Morphometric analysis of white matter lesions in MR images: method and validation

Artificial intelligence solution to classify pulmonary nodules on CT

PURPOSE

The purpose of this study was to create an algorithm to detect and classify pulmonary nodules in two categories based on their volume greater than 100 mm³ or not, using machine learning and deep learning techniques.

MATERIALS AND METHOD

The dataset used to train the model was provided by the organization team of the SFR (French Radiological Society) Data Challenge 2019. An asynchronous and parallel 3-stages pipeline was developed to process all the data (a data "pre-processing" stage; a "nodule detection" stage; a "classifier" stage). Lung segmentation was achieved using 3D U-NET algorithm; nodule detection was done using 3D Retina-UNET and classifier stage with a support vector machine algorithm on selected features. Performances were assessed using area under receiver operating characteristics curve (AUROC).

RESULTS

The pipeline showed good performance for pathological nodule detection and patient diagnosis. With the preparation dataset, an AUROC of 0.9058 (95% confidence interval [CI]: 0.8746-0.9362) was obtained, 87% yielding accuracy (95% CI: 84.83%-91.03%) for the "nodule detection" stage, corresponding to 86% specificity (95% CI: 82%-92%) and 89% sensitivity (95% CI: 84.83%-91.03%).

CONCLUSION

A fully functional pipeline using 3D U-NET, 3D Retina-UNET and classifier stage with a support vector machine algorithm was developed, resulting in high capabilities for pulmonary nodule classification.

Developing and testing an algorithm for automatic segmentation of the fetal face from three-dimensional ultrasound images

Fetal craniofacial abnormalities are challenging to detect and diagnose on prenatal ultrasound (US). Image segmentation and computer analysis of three-dimensional US volumes of the fetal face may provide an objective measure to quantify fetal facial features and identify abnormalities. We have developed and tested an atlas-based partially automated facial segmentation algorithm; however, the volumes require additional manual segmentation (MS), which is time and labour intensive and may preclude this method from clinical adoption. These manually refined segmentations can then be used as a reference (atlas) by the partially automated segmentation algorithm to improve algorithmic performance with the aim of eliminating the need for manual refinement and developing a fully automated system. This study assesses the inter- and intra-operator variability of MS and tests an optimized version of our automatic segmentation (AS) algorithm. The manual refinements of 15 fetal faces performed by three operators and repeated by one operator were assessed by Dice score, average symmetrical surface distance and volume difference. The performance of the partially automatic algorithm with difference size atlases was evaluated by Dice score and computational time. Assessment of the manual refinements showed low inter- and intra-operator variability demonstrating its suitability for optimizing the AS algorithm. The algorithm showed improved performance following an increase in the atlas size in turn reducing the need for manual refinement.

Optimization for Medical Image Segmentation: Theory and Practice When Evaluating With Dice Score or Jaccard Index

In many medical imaging and classical computer vision tasks, the Dice score and Jaccard index are used to evaluate the segmentation performance. Despite the existence and great empirical success of metric-sensitive losses, i.e. relaxations of these metrics such as soft Dice, soft Jaccard and Lovász-Softmax, many researchers still use per-pixel losses, such as (weighted) cross-entropy to train CNNs for segmentation. Therefore, the target metric is in many cases not directly optimized. We investigate from a theoretical perspective, the relation within the group of metric-sensitive loss functions and question the existence of an optimal weighting scheme for weighted cross-entropy to optimize the Dice score and Jaccard index at test time. We find that the Dice score and Jaccard index approximate each other relatively and absolutely, but we find no such approximation for a weighted Hamming similarity. For the Tversky loss, the approximation gets monotonically worse when deviating from the trivial weight setting where soft Tversky equals soft Dice. We verify these results empirically in an extensive validation on six medical segmentation tasks and can confirm that metric-sensitive losses are superior to cross-entropy based loss functions in case of evaluation with Dice Score or Jaccard Index. This further holds in a multi-class setting, and across different object sizes and foreground/background ratios. These results encourage a wider adoption of metric-sensitive loss functions for medical segmentation tasks where the performance measure of interest is the Dice score or Jaccard index.

Degeneration Affects Three-Dimensional Strains in Human Menisci: In situ MRI Acquisition Combined With Image Registration

Degenerative changes of menisci contribute to the evolution of osteoarthritis in the knee joint, because they alter the load transmission to the adjacent articular cartilage. Identifying alterations in the strain response of meniscal tissue under compression that are associated with progressive degeneration may uncover links between biomechanical function and meniscal degeneration. Therefore, the goal of this study was to investigate how degeneration effects the three-dimensional (3D; axial, circumferential, radial) strain in different anatomical regions of human menisci (anterior and posterior root attachment; anterior and posterior horn; pars intermedia) under simulated compression. Magnetic resonance imaging (MRI) was performed to acquire image sequences of 12 mild and 12 severe degenerated knee joints under unloaded and loaded [25%, 50% and 100% body weight (BW)] conditions using a customized loading device. Medial and lateral menisci as well as their root attachments were manually segmented. Intensity-based rigid and non-rigid image registration were performed to obtain 3D deformation fields under the respective load levels. Finally, the 3D voxels were transformed into hexahedral finite-element models and direction-dependent local strain distributions were determined. The axial compressive strain in menisci and meniscal root attachments significantly increased on average from 3.1% in mild degenerated joints to 7.3% in severe degenerated knees at 100% BW (p ≤ 0.021). In severe degenerated knee joints, the menisci displayed a mean circumferential strain of 0.45% (mild: 0.35%) and a mean radial strain of 0.41% (mild: 0.37%) at a load level of 100% BW. No significant changes were observed in the circumferential or radial directions between mild and severe degenerated knee joints for all load levels (p > 0.05). In conclusion, high-resolution MRI was successfully combined with image registration to investigate spatial strain distributions of the meniscus and its attachments in response to compression. The results of the current study highlight that the compressive integrity of the meniscus decreases with progressing tissue degeneration, whereas the tensile properties are maintained.

Theoretical analysis and experimental validation of volume bias of soft Dice optimized segmentation maps in the context of inherent uncertainty

The clinical interest is often to measure the volume of a structure, which is typically derived from a segmentation. In order to evaluate and compare segmentation methods, the similarity between a segmentation and a predefined ground truth is measured using popular discrete metrics, such as the Dice score. Recent segmentation methods use a differentiable surrogate metric, such as soft Dice, as part of the loss function during the learning phase. In this work, we first briefly describe how to derive volume estimates from a segmentation that is, potentially, inherently uncertain or ambiguous. This is followed by a theoretical analysis and an experimental validation linking the inherent uncertainty to common loss functions for training CNNs, namely cross-entropy and soft Dice. We find that, even though soft Dice optimization leads to an improved performance with respect to the Dice score and other measures, it may introduce a volume bias for tasks with high inherent uncertainty. These findings indicate some of the method's clinical limitations and suggest doing a closer ad-hoc volume analysis with an optional re-calibration step.

Functional connectivity of the irritative zone identified by electrical source imaging, and EEG-correlated fMRI analyses

•

The irritative zone differs on Electrical Source Imaging (ESI) and EEG-fMRI.

•

Findings differ in functional connectivity and show low temporal correlation.

•

ESI and EEG-fMRI reveal distinct aspects of the irritative zone.

•

The differences may reflect differences in temporal resolution of the techniques.

Objective

The irritative zone - the area generating epileptic spikes - can be studied non-invasively during the interictal period using Electrical Source Imaging (ESI) and simultaneous electroencephalography-functional magnetic resonance imaging (EEG-fMRI). Although the techniques yield results which may overlap spatially, differences in spatial localization of the irritative zone within the same patient are consistently observed. To investigate this discrepancy, we used Blood Oxygenation Level Dependent (BOLD) functional connectivity measures to examine the underlying relationship between ESI and EEG-fMRI findings.

Methods

Fifteen patients (age 20–54), who underwent presurgical epilepsy investigation, were scanned using a single-session resting-state EEG-fMRI protocol. Structural MRI was used to obtain the electrode localisation of a high-density 64-channel EEG cap. Electrical generators of interictal epileptiform discharges were obtained using a distributed local autoregressive average (LAURA) algorithm as implemented in Cartool EEG software. BOLD activations were obtained using both spike-related and voltage-map EEG-fMRI analysis. The global maxima of each method were used to investigate the temporal relationship of BOLD time courses and to assess the spatial similarity using the Dice similarity index between functional connectivity maps.

Results

ESI, voltage-map and spike-related EEG-fMRI methods identified peaks in 15 (100%), 13 (67%) and 8 (53%) of the 15 patients, respectively. For all methods, maxima were localised within the same lobe, but differed in sub-lobar localisation, with a median distance of 22.8 mm between the highest peak for each method. The functional connectivity analysis showed that the temporal correlation between maxima only explained 38% of the variance between the time course of the BOLD response at the maxima. The mean Dice similarity index between seed-voxel functional connectivity maps showed poor spatial agreement.

Significance

Non-invasive methods for the localisation of the irritative zone have distinct spatial and temporal sensitivity to different aspects of the local cortical network involved in the generation of interictal epileptiform discharges.

Habitual sleep is associated with both source memory and hippocampal subfield volume during early childhood

Previous research has established important developmental changes in sleep and memory during early childhood. These changes have been linked separately to brain development, yet few studies have explored their interrelations during this developmental period. The goal of this report was to explore these associations in 200 (100 female) typically developing 4- to 8-year-old children. We examined whether habitual sleep patterns (24-h sleep duration, nap status) were related to children’s performance on a source memory task and hippocampal subfield volumes. Results revealed that, across all participants, after controlling for age, habitual sleep duration was positively related to source memory performance. In addition, in younger (4–6 years, n = 67), but not older (6–8 years, n = 70) children, habitual sleep duration was related to hippocampal head subfield volume (CA2-4/DG). Moreover, within younger children, volume of hippocampal subfields varied as a function of nap status; children who were still napping (n = 28) had larger CA1 volumes in the body compared to children who had transitioned out of napping (n = 39). Together, these findings are consistent with the hypothesis that habitually napping children may have more immature cognitive networks, as indexed by hippocampal integrity. Furthermore, these results shed additional light on why sleep is important during early childhood, a period of substantial brain development.

