CNN-based fully automatic wrist cartilage volume quantification in MR images: A comparative analysis between different CNN architectures.

PURPOSE: Automatic measurement of wrist cartilage volume in MR images. METHODS: We assessed the performance of four manually optimized variants of the U-Net architecture, nnU-Net and Mask R-CNN frameworks for the segmentation of wrist cartilage. The results were compared to those from a patch-based convolutional neural network (CNN) we previously designed. The segmentation quality was assessed on the basis of a comparative analysis with manual segmentation. The best networks were compared using a cross-validation approach on a dataset of 33 3D VIBE images of mostly healthy volunteers. Influence of some image parameters on the segmentation reproducibility was assessed. RESULTS: The U-Net-based networks outperformed the patch-based CNN in terms of segmentation homogeneity and quality, while Mask R-CNN did not show an acceptable performance. The median 3D DSC value computed with the U-Net_AL (0.817) was significantly larger than DSC values computed with the other networks. In addition, the U-Net_AL provided the lowest mean volume error (17%) and the highest Pearson correlation coefficient (0.765) with respect to the ground truth values. Of interest, the reproducibility computed using U-Net_AL was larger than the reproducibility of the manual segmentation. Moreover, the results indicate that the MRI-based wrist cartilage volume is strongly affected by the image resolution. CONCLUSIONS: U-Net CNN with attention layers provided the best wrist cartilage segmentation performance. In order to be used in clinical conditions, the trained network can be fine-tuned on a dataset representing a group of specific patients. The error of cartilage volume measurement should be assessed independently using a non-MRI method.

Deep learning-based fully automatic segmentation of wrist cartilage in MR images.

The study objective was to investigate the performance of a dedicated convolutional neural network (CNN) optimized for wrist cartilage segmentation from 2D MR images. CNN utilized a planar architecture and patch-based (PB) training approach that ensured optimal performance in the presence of a limited amount of training data. The CNN was trained and validated in 20 multi-slice MRI datasets acquired with two different coils in 11 subjects (healthy volunteers and patients). The validation included a comparison with the alternative state-of-the-art CNN methods for the segmentation of joints from MR images and the ground-truth manual segmentation. When trained on the limited training data, the CNN outperformed significantly image-based and PB-U-Net networks. Our PB-CNN also demonstrated a good agreement with manual segmentation (Sørensen-Dice similarity coefficient [DSC] = 0.81) in the representative (central coronal) slices with a large amount of cartilage tissue. Reduced performance of the network for slices with a very limited amount of cartilage tissue suggests the need for fully 3D convolutional networks to provide uniform performance across the joint. The study also assessed inter- and intra-observer variability of the manual wrist cartilage segmentation (DSC = 0.78-0.88 and 0.9, respectively). The proposed deep learning-based segmentation of the wrist cartilage from MRI could facilitate research of novel imaging markers of wrist osteoarthritis to characterize its progression and response to therapy.

Improving detection of fMRI activation at 1.5 T using high permittivity ceramics.

In this work, we propose an application of high permittivity materials (HPMs) to improve functional magnetic resonance imaging (fMRI) at 1.5 T, increasing the receive (Rx) sensitivity of a commercial multi-channel head coil. To evaluate the transmit efficiency, specific absorption rate (SAR), and the signal-to-noise ratio (SNR) changes introduced by the HPMs with relative permittivity of 4500, we considered the following configurations in simulation: a whole-body birdcage coil and an Rx-only multi-channel head coil with and without the HPM blocks in the presence of a homogeneous head phantom or a human body model. Experimental studies were also performed with a phantom and with volunteers. Seven healthy volunteers enrolled in a prospective study of fMRI activation in the motor cortex with and without HPMs. fMRI data were analyzed using group-level paired T-tests between acquisitions with and without HPM blocks. Both electromagnetic simulations and experimental measurements showed ∼25% improvement in the Rx sensitivity of a commercial head coil in the areas of interest when HPM blocks were placed in close proximity. It increased the detected motor cortex fMRI activation volume by an average of 56%, thus resulting in more sensitive functional imaging at 1.5 T.

Elevated Levels of Serum Biomarkers Associated with Damage to the CNS Neurons and Endothelial Cells Are Linked with Changes in Brain Connectivity in Breast Cancer Patients with Vestibulo-Atactic Syndrome.

