for uncertainty in the estimation process. RESULTS: We validate our approach using synthetic, simulated and real data. First, we generate datasets from the proposed forward model with subject-specific binary masking and measurement noise, as well as from a variant of the model without measurement noise. We use both datasets to evaluate our model, along with two algorithms: coordinate-ascent algorithm and variational Bayes algorithm. We conclude that our proposed model with variational Bayes outperforms other competitors, even under model-misspecification. Using variational Bayes offers a significant improvement in performance, with almost no additional computational overhead. We next test our approach on simulated fMRI data. We show our approach is robust to initialization and can recover a solution close to the ground truth. Finally, we apply our proposed methodology along with baselines to a real dataset of fMRI recordings of individuals from two groups, a control group and a group suffering from depression, with recordings made while individuals were subjected to musical stimuli. Our methodology is able to identify group differences that are less clear under competing methods. CONCLUSIONS: Our model-based approach demonstrates the advantage of probabilistic models and modern algorithms that account for uncertainty in accurate identification of group-representative connectivity maps. The variational Bayes methodology yields highly accurate results without increasing the computational load compared to traditional methods. In addition, it is robust to model misspecification, and increases the ability to avoid local optima in the solution.

An audit of the polytrauma fracture detection rate of clinicians evaluating lodox statscan bodygrams in two South African public sector trauma units.

BACKGROUND: Increasing global demand for specialized radiological investigations has resulted in delayed or non-reporting of plain trauma radiographs by radiologists. This is particularly true in resource-limited environments, where referring clinicians rely largely on their own radiographic interpretation. A wide accuracy range has been documented for non-radiologist reporting of conventional trauma radiographs. The Lodox Statscan whole-body digital X-ray machine is a relatively new technology that poses unique interpretive challenges. The fracture detection rate of trauma clinicians utilizing this modality has not been determined. OBJECTIVE: An audit of the polytrauma fracture detection rate of clinicians evaluating Lodox Statscan bodygrams in two South African public-sector Trauma Units. METHODS: A retrospective descriptive study of imaging data of Cape Town Level 1-equivalent public-sector Trauma Units during March-April 2015. Statscan bodygrams acquired for adult polytrauma triage were reviewed and correlated with follow-up imaging and patient records. Missed fractures were stratified by body part, mechanism of injury and ventilatory support. The fracture detection rate was determined with 95% confidence. The Generalised Fischer Exact Test assessed any association between the fracture site and failure of detection. Specialist orthopaedic review assessed the potential need for surgical management of missed fractures. RESULTS: 227 patients (male = 193, 85%; mean age: 33 years) were included; 195 fractures were demonstrated on the whole-body triage projections. Lower limb fractures predominated (n = 66, 34%). The fracture detection rate was 89% (95% CI = 86-93%), with the site of fracture associated with failure of detection (p = 0.01). Twelve of 21 undetected fractures (57%) involved the elbow or shoulder girdle. All elbow fractures (n = 3, 100%), more than half the shoulder girdle fractures (9/13,69%) and 12% (15/123) of extremity fractures were undetected. One missed fracture (1/21,4.7%) unequivocally required surgical management, while a further 7 (7/21, 33.3%) could potentially have benefitted from surgery, depending on follow-up imaging findings. CONCLUSION: This is the first analysis of the accuracy of bodygram polytrauma fracture detection by clinicians. Particular review of the shoulder girdle, elbow and extremities for subtle fractures, in addition to standardized limb positioning, are recommended for improved diagnostic accuracy in this setting. These findings can inform clinician training courses in this domain.

Reliability and Accuracy of Clinical Assessment and Digital Image Analysis for Steatosis Evaluation in Discarded Human Livers.

BACKGROUND: Accurate assessment of steatosis in procured livers is crucial to reduce the poor outcome associated with high-grade steatosis and to optimize the utilization of donor grafts. Clinical examination and digital image analysis (DIA) have been used for steatosis evaluation, but the validity of these methods is debated. This study aimed to compare these methods with standard histology for assessment of steatosis severity in human livers and to evaluate a revised classification system for automated fat measurement. METHODS: Clinical assessment of liver steatosis at time of retrieval and automated measurement were compared with standard histology in paraffinized and hematoxylin and eosin-stained slides, using a 4-grade scale for ordinal data and percentages for numerical values. RESULTS: Analysis of 42 human livers that were retrieved but not transplanted showed that clinical examination was not reliable for assigning steatosis grades (κw, 0.12; 95% CI, -0.06 to 0.30), overestimated steatosis severity, and had an accuracy of 67% for discriminating low- and high-grade steatosis. Digital image analysis had a substantial agreement on absolute fat percentage (intraclass correlation coefficient, 0.76; 95% CI, 0.63-0.84) and steatosis grades (κw, 0.70; 95% CI, 0.57-0.82), with 88% accuracy using the revised classification system. CONCLUSIONS: Clinical assessment of steatosis is inaccurate, and relying on this method alone could result in unnecessary discard of livers. Digital image analysis is feasible with higher accuracy and reliability, but further clinical studies are required to evaluate its clinical validity.

Non-invasive assessment of liver quality in transplantation based on thermal imaging analysis.

BACKGROUND AND OBJECTIVE: Liver quality evaluation is one of the vital steps for predicting the success of liver transplantation. Current evaluation methods, such as biopsy and visual inspection, which are either invasive or lack of consistent standards, provide limited predictive value of long-term transplant viability. Objective analytical models, based on the real-time infrared images of livers during perfusion and preservation, are proposed as novel methods to precisely evaluate donated liver quality. METHODS: In this study, by using principal component analysis to extract infrared image features as predictors, we construct a multivariate logistic regression model for single liver quality evaluation, and a multi-task learning logistic regression model for cross-liver quality evaluation. RESULTS: The single liver quality predictions show testing errors of 0%. The leave-one-liver-out predictions show testing errors ranging from 9% to 36%. CONCLUSIONS: It is found that there is a strong correlation between the viability of livers and the infrared image features in both single liver and cross-liver quality evaluations. These analytical methods also determine that the selected significant infrared image features indicate regional difference in viability and show that more stringent pre-implantation evaluation may be needed to predict transplant outcomes.

Discrimination surfaces with application to region-specific brain asymmetry analysis.

Discrimination surfaces are here introduced as a diagnostic tool for localizing brain regions where discrimination between diseased and nondiseased participants is higher. To estimate discrimination surfaces, we introduce a Mann-Whitney type of statistic for random fields and present large-sample results characterizing its asymptotic behavior. Simulation results demonstrate that our estimator accurately recovers the true surface and corresponding interval of maximal discrimination. The empirical analysis suggests that in the anterior region of the brain, schizophrenic patients tend to present lower local asymmetry scores in comparison with participants in the control group.

Positron emission and computed X-ray tomography: a coming together.

We describe the introduction of positron emission tomography/computed tomography (PET/CT) to the investigation of patients with cancer. The first such unit in the UK and its mode of operation is discussed and initial applications shown. Five hundred and thirty-five patients have been scanned with 2-[18F]fluoro-2-deoxy-D-glucose from mid-January 2002 to the end of August 2002. From this initial experience a clear view of the impact of this technology is emerging. It can now be stated that (1) PET/CT does speed up the throughput of patient studies by at least 25% and hence adds to the comfort of patients scanned; and (2) PET/CT leads to greater accuracy in the interpretation of data. In view of the routine availability of high quality PET and CT fused maps a significant development in radiotherapy planning is on the horizon. We discuss our experience at present and point to further developments in the near future.