Validation of a Web-Based Planning Tool for Percutaneous Cryoablation of Renal Tumors

Purpose

To validate a simulation environment for virtual planning of percutaneous cryoablation of renal tumors.

Materials and Methods

Prospectively collected data from 19 MR-guided procedures were used for validation of the simulation model. Volumetric overlap of the simulated ablation zone volume (Σ) and the segmented ablation zone volume (S; assessed on 1-month follow-up scan) was quantified. Validation metrics were DICE Similarity Coefficient (DSC; the ratio between twice the overlapping volume of both ablation zones divided by the sum of both ablation zone volumes), target overlap (the ratio between the overlapping volume of both ablation zones to the volume of S; low ratio means S is underestimated), and positive predictive value (the ratio between the overlapping volume of both ablation zones to the volume of Σ; low ratio means S is overestimated). Values were between 0 (no alignment) and 1 (perfect alignment), a value > 0.7 is considered good.

Results

Mean volumes of S and Σ were 14.8 cm³ (± 9.9) and 26.7 cm³ (± 15.0), respectively. Mean DSC value was 0.63 (± 0.2), and ≥ 0.7 in 9 cases (47%). Mean target overlap and positive predictive value were 0.88 (± 0.11) and 0.53 (± 0.24), respectively. In 17 cases (89%), target overlap was ≥ 0.7; positive predictive value was ≥ 0.7 in 4 cases (21%) and < 0.6 in 13 cases (68%). This indicates S is overestimated in the majority of cases.

Conclusion

The validation results showed a tendency of the simulation model to overestimate the ablation effect. Model adjustments are necessary to make it suitable for clinical use.

Association between MRI histogram features and treatment response in locally advanced cervical cancer treated by chemoradiotherapy

OBJECTIVE

To examine the associations of histogram features of T2-weighted (T2W) images and apparent diffusion coefficient (ADC) with treatment response in locally advanced cervical cancer (LACC) following concurrent chemoradiotherapy (CCRT).

MATERIALS AND METHODS

Fifty-eight patients who underwent a 4-week CCRT regimen with MRI prior to treatment (pre-CCRT) and after treatment (post-CCRT) were retrospectively analysed. Histogram features were calculated from volumes of interest (VOIs) from one radiologist on T2W images and ADC maps. VOIs from two radiologists were used to assess observer repeatability in delineation and feature values at both time-points with the Dice similarity coefficient (DSC) and intraclass correlation coefficient (ICC). Treatment response was defined as a 90% reduction in tumour volume. Paired Mann-Whitney U tests were used to determine if features changed significantly between examinations. Two-sample Mann-Whitney U tests were used to identify features that were significantly different between response groups. Receiver operating characteristic (ROC) analysis was done on significantly different MRI features between treatment response groups.

RESULTS

Pre-CCRT delineation and feature repeatability were generally good (DSC > 0.700; ICC > 0.750). Post-CCRT repeatability was low (DSC < 0.700; ICC < 0.750), but ADC mean and percentiles retained good ICC scores. All features, except for T2W_Kurtosis, significantly changed between examinations. Post-CCRT ADC₅₀ was the only feature that demonstrated both good observer variability and significant differences between treatment response groups (p = 0.036) and had an AUC of 0.701 with a cut-off of 1.357 × 10^-6 mm²/s.

CONCLUSION

ADC and T2W histogram features could be used to track changes in LACC tumours undergoing CCRT. Post-CCRT ADC₅₀ was associated with treatment response with good observer repeatability.

KEY POINTS

• Pre-treatment tumour delineation and histogram feature values had good observer repeatability, while these were less repeatable at post-treatment. • MRI histogram analysis could be used to track changes in the tumour as it undergoes concurrent chemoradiotherapy. • Post-treatment median ADC was associated with treatment response and had good repeatability.

Quantitative identification and segmentation repeatability of thoracic spinal muscle morphology

OBJECTIVE

MRI derived spinal-muscle morphology measurements have potential diagnostic, prognostic, and therapeutic applications in spinal health. Muscle morphology in the thoracic spine is an important determinant of kyphosis severity in older adults. However, the literature on quantification of spinal muscles to date has been limited to cervical and lumbar regions. Hence, we aim to propose a method to quantitatively identify regions of interest of thoracic spinal muscle in axial MR images and investigate the repeatability of their measurements.

METHODS

Middle (T4-T5) and lower (T8-T9) thoracic levels of six healthy volunteers (age 26 ± 6 years) were imaged in an upright open scanner (0.5T MROpen, Paramed, Genoa, Italy). A descriptive methodology for defining the regions of interest of trapezius, erector spinae, and transversospinalis in axial MR images was developed. The guidelines for segmentation are laid out based on the points of origin and insertion, probable size, shape, and the position of the muscle groups relative to other recognizable anatomical landmarks as seen from typical axial MR images. 2D parameters such as muscle cross-sectional area (CSA) and muscle position (radius and angle) with respect to the vertebral body centroid were computed and 3D muscle geometries were generated. Intra and inter-rater segmentation repeatability was assessed with intraclass correlation coefficient (ICC (3,1)) for 2D parameters and with dice coefficient (DC) for 3D parameters.

RESULTS

Intra and inter-rater repeatability for 2D and 3D parameters for all muscles was generally good/excellent (average ICC (3,1) = 0.9 with ranges of 0.56-0.98; average DC = 0.92 with ranges from 0.85-0.95).

CONCLUSION

The guidelines proposed are important for reliable MRI-based measurements and allow meaningful comparisons of muscle morphometry in the thoracic spine across different studies globally. Good segmentation repeatability suggests we can further investigate the effect of posture and spinal curvature on muscle morphology in the thoracic spine.

Infant FreeSurfer: An automated segmentation and surface extraction pipeline for T1-weighted neuroimaging data of infants 0-2 years

The development of automated tools for brain morphometric analysis in infants has lagged significantly behind analogous tools for adults. This gap reflects the greater challenges in this domain due to: 1) a smaller-scaled region of interest, 2) increased motion corruption, 3) regional changes in geometry due to heterochronous growth, and 4) regional variations in contrast properties corresponding to ongoing myelination and other maturation processes. Nevertheless, there is a great need for automated image-processing tools to quantify differences between infant groups and other individuals, because aberrant cortical morphologic measurements (including volume, thickness, surface area, and curvature) have been associated with neuropsychiatric, neurologic, and developmental disorders in children. In this paper we present an automated segmentation and surface extraction pipeline designed to accommodate clinical MRI studies of infant brains in a population 0-2 year-olds. The algorithm relies on a single channel of T1-weighted MR images to achieve automated segmentation of cortical and subcortical brain areas, producing volumes of subcortical structures and surface models of the cerebral cortex. We evaluated the algorithm both qualitatively and quantitatively using manually labeled datasets, relevant comparator software solutions cited in the literature, and expert evaluations. The computational tools and atlases described in this paper will be distributed to the research community as part of the FreeSurfer image analysis package.

SelectStitch: Automated Frame Segmentation and Stitching to Create Composite Images from Otoscope Video Clips

Background and Objective: the aim of this study is to develop and validate an automated image segmentation-based frame selection and stitching framework to create enhanced composite images from otoscope videos. The proposed framework, called SelectStitch, is useful for classifying eardrum abnormalities using a single composite image instead of the entire raw otoscope video dataset. Methods: SelectStitch consists of a convolutional neural network (CNN) based semantic segmentation approach to detect the eardrum in each frame of the otoscope video, and a stitching engine to generate a high-quality composite image from the detected eardrum regions. In this study, we utilize two separate datasets: the first one has 36 otoscope videos that were used to train a semantic segmentation model, and the second one, containing 100 videos, which was used to test the proposed method. Cases from both adult and pediatric patients were used in this study. A configuration of 4-levels depth U-Net architecture was trained to automatically find eardrum regions in each otoscope video frame from the first dataset. After the segmentation, we automatically selected meaningful frames from otoscope videos by using a pre-defined threshold, i.e., it should contain at least an eardrum region of 20% of a frame size. We have generated 100 composite images from the test dataset. Three ear, nose, and throat (ENT) specialists (ENT-I, ENT-II, ENT-III) compared in two rounds the composite images produced by SelectStitch against the composite images that were generated by the base processes, i.e., stitching all the frames from the same video data, in terms of their diagnostic capabilities. Results: In the first round of the study, ENT-I, ENT-II, ENT-III graded improvement for 58, 57, and 71 composite images out of 100, respectively, for SelectStitch over the base composite, reflecting greater diagnostic capabilities. In the repeat assessment, these numbers were 56, 56, and 64, respectively. We observed that only 6%, 3%, and 3% of the cases received a lesser score than the base composite images, respectively, for ENT-I, ENT-II, and ENT-III in Round-1, and 4%, 0%, and 2% of the cases in Round-2. Conclusions: We conclude that the frame selection and stitching will increase the probability of detecting a lesion even if it appears in a few frames.