Vestibulo-atactic syndrome (VAS), which represents a combination of motor and vestibular disorders, can be manifested as a clinical complication of breast cancer treatment and has a significant impact on patients' quality of life. The identification of novel potential biomarkers that might help to predict the onset of VAS and its progression could improve the management of this group of patients. In the current study, the levels of intercellular cell adhesion molecule 1 (ICAM-1), platelet/endothelial cell adhesion molecule 1 (PECAM-1), NSE (neuron-specific enolase), and the antibodies recognizing NR-2 subunit of NMDA receptor (NR-2-ab) were measured in the blood serum of BC survivor patients with vestibulo-atactic syndrome (VAS) and associated with the brain connectome data obtained via functional magnetic resonance imaging (fMRI) studies. A total of 21 patients were registered in this open, single-center trial and compared to age-matched healthy female volunteers (control group) (n = 17). BC patients with VAS demonstrated higher serum levels of ICAM-1, PECAM-1, and NSE and a lower value of NR-2-ab, with values of 654.7 ± 184.8, 115.3 ± 37.03, 49.9 ± 103.9, and 0.5 ± 0.3 pg/mL, respectively, as compared to the healthy volunteers, with 230.2 ± 44.8, 62.8 ± 15.6, 15.5 ± 6.4, and 1.4 ± 0.7 pg/mL. According to the fMRI data (employing seed-to-voxel and ROI-to-ROI methods), in BC patients with VAS, significant changes were detected in the functional connectivity in the areas involved in the regulation of postural-tonic reflexes, the coordination of movements, and the regulation of balance. In conclusion, the detected elevated levels of serum biomarkers may reveal damage to the CNS neurons and endothelial cells that is, in turn, associated with the change in the brain connectivity in this group of patients.

Microstructural Properties of Brain White Matter Tracts in Breast Cancer Survivors: A Diffusion Tensor Imaging Study.

Complex breast cancer (BC) treatment can cause various neurological and psychiatric complications, such as postmastectomy pain syndrome, vestibulocerebellar ataxia, and depression, which can lead to microstructural damage of the white matter tracts of the brain. The purpose of the study is to assess microstructural changes in the white matter tracts of the brain in BC survivors using diffusion tensor imaging (DTI). Single DTI scans were performed on patients (n = 84) after complex BC treatment (i.e., surgery, chemotherapy and/or radiation therapy) and on the control group (n = 40). According to the results, a decrease in the quantitative anisotropy (FDR ≤ 0.05) was revealed in the bilateral corticospinal tracts, cerebellar tracts, corpus callosum, fornix, left superior corticostriatal and left corticopontine parietal in patients after BC treatment in comparison to the control group. A decrease in the quantitative anisotropy (FDR ≤ 0.05) was also revealed in the corpus callosum and right cerebellar tracts in patients after BC treatment with the presence of postmastectomy pain syndrome and vestibulocerebellar ataxia. The use of DTI in patients after BC treatment reveals microstructural properties of the white matter tracts in the brain. The results will allow for the improvement of treatment and rehabilitation approaches in patients receiving treatment for breast cancer.

Functional Network Connectivity Reveals the Brain Functional Alterations in Breast Cancer Survivors.

Different neurological and psychiatric disorders such as vertebrobasilar insufficiency, chronic pain syndrome, anxiety, and depression are observed in more than 90% of patients after treatment for breast cancer and may cause alterations in the functional connectivity of the default mode network. The purpose of the present study is to assess changes in the functional connectivity of the default mode network in patients after breast cancer treatment using resting state functional magnetic resonance imaging (rs-fMRI). Rs-fMRI was performed using a 3.0T MR-scanner on patients (N = 46, women) with neurological disorders (chronic pain, dizziness, headaches, and/or tinnitus) in the late postoperative period (>12 months) after Patey radical mastectomy for breast cancer. According to the intergroup statistical analysis, there were differences in the functional connectivity of the default mode network in all 46 patients after breast cancer treatment compared to the control group (p < 0.01). The use of rs-fMRI in in breast cancer survivors allowed us to identify changes in the functional connectivity in the brain caused by neurological disorders, which correlated with a decreased quality of life in these patients. The results indicate the necessity to improve treatment and rehabilitation methods in this group of patients.