MRI-based radiomics in breast cancer: feature robustness with respect to inter-observer segmentation variability

Radiomics is an emerging field using the extraction of quantitative features from medical images for tissue characterization. While MRI-based radiomics is still at an early stage, it showed some promising results in studies focusing on breast cancer patients in improving diagnoses and therapy response assessment. Nevertheless, the use of radiomics raises a number of issues regarding feature quantification and robustness. Therefore, our study aim was to determine the robustness of radiomics features extracted by two commonly used radiomics software with respect to variability in manual breast tumor segmentation on MRI. A total of 129 histologically confirmed breast tumors were segmented manually in three dimensions on the first post-contrast T1-weighted MR exam by four observers: a dedicated breast radiologist, a resident, a Ph.D. candidate, and a medical student. Robust features were assessed using the intraclass correlation coefficient (ICC > 0.9). The inter-observer variability was evaluated by the volumetric Dice Similarity Coefficient (DSC). The mean DSC for all tumors was 0.81 (range 0.19–0.96), indicating a good spatial overlap of the segmentations based on observers of varying expertise. In total, 41.6% (552/1328) and 32.8% (273/833) of all RadiomiX and Pyradiomics features, respectively, were identified as robust and were independent of inter-observer manual segmentation variability.

Comparison between atlas and convolutional neural network based automatic segmentation of multiple organs at risk in non-small cell lung cancer

Delineation of organs at risk (OARs) is important but time consuming for radiotherapy planning. Automatic segmentation of OARs based on convolutional neural network (CNN) has been established for lung cancer patients at our institution. The aim of this study is to compare automatic segmentation based on CNN (AS-CNN) with automatic segmentation based on atlas (AS-Atlas) in terms of the efficiency and accuracy of OARs contouring.The OARs, including the lungs, esophagus, heart, liver, and spinal cord, of 19 non-small cell lung cancer patients were delineated using three methods: AS-CNN, AS-Atlas in the Pinnacle-software, and manual delineation (MD) by a senior radiation oncologist. MD was used as the ground-truth reference, and the segmentation efficiency was evaluated by the time spent per patient. The accuracy was evaluated using the Mean surface distance (MSD) and Dice similarity coefficient (DSC). The paired t-test or Wilcoxon signed-rank test was used to compare these indexes between the 2 automatic segmentation models.In the 19 testing cases, both AS-CNN and AS-Atlas saved substantial time compared with MD. AS-CNN was more efficient than AS-Atlas (1.6 min vs 2.4 min, P < .001). In terms of the accuracy, AS-CNN performed well in the esophagus, with a DSC of 73.2%. AS-CNN was better than AS-Atlas in segmenting the left lung (DSC: 94.8% vs 93.2%, P = .01; MSD: 1.10 cm vs 1.73 cm, P < .001) and heart (DSC: 89.3% vs 85.8%, P = .05; MSD: 1.65 cm vs 3.66 cm, P < .001). Furthermore, AS-CNN exhibited superior performance in segmenting the liver (DSC: 93.7% vs 93.6%, P = .81; MSD: 2.03 cm VS 2.11 cm, P = .66). The results obtained from AS-CNN and AS-Atlas were similar in segmenting the right lung. However, the performance of AS-CNN in the spinal cord was inferior to that of AS-Atlas (DSC: 82.1% vs 86.8%, P = .01; MSD: 0.87 cm vs 0.66 cm, P = .01).Our study demonstrated that AS-CNN significantly reduced the contouring time and outperformed AS-Atlas in most cases. AS-CNN can potentially be used for OARs segmentation in patients with pathological N2 (pN2) non-small cell lung cancer.

Newborn left amygdala volume associates with attention disengagement from fearful faces at eight months

After 5 months of age, infants begin to prioritize attention to fearful over other facial expressions. One key proposition is that amygdala and related early-maturing subcortical network, is important for emergence of this attentional bias - however, empirical data to support these assertions are lacking. In this prospective longitudinal study, we measured amygdala volumes from MR images in 65 healthy neonates at 2-5 weeks of gestation corrected age and attention disengagement from fearful vs. non-fearful facial expressions at 8 months with eye tracking. Overall, infants were less likely to disengage from fearful than happy/neutral faces, demonstrating an age-typical bias for fear. Left, but not right, amygdala volume (corrected for intracranial volume) was positively associated with the likelihood of disengaging attention from fearful faces to a salient lateral distractor (r = .302, p = .014). No association was observed with the disengagement from neutral or happy faces in equivalent conditions (r = .166 and .125, p = .186 and .320, respectively). These results are the first to link the amygdala volume with the emerging perceptual vigilance for fearful faces during infancy. They suggest a link from the prenatally defined variability in the amygdala size to early postnatal emotional and social traits.

Improved Vim targeting for focused ultrasound ablation treatment of essential tremor: A probabilistic and patient-specific approach

Magnetic resonance-guided focused ultrasound (MRgFUS) ablation of the ventral intermediate (Vim) thalamic nucleus is an incisionless treatment for essential tremor (ET). The standard initial targeting method uses an approximate, atlas-based stereotactic approach. We developed a new patient-specific targeting method to identify an individual's Vim and the optimal MRgFUS target region therein for suppression of tremor. In this retrospective study of 14 ET patients treated with MRgFUS, we investigated the ability of WMnMPRAGE, a highly sensitive and robust sequence for imaging gray matter-white matter contrast, to identify the Vim, FUS ablation, and a clinically efficacious region within the Vim in individual patients. We found that WMnMPRAGE can directly visualize the Vim in ET patients, segmenting this nucleus using manual or automated segmentation capabilities developed by our group. WMnMPRAGE also delineated the ablation's core and penumbra, and showed that all patients' ablation cores lay primarily within their Vim segmentations. We found no significant correlations between standard ablation features (e.g., ablation volume, Vim-ablation overlap) and 1-month post-treatment clinical outcome. We then defined a group-based probabilistic target, which was nonlinearly warped to individual brains; this target was located within the Vim for all patients. The overlaps between this target and patient ablation cores correlated significantly with 1-month clinical outcome (r = -.57, p = .03), in contrast to the standard target (r = -.23, p = .44). We conclude that WMnMPRAGE is a highly sensitive sequence for segmenting Vim and ablation boundaries in individual patients, allowing us to find a novel tremor-associated center within Vim and potentially improving MRgFUS treatment for ET.

Gabor Feature Based LogDemons with Inertial Constraint for Nonrigid Image Registration

Nonrigid image registration plays an important role in the field of computer vision and medical application. The methods based on Demons algorithm for image registration usually use intensity difference as similarity criteria. However, intensity based methods can not preserve image texture details well and are limited by local minima. In order to solve these problems, we propose a Gabor feature based LogDemons registration method in this paper, called GFDemons. We extract Gabor features of the registered images to construct feature similarity metric since Gabor filters are suitable to extract image texture information. Furthermore, because of the weak gradients in some image regions, the update fields are too small to transform the moving image to the fixed image correctly. In order to compensate this deficiency, we propose an inertial constraint strategy based on GFDemons, named IGFDemons, using the previous update fields to provide guided information for the current update field. The inertial constraint strategy can further improve the performance of the proposed method in terms of accuracy and convergence. We conduct experiments on three different types of images and the results demonstrate that the proposed methods achieve better performance than some popular methods.

Evaluation of Ultrafast Wave-CAIPI MPRAGE for Visual Grading and Automated Measurement of Brain Tissue Volume

BACKGROUND AND PURPOSE

Volumetric brain MR imaging typically has long acquisition times. We sought to evaluate an ultrafast MPRAGE sequence based on Wave-CAIPI (Wave-MPRAGE) compared with standard MPRAGE for evaluation of regional brain tissue volumes.

MATERIALS AND METHODS

We performed scan-rescan experiments in 10 healthy volunteers to evaluate the intraindividual variability of the brain volumes measured using the standard and Wave-MPRAGE sequences. We then evaluated 43 consecutive patients undergoing brain MR imaging. Patients underwent 3T brain MR imaging, including a standard MPRAGE sequence (acceleration factor [R] = 2, acquisition time [TA] = 5.2 minutes) and an ultrafast Wave-MPRAGE sequence (R = 9, TA = 1.15 minutes for the 32-channel coil; R = 6, TA = 1.75 minutes for the 20-channel coil). Automated segmentation of regional brain volume was performed. Two radiologists evaluated regional brain atrophy using semiquantitative visual rating scales.

RESULTS

The mean absolute symmetrized percent change in the healthy volunteers participating in the scan-rescan experiments was not statistically different in any brain region for both the standard and Wave-MPRAGE sequences. In the patients undergoing evaluation for neurodegenerative disease, the Dice coefficient of similarity between volumetric measurements obtained from standard and Wave-MPRAGE ranged from 0.86 to 0.95. Similarly, for all regions, the absolute symmetrized percent change for brain volume and cortical thickness showed <6% difference between the 2 sequences. In the semiquantitative visual comparison, the differences between the 2 radiologists' scores were not clinically or statistically significant.

CONCLUSIONS

Brain volumes estimated using ultrafast Wave-MPRAGE show low intraindividual variability and are comparable with those estimated using standard MPRAGE in patients undergoing clinical evaluation for suspected neurodegenerative disease.

Reduced accuracy of MRI deep grey matter segmentation in multiple sclerosis: an evaluation of four automated methods against manual reference segmentations in a multi-center cohort

Background

Deep grey matter (DGM) atrophy in multiple sclerosis (MS) and its relation to cognitive and clinical decline requires accurate measurements. MS pathology may deteriorate the performance of automated segmentation methods. Accuracy of DGM segmentation methods is compared between MS and controls, and the relation of performance with lesions and atrophy is studied.

Methods

On images of 21 MS subjects and 11 controls, three raters manually outlined caudate nucleus, putamen and thalamus; outlines were combined by majority voting. FSL-FIRST, FreeSurfer, Geodesic Information Flow and volBrain were evaluated. Performance was evaluated volumetrically (intra-class correlation coefficient (ICC)) and spatially (Dice similarity coefficient (DSC)). Spearman's correlations of DSC with global and local lesion volume, structure of interest volume (ROIV), and normalized brain volume (NBV) were assessed.

Results

ICC with manual volumes was mostly good and spatial agreement was high. MS exhibited significantly lower DSC than controls for thalamus and putamen. For some combinations of structure and method, DSC correlated negatively with lesion volume or positively with NBV or ROIV. Lesion-filling did not substantially change segmentations.

Conclusions

Automated methods have impaired performance in patients. Performance generally deteriorated with higher lesion volume and lower NBV and ROIV, suggesting that these may contribute to the impaired performance.

Electronic supplementary material

The online version of this article (10.1007/s00415-020-10023-1) contains supplementary material, which is available to authorized users.

An improved matrix-based endovascular guidewire position simulation using fusiform ternary tree

In this paper, a new matrix-based method is proposed to real-time determine, the guidewire position inside an arterial system. The guidewire path is obtained by the optimal path method, particularly, the fusiform ternary tree method according to the principle of minimum output value of root node. An adaptive sampling strategy, and an optimization strategy based on the proximal end and distal end of the guidewire are proposed to change the guidewire position for obtaining an ideal guidewire path. Compared to the existing methods, the proposed method can achieve 74%, 64%, and 70% improvements in accuracy for phantoms 1, 2, and 3, respectively, investigated in this work.

Evaluating White Matter Lesion Segmentations with Refined Sørensen-Dice Analysis

The Sørensen-Dice index (SDI) is a widely used measure for evaluating medical image segmentation algorithms. It offers a standardized measure of segmentation accuracy which has proven useful. However, it offers diminishing insight when the number of objects is unknown, such as in white matter lesion segmentation of multiple sclerosis (MS) patients. We present a refinement for finer grained parsing of SDI results in situations where the number of objects is unknown. We explore these ideas with two case studies showing what can be learned from our two presented studies. Our first study explores an inter-rater comparison, showing that smaller lesions cannot be reliably identified. In our second case study, we demonstrate fusing multiple MS lesion segmentation algorithms based on the insights into the algorithms provided by our analysis to generate a segmentation that exhibits improved performance. This work demonstrates the wealth of information that can be learned from refined analysis of medical image segmentations.

TAPAS: A Thresholding Approach for Probability Map Automatic Segmentation in Multiple Sclerosis

Total brain white matter lesion (WML) volume is the most widely established magnetic resonance imaging (MRI) outcome measure in studies of multiple sclerosis (MS). To estimate WML volume, there are a number of automatic segmentation methods available, yet manual delineation remains the gold standard approach. Automatic approaches often yield a probability map to which a threshold is applied to create lesion segmentation masks. Unfortunately, few approaches systematically determine the threshold employed; many methods use a manually selected threshold, thus introducing human error and bias into the automated procedure. In this study, we propose and validate an automatic thresholding algorithm, Thresholding Approach for Probability Map Automatic Segmentation in Multiple Sclerosis (TAPAS), to obtain subject-specific threshold estimates for probability map automatic segmentation of T2-weighted (T2) hyperintense WMLs. Using multimodal MRI, the proposed method applies an automatic segmentation algorithm to obtain probability maps. We obtain the true subject-specific threshold that maximizes the Sørensen-Dice similarity coefficient (DSC). Then the subject-specific thresholds are modeled on a naive estimate of volume using a generalized additive model. Applying this model, we predict a subject-specific threshold in data not used for training. We ran a Monte Carlo-resampled split-sample cross-validation (100 validation sets) using two data sets: the first obtained from the Johns Hopkins Hospital (JHH) on a Philips 3 Tesla (3T) scanner (n = 94) and a second collected at the Brigham and Women's Hospital (BWH) using a Siemens 3T scanner (n = 40). By means of the proposed automated technique, in the JHH data we found an average reduction in subject-level absolute error of 0.1 mL per one mL increase in manual volume. Using Bland-Altman analysis, we found that volumetric bias associated with group-level thresholding was mitigated when applying TAPAS. The BWH data showed similar absolute error estimates using group-level thresholding or TAPAS likely since Bland-Altman analyses indicated no systematic biases associated with group or TAPAS volume estimates. The current study presents the first validated fully automated method for subject-specific threshold prediction to segment brain lesions.

Amygdalar nuclei and hippocampal subfields on MRI: Test-retest reliability of automated volumetry across different MRI sites and vendors

BACKGROUND

The amygdala and the hippocampus are two limbic structures that play a critical role in cognition and behavior, however their manual segmentation and that of their smaller nuclei/subfields in multicenter datasets is time consuming and difficult due to the low contrast of standard MRI. Here, we assessed the reliability of the automated segmentation of amygdalar nuclei and hippocampal subfields across sites and vendors using FreeSurfer in two independent cohorts of older and younger healthy adults.

METHODS

Sixty-five healthy older (cohort 1) and 68 younger subjects (cohort 2), from the PharmaCog and CoRR consortia, underwent repeated 3D-T1 MRI (interval 1-90 days). Segmentation was performed using FreeSurfer v6.0. Reliability was assessed using volume reproducibility error (ε) and spatial overlapping coefficient (DICE) between test and retest session.

RESULTS

Significant MRI site and vendor effects (p ​< ​.05) were found in a few subfields/nuclei for the ε, while extensive effects were found for the DICE score of most subfields/nuclei. Reliability was strongly influenced by volume, as ε correlated negatively and DICE correlated positively with volume size of structures (absolute value of Spearman's r correlations >0.43, p ​< ​1.39E-36). In particular, volumes larger than 200 ​mm³ (for amygdalar nuclei) and 300 ​mm³ (for hippocampal subfields, except for molecular layer) had the best test-retest reproducibility (ε ​< ​5% and DICE ​> ​0.80).

CONCLUSION

Our results support the use of volumetric measures of larger amygdalar nuclei and hippocampal subfields in multisite MRI studies. These measures could be useful for disease tracking and assessment of efficacy in drug trials.

Tissue-specific and interpretable sub-segmentation of whole tumour burden on CT images by unsupervised fuzzy clustering

BACKGROUND

Cancer typically exhibits genotypic and phenotypic heterogeneity, which can have prognostic significance and influence therapy response. Computed Tomography (CT)-based radiomic approaches calculate quantitative features of tumour heterogeneity at a mesoscopic level, regardless of macroscopic areas of hypo-dense (i.e., cystic/necrotic), hyper-dense (i.e., calcified), or intermediately dense (i.e., soft tissue) portions.

METHOD

With the goal of achieving the automated sub-segmentation of these three tissue types, we present here a two-stage computational framework based on unsupervised Fuzzy C-Means Clustering (FCM) techniques. No existing approach has specifically addressed this task so far. Our tissue-specific image sub-segmentation was tested on ovarian cancer (pelvic/ovarian and omental disease) and renal cell carcinoma CT datasets using both overlap-based and distance-based metrics for evaluation.

RESULTS

On all tested sub-segmentation tasks, our two-stage segmentation approach outperformed conventional segmentation techniques: fixed multi-thresholding, the Otsu method, and automatic cluster number selection heuristics for the K-means clustering algorithm. In addition, experiments showed that the integration of the spatial information into the FCM algorithm generally achieves more accurate segmentation results, whilst the kernelised FCM versions are not beneficial. The best spatial FCM configuration achieved average Dice similarity coefficient values starting from 81.94±4.76 and 83.43±3.81 for hyper-dense and hypo-dense components, respectively, for the investigated sub-segmentation tasks.

CONCLUSIONS

The proposed intelligent framework could be readily integrated into clinical research environments and provides robust tools for future radiomic biomarker validation.

CNN-Based Quality Assurance for Automatic Segmentation of Breast Cancer in Radiotherapy

Purpose: More and more automatic segmentation tools are being introduced in routine clinical practice. However, physicians need to spend a considerable amount of time in examining the generated contours slice by slice. This greatly reduces the benefit of the tool's automaticity. In order to overcome this shortcoming, we developed an automatic quality assurance (QA) method for automatic segmentation using convolutional neural networks (CNNs).

Materials and Methods: The study cohort comprised 680 patients with early-stage breast cancer who received whole breast radiation. The overall architecture of the automatic QA method for deep learning-based segmentation included the following two main parts: a segmentation CNN model and a QA network that was established based on ResNet-101. The inputs were from computed tomography, segmentation probability maps, and uncertainty maps. Two kinds of Dice similarity coefficient (DSC) outputs were tested. One predicted the DSC quality level of each slice ([0.95, 1] for “good,” [0.8, 0.95] for “medium,” and [0, 0.8] for “bad” quality), and the other predicted the DSC value of each slice directly. The performances of the method to predict the quality levels were evaluated with quantitative metrics: balanced accuracy, F score, and the area under the receiving operator characteristic curve (AUC). The mean absolute error (MAE) was used to evaluate the DSC value outputs.

Results: The proposed methods involved two types of output, both of which achieved promising accuracy in terms of predicting the quality level. For the good, medium, and bad quality level prediction, the balanced accuracy was 0.97, 0.94, and 0.89, respectively; the F score was 0.98, 0.91, and 0.81, respectively; and the AUC was 0.96, 0.93, and 0.88, respectively. For the DSC value prediction, the MAE was 0.06 ± 0.19. The prediction time was approximately 2 s per patient.

Conclusions: Our method could predict the segmentation quality automatically. It can provide useful information for physicians regarding further verification and revision of automatic contours. The integration of our method into current automatic segmentation pipelines can improve the efficiency of radiotherapy contouring.

White matter characteristics of motor, sensory and interhemispheric tracts underlying impaired upper limb function in children with unilateral cerebral palsy

This study explored the role of lesion timing (periventricular white matter versus cortical and deep grey matter lesions) and type of corticospinal tract (CST) wiring pattern (contralateral, bilateral, ipsilateral) on white matter characteristics of the CST, medial lemniscus, superior thalamic radiations and sensorimotor transcallosal fibers in children with unilateral cerebral palsy (CP), and examined the association with upper limb function. Thirty-four children (mean age 10 years 7 months ± 2 years 3 months) with unilateral CP underwent a comprehensive upper limb evaluation and diffusion weighted imaging (75 directions, b value 2800). Streamline count, fractional anisotropy and mean diffusivity were extracted from the targeted tracts and asymmetry indices were additionally calculated. Transcranial magnetic stimulation was applied to assess the CST wiring pattern. Results showed a more damaged CST in children with cortical and deep grey matter lesions (N = 10) and ipsilateral CST projections (N = 11) compared to children with periventricular white matter lesions (N = 24; p < 0.02) and contralateral CST projections (N = 9; p < 0.025), respectively. Moderate to high correlations were found between diffusion metrics of the targeted tracts and upper limb function (r = 0.45-0.72; p < 0.01). Asymmetry indices of the CST and sensory tracts could best explain bimanual performance (74%, p < 0.0001) and unimanual capacity (50%, p = 0.004). Adding lesion timing and CST wiring pattern did not further improve the model of bimanual performance, while for unimanual capacity lesion timing was additionally retained (58%, p = 0.0002). These results contribute to a better understanding of the underlying neuropathology of upper limb function in children with unilateral CP and point towards a clinical potential of tractography.

Use of Artificial Intelligence to Reduce Radiation Exposure at Fluoroscopy-Guided Endoscopic Procedures

OBJECTIVES

Exposure to ionizing radiation remains a hazard for patients and healthcare providers. We evaluated the utility of an artificial intelligence (AI)-enabled fluoroscopy system to minimize radiation exposure during image-guided endoscopic procedures.

METHODS

We conducted a prospective study of 100 consecutive patients who underwent fluoroscopy-guided endoscopic procedures. Patients underwent interventions using either conventional or AI-equipped fluoroscopy system that uses ultrafast collimation to limit radiation exposure to the region of interest. The main outcome measure was to compare radiation exposure with patients, which was measured by dose area product. Secondary outcome was radiation scatter to endoscopy personnel measured using dosimeter.

RESULTS

Of 100 patients who underwent procedures using traditional (n = 50) or AI-enabled (n = 50) fluoroscopy systems, there was no significant difference in demographics, body mass index, procedural type, and procedural or fluoroscopy time between the conventional and the AI-enabled fluoroscopy systems. Radiation exposure to patients was lower (median dose area product 2,178 vs 5,708 mGym, P = 0.001) and scatter effect to endoscopy personnel was less (total deep dose equivalent 0.28 vs 0.69 mSv; difference of 59.4%) for AI-enabled fluoroscopy as compared to conventional system. On multivariate linear regression analysis, after adjusting for patient characteristics, procedural/fluoroscopy duration, and type of fluoroscopy system, only AI-equipped fluoroscopy system (coefficient 3,331.9 [95% confidence interval: 1,926.8-4,737.1, P < 0.001) and fluoroscopy duration (coefficient 813.2 [95% confidence interval: 640.5-985.9], P < 0.001) were associated with radiation exposure.

DISCUSSION

The AI-enabled fluoroscopy system significantly reduces radiation exposure to patients and scatter effect to endoscopy personnel (see Graphical abstract, Supplementary Digital Content, http://links.lww.com/AJG/B461).

Validation of automated magnetic resonance image segmentation for radiation therapy planning in prostate cancer

Background and purpose

Magnetic resonance imaging (MRI) is increasingly used in radiation therapy planning of prostate cancer (PC) to reduce target volume delineation uncertainty. This study aimed to assess and validate the performance of a fully automated segmentation tool (AST) in MRI based radiation therapy planning of PC.

Material and methods

Pelvic structures of 65 PC patients delineated in an MRI-only workflow according to established guidelines were included in the analysis. Automatic vs manual segmentation by an experienced oncologist was compared with geometrical parameters, such as the dice similarity coefficient (DSC). Fifteen patients had a second MRI within 15 days to assess repeatability of the AST for prostate and seminal vesicles. Furthermore, we investigated whether hormonal therapy or body mass index (BMI) affected the AST results.

Results

The AST showed high agreement with manual segmentation expressed as DSC (mean, SD) for delineating prostate (0.84, 0.04), bladder (0.92, 0.04) and rectum (0.86, 0.04). For seminal vesicles (0.56, 0.17) and penile bulb (0.69, 0.12) the respective agreement was moderate. Performance of AST was not influenced by neoadjuvant hormonal therapy, although those on treatment had significantly smaller prostates than the hormone-naïve patients (p < 0.0001). In repeat assessment, consistency of prostate delineation resulted in mean DSC of 0.89, (SD 0.03) between the paired MRI scans for AST, while mean DSC of manual delineation was 0.82, (SD 0.05).

Conclusion

Fully automated MRI segmentation tool showed good agreement and repeatability compared with manual segmentation and was found clinically robust in patients with PC. However, manual review and adjustment of some structures in individual cases remain important in clinical use.

Toward Effective Medical Image Analysis Using Hybrid Approaches—Review, Challenges and Applications

Accurate medical images analysis plays a vital role for several clinical applications. Nevertheless, the immense and complex data volume to be processed make difficult the design of effective algorithms. The first aim of this paper is to examine this area of research and to provide some relevant reference sources related to the context of medical image analysis. Then, an effective hybrid solution to further improve the expected results is proposed here. It allows to consider the benefits of the cooperation of different complementary approaches such as statistical-based, variational-based and atlas-based techniques and to reduce their drawbacks. In particular, a pipeline framework that involves different steps such as a preprocessing step, a classification step and a refinement step with variational-based method is developed to identify accurately pathological regions in biomedical images. The preprocessing step has the role to remove noise and improve the quality of the images. Then the classification is based on both symmetry axis detection step and non linear learning with SVM algorithm. Finally, a level set-based model is performed to refine the boundary detection of the region of interest. In this work we will show that an accurate initialization step could enhance final performances. Some obtained results are exposed which are related to the challenging application of brain tumor segmentation.

Detection and localization of distal radius fractures: Deep learning system versus radiologists

PURPOSE

To evaluate a deep learning based image analysis software for the detection and localization of distal radius fractures.

METHOD

A deep learning system (DLS) was trained on 524 wrist radiographs (166 showing fractures). Performance was tested on internal (100 radiographs, 42 showing fractures) and external test sets (200 radiographs, 100 showing fractures). Single and combined views of the radiographs were shown to DLS and three readers. Readers were asked to indicate fracture location with regions of interest (ROI). The DLS yielded scores (range 0-1) and a heatmap. Detection performance was expressed as AUC, sensitivity and specificity at the optimal threshold and compared to radiologists' performance. Heatmaps were compared to radiologists' ROIs.

RESULTS

The DLS showed excellent performance on the internal test set (AUC 0.93 (95% confidence interval (CI) 0.82-0.98) - 0.96 (0.87-1.00), sensitivity 0.81 (0.58-0.95) - 0.90 (0.70-0.99), specificity 0.86 (0.68-0.96) - 1.0 (0.88-1.0)). DLS performance decreased on the external test set (AUC 0.80 (0.71-0.88) - 0.89 (0.81-0.94), sensitivity 0.64 (0.49-0.77) - 0.92 (0.81-0.98), specificity 0.60 (0.45-0.74) - 0.90 (0.78-0.97)). Radiologists' performance was comparable on internal data (sensitivity 0.71 (0.48-0.89) - 0.95 (0.76-1.0), specificity 0.52 (0.32-0.71) - 0.97 (0.82-1.0)) and better on external data (sensitivity 0.88 (0.76-0.96) - 0.98 (0.89-1.0), specificities 0.66 (0.51-0.79) - 1.0 (0.93-1.0), p < 0.05). In over 90%, the areas of peak activation aligned with radiologists' annotations.

CONCLUSIONS

The DLS was able to detect and localize wrist fractures with a performance comparable to radiologists, using only a small dataset for training.

Fully automatic segmentation of glottis and vocal folds in endoscopic laryngeal high-speed videos using a deep Convolutional LSTM Network

The objective investigation of the dynamic properties of vocal fold vibrations demands the recording and further quantitative analysis of laryngeal high-speed video (HSV). Quantification of the vocal fold vibration patterns requires as a first step the segmentation of the glottal area within each video frame from which the vibrating edges of the vocal folds are usually derived. Consequently, the outcome of any further vibration analysis depends on the quality of this initial segmentation process. In this work we propose for the first time a procedure to fully automatically segment not only the time-varying glottal area but also the vocal fold tissue directly from laryngeal high-speed video (HSV) using a deep Convolutional Neural Network (CNN) approach. Eighteen different Convolutional Neural Network (CNN) network configurations were trained and evaluated on totally 13,000 high-speed video (HSV) frames obtained from 56 healthy and 74 pathologic subjects. The segmentation quality of the best performing Convolutional Neural Network (CNN) model, which uses Long Short-Term Memory (LSTM) cells to take also the temporal context into account, was intensely investigated on 15 test video sequences comprising 100 consecutive images each. As performance measures the Dice Coefficient (DC) as well as the precisions of four anatomical landmark positions were used. Over all test data a mean Dice Coefficient (DC) of 0.85 was obtained for the glottis and 0.91 and 0.90 for the right and left vocal fold (VF) respectively. The grand average precision of the identified landmarks amounts 2.2 pixels and is in the same range as comparable manual expert segmentations which can be regarded as Gold Standard. The method proposed here requires no user interaction and overcomes the limitations of current semiautomatic or computational expensive approaches. Thus, it allows also for the analysis of long high-speed video (HSV)-sequences and holds the promise to facilitate the objective analysis of vocal fold vibrations in clinical routine. The here used dataset including the ground truth will be provided freely for all scientific groups to allow a quantitative benchmarking of segmentation approaches in future.

Manual and semiautomatic segmentation of bone sarcomas on MRI have high similarity

The aims of this study were to evaluate the intra- and interobserver reproducibility of manual segmentation of bone sarcomas in magnetic resonance imaging (MRI) studies and to compare manual and semiautomatic segmentation methods. This retrospective study included twelve osteosarcoma and eight Ewing sarcoma MRI studies performed prior to any therapeutic intervention. All cases were histopathologically confirmed. Three radiologists used 3D-Slicer software to perform manual segmentation of bone sarcomas in a blinded and independent manner. One radiologist segmented manually and also performed semiautomatic segmentation with the GrowCut tool. Segmentation exercises were timed for comparison. The dice similarity coefficient (DSC) and Hausdorff distance (HD) were used to evaluate similarity between the segmentation results and further statistical analyses were performed to compare DSC, HD, and volumetric results. Manual segmentation was reproducible with intraobserver DSC varying from 0.83 to 0.97 and HD from 3.37 to 28.73 mm. Interobserver DSC of manual segmentation showed variation from 0.73 to 0.97 and HD from 3.93 to 33.40 mm. Semiautomatic segmentation compared to manual segmentation resulted in DSCs of 0.71−0.96 and HDs of 5.38−31.54 mm. Semiautomatic segmentation required significantly less time compared to manual segmentation (P value ≤0.05). Among all situations compared, tumor volumetry did not show significant statistical differences (P value >0.05). We found excellent intra- and interobserver agreement for manual segmentation of osteosarcoma and Ewing sarcoma. There was high similarity between manual and semiautomatic segmentation, with a significant reduction of segmentation time using the semiautomatic method.

pBrain: A novel pipeline for Parkinson related brain structure segmentation

•

A novel pipeline for Parkinson`s disease structure segmentation is presented.

•

State-of-the-art fast multiatlas patch-based label fusion with systematic error correction is used to accurately and efficiently produce very competitive results in around 5 min.

•

The proposed pipeline works at high resolution (0.5 mm) but it can work also with standard resolution (1 mm) T2 images allowing the analysis of large legacy databases.

•

The proposed pipeline will be made publically available online through our volBrain platform.

Parkinson is a very prevalent neurodegenerative disease impacting the life of millions of people worldwide. Although its cause remains unknown, its functional and structural analysis is fundamental to advance in the search of a cure or symptomatic treatment. The automatic segmentation of deep brain structures related to Parkinson`s disease could be beneficial for the follow up and treatment planning. Unfortunately, there is not broadly available segmentation software to automatically measure Parkinson related structures. In this paper, we present a novel pipeline to segment three deep brain structures related to Parkinson's disease (substantia nigra, subthalamic nucleus and red nucleus). The proposed method is based on the multi-atlas label fusion technology that works on standard and high-resolution T2-weighted images. The proposed method also includes as post-processing a new neural network-based error correction step to minimize systematic segmentation errors. The proposed method has been compared to other state-of-the-art methods showing competitive results in terms of accuracy and execution time.

Knee Meniscus Segmentation and Tear Detection from MRI: A Review

Background:

Automatic diagnostic systems in medical imaging provide useful information to support radiologists and other relevant experts. The systems that help radiologists in their analysis and diagnosis appear to be increasing.

Discussion:

Knee joints are intensively studied structures, as well. In this review, studies that automatically segment meniscal structures from the knee joint MR images and detect tears have been investigated. Some of the studies in the literature merely perform meniscus segmentation, while others include classification procedures that detect both meniscus segmentation and anomalies on menisci. The studies performed on the meniscus were categorized according to the methods they used. The methods used and the results obtained from such studies were analyzed along with their drawbacks, and the aspects to be developed were also emphasized.

Conclusion:

The work that has been done in this area can effectively support the decisions that will be made by radiology and orthopedics specialists. Furthermore, these operations, which were performed manually on MR images, can be performed in a shorter time with the help of computeraided systems, which enables early diagnosis and treatment.

The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation

BACKGROUND

To evaluate binary classifications and their confusion matrices, scientific researchers can employ several statistical rates, accordingly to the goal of the experiment they are investigating. Despite being a crucial issue in machine learning, no widespread consensus has been reached on a unified elective chosen measure yet. Accuracy and F1 score computed on confusion matrices have been (and still are) among the most popular adopted metrics in binary classification tasks. However, these statistical measures can dangerously show overoptimistic inflated results, especially on imbalanced datasets.

RESULTS

The Matthews correlation coefficient (MCC), instead, is a more reliable statistical rate which produces a high score only if the prediction obtained good results in all of the four confusion matrix categories (true positives, false negatives, true negatives, and false positives), proportionally both to the size of positive elements and the size of negative elements in the dataset.

CONCLUSIONS

In this article, we show how MCC produces a more informative and truthful score in evaluating binary classifications than accuracy and F1 score, by first explaining the mathematical properties, and then the asset of MCC in six synthetic use cases and in a real genomics scenario. We believe that the Matthews correlation coefficient should be preferred to accuracy and F1 score in evaluating binary classification tasks by all scientific communities.

A Partially-Learned Algorithm for Joint Photo-acoustic Reconstruction and Segmentation

In an inhomogeneously illuminated photoacoustic image, important information like vascular geometry is not readily available, when only the initial pressure is reconstructed. To obtain the desired information, algorithms for image segmentation are often applied as a post-processing step. In this article, we propose to jointly acquire the photoacoustic reconstruction and segmentation, by modifying a recently developed partially learned algorithm based on a convolutional neural network. We investigate the stability of the algorithm against changes in initial pressures and photoacoustic system settings. These insights are used to develop an algorithm that is robust to input and system settings. Our approach can easily be applied to other imaging modalities and can be modified to perform other high-level tasks different from segmentation. The method is validated on challenging synthetic and experimental photoacoustic tomography data in limited angle and limited view scenarios. It is computationally less expensive than classical iterative methods and enables higher quality reconstructions and segmentations than the state-of-the-art learned and non-learned methods.

Tractography in the presence of multiple sclerosis lesions

Accurate anatomical localisation of specific white matter tracts and the quantification of their tract-specific microstructural damage in conditions such as multiple sclerosis (MS) can contribute to a better understanding of symptomatology, disease evolution and intervention effects. Diffusion MRI-based tractography is being used increasingly to segment white matter tracts as regions-of-interest for subsequent quantitative analysis. Since MS lesions can interrupt the tractography algorithm’s tract reconstruction, clinical studies frequently resort to atlas-based approaches, which are convenient but ignorant to individual variability in tract size and shape. Here, we revisit the problem of individual tractography in MS, comparing tractography algorithms using: (i) The diffusion tensor framework; (ii) constrained spherical deconvolution (CSD); and (iii) damped Richardson-Lucy (dRL) deconvolution. Firstly, using simulated and in vivo data from 29 MS patients and 19 healthy controls, we show that the three tracking algorithms respond differentially to MS pathology. While the tensor-based approach is unable to deal with crossing fibres, CSD produces spurious streamlines, in particular in tissue with high fibre loss and low diffusion anisotropy. With dRL, streamlines are increasingly interrupted in pathological tissue. Secondly, we demonstrate that despite the effects of lesions on the fibre orientation reconstruction algorithms, fibre tracking algorithms are still able to segment tracts that pass through areas with a high prevalence of lesions. Combining dRL-based tractography with an automated tract segmentation tool on data from 131 MS patients, the corticospinal tracts and arcuate fasciculi could be reconstructed in more than 90% of individuals. Comparing tract-specific microstructural parameters (fractional anisotropy, radial diffusivity and magnetisation transfer ratio) in individually segmented tracts to those from a tract probability map, we show that there is no systematic disease-related bias in the individually reconstructed tracts, suggesting that lesions and otherwise damaged parts are not systematically omitted during tractography. Thirdly, we demonstrate modest anatomical correspondence between the individual and tract probability-based approach, with a spatial overlap between 35 and 55%. Correlations between tract-averaged microstructural parameters in individually segmented tracts and the probability-map approach ranged between r = .53 (p < .001) for radial diffusivity in the right cortico-spinal tract and r = .97 (p < .001) for magnetisation transfer ratio in the arcuate fasciculi. Our results show that MS white matter lesions impact fibre orientation reconstructions but this does not appear to hinder the ability to anatomically reconstruct white matter tracts in MS. Individual tract segmentation in MS is feasible on a large scale and could prove a powerful tool for investigating diagnostic and prognostic markers.

A multi-path 2.5 dimensional convolutional neural network system for segmenting stroke lesions in brain MRI images

Automatic identification of brain lesions from magnetic resonance imaging (MRI) scans of stroke survivors would be a useful aid in patient diagnosis and treatment planning. It would also greatly facilitate the study of brain-behavior relationships by eliminating the laborious step of having a human expert manually segment the lesion on each brain scan. We propose a multi-modal multi-path convolutional neural network system for automating stroke lesion segmentation. Our system has nine end-to-end UNets that take as input 2-dimensional (2D) slices and examines all three planes with three different normalizations. Outputs from these nine total paths are concatenated into a 3D volume that is then passed to a 3D convolutional neural network to output a final lesion mask. We trained and tested our method on datasets from three sources: Medical College of Wisconsin (MCW), Kessler Foundation (KF), and the publicly available Anatomical Tracings of Lesions After Stroke (ATLAS) dataset. To promote wide applicability, lesions were included from both subacute (1 to 5 weeks) and chronic ( >  3 months) phases post stroke, and were of both hemorrhagic and ischemic etiology. Cross-study validation results (with independent training and validation datasets) were obtained to compare with previous methods based on naive Bayes, random forests, and three recently published convolutional neural networks. Model performance was quantified in terms of the Dice coefficient, a measure of spatial overlap between the model-identified lesion and the human expert-identified lesion, where 0 is no overlap and 1 is complete overlap. Training on the KF and MCW images and testing on the ATLAS images yielded a mean Dice coefficient of 0.54. This was reliably better than the next best previous model, UNet, at 0.47. Reversing the train and test datasets yields a mean Dice of 0.47 on KF and MCW images, whereas the next best UNet reaches 0.45. With all three datasets combined, the current system compared to previous methods also attained a reliably higher cross-validation accuracy. It also achieved high Dice values for many smaller lesions that existing methods have difficulty identifying. Overall, our system is a clear improvement over previous methods for automating stroke lesion segmentation, bringing us an important step closer to the inter-rater accuracy level of human experts.

Computed Tomography to Cone Beam Computed Tomography Deformable Image Registration for Contour Propagation Using Head and Neck, Patient-Based Computational Phantoms: A Multicenter Study

PURPOSE

To investigate the performance of various algorithms for deformable image registration (DIR) for propagating regions of interest (ROIs) using multiple commercial platforms, from computed tomography to cone beam computed tomography (CBCT) and megavoltage computed tomography.

METHODS AND MATERIALS

Fourteen institutions participated in the study using 5 commercial platforms: RayStation (RaySearch Laboratories, Stockholm, Sweden), MIM (Cleveland, OH), VelocityAI and SmartAdapt (Varian Medical Systems, Palo Alto, CA), and ABAS (Elekta AB, Stockholm, Sweden). Algorithms were tested on synthetic images generated with the ImSimQA (Oncology Systems Limited, Shrewsbury, UK) package by applying 2 specific deformation vector fields (DVF) to real head and neck patient datasets. On-board images from 3 systems were used: megavoltage computed tomography from Tomotherapy and 2 kinds of CBCT from a clinical linear accelerator. Image quality of the system was evaluated. The algorithms' accuracy was assessed by comparing the DIR-mapped ROIs returned by each center with those of the reference, using the Dice similarity coefficient and mean distance to conformity metrics. Statistical inference on the validation results was carried out to identify the prognostic factors of DIR performance.

RESULTS

Analyzing 840 DIR-mapped ROIs returned by the centers, it was demonstrated that DVF intensity and image quality were significant prognostic factors of DIR performance. The accuracy of the propagated contours was generally high, and acceptable DIR performance can be obtained with lower-dose CBCT image protocols.

CONCLUSIONS

The performance of the systems proved to be image quality specific, depending on the DVF type and only partially on the platforms. All systems proved to be robust against image artifacts and noise, except the demon-based software.

Predicting high-intensity focused ultrasound thalamotomy lesions using 2D magnetic resonance thermometry and 3D Gaussian modeling

PURPOSE

To develop a method of using two-dimensional (2D) magnetic resonance thermometry, and three-dimensional (3D) Gaussian modeling to predict the volume, shape, and location of 1 day postoperative T1w high-intensity focused ultrasound lesions in medication refractory tremor patients; thereby facilitating a better comprehension of thermal damage thresholds, which can be utilized to reduce adverse events, and improve patient outcome.

METHODS

Fifteen patients underwent magnetic resonance-guided focused ultrasound (MRgFUS) thalamotomy, which was performed at our center using an InSightec ExAblate 4000 system (Haifa, Israel), and guided by magnetic resonance imaging using a 3 T Discovery 750 (General Electric Healthcare, Waukesha, WI, USA). For treatment monitoring, 2D MR thermometry (temperature sensitivity: -0.00909 ppm/°C, bandwidth: 279 Hz/pixel) was performed in multiple orthogonal planes (sagittal, coronal, and axial) intraoperatively. These images were temporally filtered using a general linear model approach to reduce noise. Temporal volumes of filtered temperature maps with a peak temperature ≥ 47°C were aligned and fitted with a 3D Gaussian to create a canonical heating model. We then fitted the filtered 2D temperature maps with a 3D Gaussian, and used the relationships derived from the 3D heating model to estimate the 3D temperature distribution. These temperature distributions were converted into thermal dose distributions and accumulated across time to create an accumulated thermal dose (ATD) profile. Thresholded ATD profiles were then correlated with manually traced T1-weighted 1 day postoperative lesion volumes across patients, and linear regression slopes were plotted against varying ATD thresholds. Additionally, the Dice-Sørensen coefficient (DSC) was calculated to quantify the volumetric overlap between predicted, and actual lesions.

RESULTS

On average, 18.1 (standard deviation (SD): ±4.6, range: 10-29) sonications were performed with an average peak temperature achieved of 62.4°C (SD: ±2.4, range: 58.2-67.7). An ATD threshold of 35.8 CEM43 was found to give a unity linear regression slope; this corresponded to an average DSC of 0.689 (SD: ±0.090, range: 0.476-0.815).

CONCLUSIONS

Using multiplanar 2D MR thermometry and 3D Gaussian modeling, we were able to achieve very good (DSC = 0.689) predictions of T1w 1 day postoperative lesion volume, shape and location at an ATD threshold of approximately 36 CEM43. Furthermore, this method has the potential to be used in clinical evaluations to further elucidate the relationship between thermal damage and clinical outcome. Accurate 3D lesion prediction will facilitate improved clinical decision making in future MRgFUS thalamotomies.

Breast Cancer Identification via Thermography Image Segmentation with a Gradient Vector Flow and a Convolutional Neural Network

Breast cancer is the most common cancer among women worldwide with about half a million cases reported each year. Mammary thermography can offer early diagnosis at low cost if adequate thermographic images of the breasts are taken. The identification of breast cancer in an automated way can accelerate many tasks and applications of pathology. This can help complement diagnosis. The aim of this work is to develop a system that automatically captures thermographic images of breast and classifies them as normal and abnormal (without cancer and with cancer). This paper focuses on a segmentation method based on a combination of the curvature function k and the gradient vector flow, and for classification, we proposed a convolutional neural network (CNN) using the segmented breast. The aim of this paper is to compare CNN results with other classification techniques. Thus, every breast is characterized by its shape, colour, and texture, as well as left or right breast. These data were used for training as well as to compare the performance of CNN with three classification techniques: tree random forest (TRF), multilayer perceptron (MLP), and Bayes network (BN). CNN presents better results than TRF, MLP, and BN.

Appropriate Methodology for EBRT and HDR Intracavitary/Interstitial Brachytherapy Dose Composite and Clinical Plan Evaluation for Patients With Cervical Cancer

PURPOSE

This study assessed the appropriateness of full parameter addition (FPA) methods with respect to the 3-dimensional deformable dose composite method for evaluating combined external beam radiation therapy (EBRT) and intracavitary brachytherapy (ICBT).

METHODS AND MATERIALS

A total of 22 patients who received EBRT and high-dose-rate ICBT were retrospectively evaluated. Split-ring and tandem applicators were used for all patients. Additional interstitial needles were used for 5 patients to supplement the implant. Deformable image registrations were performed to deform the secondary EBRT and ICBT planning computed tomography (CT) images onto the reference CT from the third fraction of ICBT. The Dice similarity coefficient was used to evaluate the quality of deformable registration. Doses were transferred to the reference CT, scaled to the equivalent dose in 2-Gy fractions and combined to create the dose composite. Eight dose-accumulation methods were evaluated and compared. D_2cc and D_0.1cc for organs at risk were investigated.

RESULTS

The differences in D_2cc for rectum, bladder, sigmoid, and bowel between the FPA method for whole-pelvis EBRT and ICBT, calculated using an old American Brachytherapy Society worksheet (FPA_E_h + I_old) and deformable composite for EBRT with boosts and ICBT (Def_E + B + I) were -2.19 ± 1.37 Gy_{α/β = 3}, -0.64 ± 1.13 Gy_{α/β = 3}, -2.06 ± 2.71 Gy_{α/β = 3}, and -1.59 ± 0.89 Gy_{α/β = 3}, respectively. The differences in D_2cc for rectum, bladder, sigmoid, and bowel between the new ABS worksheet (FPA_E_h + B + I_abs) and the Def_E + B + I method were 1.21 ± 1.22 Gy _{α/β = 3}, 1.93 ± 1.38 Gy_{α/β = 3}, 0.72 ± 1.12 Gy_{α/β = 3}, and 1.19 ± 1.46 Gy_{α/β = 3}, respectively. Differences in dose-volume histogram parameter values among Def_E + B + I and other FPA methods were not statistically significant (P > .05).

CONCLUSIONS

Compared with the FPA-based method, deformable registration-based dose composites demonstrated lower OAR D_2cc and D_0.1cc values; however, the differences were not statistically significant. The current ABS-recommended FPA-based sheet can serve as an acceptable plan evaluation tool for clinical purposes.

Comparison of manual and automated fiber quantification tractography in patients with temporal lobe epilepsy

•

Tractography approaches showed moderate to good agreement for tract morphology.

•

Along- and whole-tract diffusivity was significantly correlated across approaches.

•

Whole-tract AFQ but not manual tract diffusivity correlated with clinical variables.

•

Absence of excellent agreement between approaches warrants caution.

Objective

To investigate the agreement between manually and automatically generated tracts from diffusion tensor imaging (DTI) in patients with temporal lobe epilepsy (TLE). Whole and along-the-tract diffusivity metrics and correlations with patient clinical characteristics were analyzed with respect to tractography approach.

Methods

We recruited 40 healthy controls and 24 patients with TLE who underwent conventional T1-weighted imaging and 60-direction DTI. An automated (Automated Fiber Quantification, AFQ) and manual (TrackVis) deterministic tractography approach was used to identify the uncinate fasciculus (UF) and parahippocampal white matter bundle (PHWM). Tract diffusion scalar metrics were analyzed with respect to agreement across automated and manual approaches (Dice Coefficient and Spearman correlations), to side of onset of epilepsy and patient clinical characteristics, including duration of epilepsy, age of onset and presence of hippocampal sclerosis.

Results

Across approaches the analysis of tract morphology similarity revealed Dice coefficients at moderate to good agreement (0.54 - 0.6) and significant correlations between diffusion values (Spearman's Rho=0.4–0.9). However, within bilateral PHWM, AFQ yielded significantly lower FA (left: Z = 4.4, p<0.001; right: Z = 5.1, p<0.001) and higher MD values (left: Z=-4.7, p<0.001; right: Z=-3.7, p<0.001) compared to the manual approach. Whole tract DTI metrics determined using AFQ were significantly correlated with patient characteristics, including age of epilepsy onset in FA (R = 0.6, p = 0.02) and MD of the ipsilateral PHWM (R=-0.6, p = 0.02), while duration of epilepsy corrected for age correlated with MD in ipsilateral PHWM (R = 0.7, p<0.01). Correlations between clinical metrics and diffusion values extracted using the manual whole tract technique did not survive correction for multiple comparisons. Both manual and automated along-the-tract analyses demonstrated significant correlations with patient clinical characteristics such as age of onset and epilepsy duration. The strongest and most widespread localized ipsi- and contralateral diffusivity alterations were observed in patients with left TLE and patients with HS compared to controls, while patients with right TLE and patients without HS did not show these strong effects.

Conclusions

Manual and AFQ tractography approaches revealed significant correlations in the reconstruction of tract morphology and extracted whole and along-tract diffusivity values. However, as non-identical methods they differed in the respective yield of significant results across clinical correlations and group-wise statistics. Given the absence of excellent agreement between manual and AFQ techniques as demonstrated in the present study, caution should be considered when using AFQ particularly when used without reference to benchmark manual measures.

Non-invasive assessment of glioma microstructure using VERDICT MRI: correlation with histology

PURPOSE

This prospective study evaluated the use of vascular, extracellular and restricted diffusion for cytometry in tumours (VERDICT) MRI to investigate the tissue microstructure in glioma. VERDICT-derived parameters were correlated with both histological features and tumour subtype and were also used to explore the peritumoural region.

METHODS

Fourteen consecutive treatment-naïve patients (43.5 years ± 15.1 years, six males, eight females) with suspected glioma underwent diffusion-weighted imaging including VERDICT modelling. Tumour cell radius and intracellular and combined extracellular/vascular volumes were estimated using a framework based on linearisation and convex optimisation. An experienced neuroradiologist outlined the peritumoural oedema, enhancing tumour and necrosis on T2-weighted imaging and contrast-enhanced T1-weighted imaging. The same regions of interest were applied to the co-registered VERDICT maps to calculate the microstructure parameters. Pathology sections were analysed with semi-automated software to measure cellularity and cell size.

RESULTS

VERDICT parameters were successfully calculated in all patients. The imaging-derived results showed a larger intracellular volume fraction in high-grade glioma compared to low-grade glioma (0.13 ± 0.07 vs. 0.08 ± 0.02, respectively; p = 0.05) and a trend towards a smaller extracellular/vascular volume fraction (0.88 ± 0.07 vs. 0.92 ± 0.04, respectively; p = 0.10). The conventional apparent diffusion coefficient was higher in low-grade gliomas compared to high-grade gliomas, but this difference was not statistically significant (1.22 ± 0.13 × 10-3 mm2/s vs. 0.98 ± 0.38 × 10-3 mm2/s, respectively; p = 0.18).

CONCLUSION

This feasibility study demonstrated that VERDICT MRI can be used to explore the tissue microstructure of glioma using an abbreviated protocol. The VERDICT parameters of tissue structure correlated with those derived on histology. The method shows promise as a potential test for diagnostic stratification and treatment response monitoring in the future.

KEY POINTS

• VERDICT MRI is an advanced diffusion technique which has been correlated with histopathological findings obtained at surgery from patients with glioma in this study. • The intracellular volume fraction measured with VERDICT was larger in high-grade tumours compared to that in low-grade tumours. • The results were complementary to measurements from conventional diffusion-weighted imaging, and the technique could be performed in a clinically feasible timescale.

Volumetric and Spatial Accuracy of Computed Tomography Perfusion Estimated Ischemic Core Volume in Patients With Acute Ischemic Stroke

Background and Purpose- The volume of estimated ischemic core using computed tomography perfusion (CTP) imaging can identify ischemic stroke patients who are likely to benefit from reperfusion, particularly beyond standard time windows. We assessed the accuracy of pretreatment CTP estimated ischemic core in patients with successful endovascular reperfusion. Methods- Patients from the HERMES (Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke Trials) and EXTEND-IA TNK (Tenecteplase Versus Alteplase Before Endovascular Therapy for Ischemic Stroke) databases who had pretreatment CTP, >50% angiographic reperfusion, and follow-up magnetic resonance imaging at 24 hours were included. Ischemic core volume on baseline CTP data was estimated using relative cerebral blood flow <30% (RAPID, iSchemaView). Follow-up diffusion magnetic resonance imaging was registered to CTP, and the diffusion lesion was outlined using a semiautomated algorithm. Volumetric and spatial agreement (using Dice similarity coefficient, average Hausdorff distance, and precision) was assessed, and expert visual assessment of quality was performed. Results- In 120 patients, median CTP estimated ischemic core volume was 7.8 mL (IQR, 1.8-19.9 mL), and median diffusion lesion volume at 24 hours was 30.8 mL (IQR, 14.9-67.6 mL). Median volumetric difference was 4.4 mL (IQR, 1.2-12.0 mL). Dice similarity coefficient was low (median, 0.24; IQR, 0.15-0.37). The median precision (positive predictive value) of 0.68 (IQR, 0.40-0.88) and average Hausdorff distance (median, 3.1; IQR, 1.8-5.7 mm) indicated reasonable spatial agreement for regions estimated as ischemic core at baseline. Overestimation of total ischemic core volume by CTP was uncommon. Expert visual review revealed overestimation predominantly in white matter regions. Conclusions- CTP estimated ischemic core volumes were substantially smaller than follow-up diffusion-weighted imaging lesions at 24 hours despite endovascular reperfusion within 2 hours of imaging. This may be partly because of infarct growth. Volumetric CTP core overestimation was uncommon and not related to imaging-to-reperfusion time. Core overestimation in white matter should be a focus of future efforts to improve CTP accuracy.

A new approach to automatic fetal brain extraction from MRI using a variational level set method

BACKGROUND AND PURPOSE

Appropriate images extracted from the MRI of mothers' wombs can be of great help in the medical diagnosis of fetal abnormalities. As maternal tissue may appear in such images, affecting visualization of myelination of the fetal brain, it is not possible to use methods routinely used for extraction of adult brains for fetal brains. The aim of the present study was to use a variational level set approach to extract fetal brain from T2-weighted MR images of the womb.

METHODS

Coronal T2-weighted images were acquired using fast MRI protocols (to avoid artifacts). The database includes 105 MR images from eight subjects. After correcting the inhomogeneity of the images, the fetal eyes were located, and from that information, the location of the fetus brain was automatically determined. Then, the variational level set was used for fetus brain extraction. The results were analyzed by a clinical specialist (radiologist) and the similarity (Dice and Jaccard coefficients), sensitivity and specificity were calculated.

RESULTS AND CONCLUSIONS

The means of the statistical analysis for the Dice and Jaccard coefficients, sensitivity and specificity, were 99.56%, 96.89%, 95.71%, and 97.96%, respectively. Thus, extraction of fetal brain from MR images was confirmed, both statistically and visually through cross-validation.

Rhesus Macaque Brain Atlas Regions Aligned to an MRI Template

To aid in the analysis of rhesus macaque brain images, we aligned digitized anatomical regions from the widely used atlas of Paxinos et al. to a published magnetic resonance imaging (MRI) template based on a large number of subjects. Digitally labelled atlas images were aligned to the template in 2D and then in 3D. The resulting grey matter regions appear qualitatively to be well registered to the template. To quantitatively validate the procedure, MR brain images of 20 rhesus macaques were aligned to the template along with regions drawn by hand in striatal and cortical areas in each subject's MRI. There was good geometric overlap between the hand drawn regions and the template regions. Positron emission tomography (PET) images of the same subjects showing uptake of a dopamine D₂ receptor ligand were aligned to the template space, and good agreement was found between tracer binding measures calculated using the hand drawn and template regions. In conclusion, an anatomically defined set of rhesus macaque brain regions has been aligned to an MRI template and has been validated for analysis of PET imaging in a subset of striatal and cortical areas. The entire set of over 200 regions is publicly available at https://www.nitrc.org/ . Graphical Abstract ᅟ.

Embedding MRI information into MRSI data source extraction improves brain tumour delineation in animal models

Glioblastoma is the most frequent malignant intra-cranial tumour. Magnetic resonance imaging is the modality of choice in diagnosis, aggressiveness assessment, and follow-up. However, there are examples where it lacks diagnostic accuracy. Magnetic resonance spectroscopy enables the identification of molecules present in the tissue, providing a precise metabolomic signature. Previous research shows that combining imaging and spectroscopy information results in more accurate outcomes and superior diagnostic value. This study proposes a method to combine them, which builds upon a previous methodology whose main objective is to guide the extraction of sources. To this aim, prior knowledge about class-specific information is integrated into the methodology by setting the metric of a latent variable space where Non-negative Matrix Factorisation is performed. The former methodology, which only used spectroscopy and involved combining spectra from different subjects, was adapted to use selected areas of interest that arise from segmenting the T2-weighted image. Results showed that embedding imaging information into the source extraction (the proposed semi-supervised analysis) improved the quality of the tumour delineation, as compared to those obtained without this information (unsupervised analysis). Both approaches were applied to pre-clinical data, involving thirteen brain tumour-bearing mice, and tested against histopathological data. On results of twenty-eight images, the proposed Semi-Supervised Source Extraction (SSSE) method greatly outperformed the unsupervised one, as well as an alternative semi-supervised approach from the literature, with differences being statistically significant. SSSE has proven successful in the delineation of the tumour, while bringing benefits such as 1) not constricting the metabolomic-based prediction to the image-segmented area, 2) ability to deal with signal-to-noise issues, 3) opportunity to answer specific questions by allowing researchers/radiologists define areas of interest that guide the source extraction, 4) creation of an intra-subject model and avoiding contamination from inter-subject overlaps, and 5) extraction of meaningful, good-quality sources that adds interpretability, conferring validation and better understanding of each case.