Deep-learning-based reconstruction of T2-weighted magnetic resonance imaging of the prostate accelerated by compressed sensing provides improved image quality at half the acquisition time

Background

Deep-learning-based reconstruction (DLR) improves the quality of magnetic resonance (MR) images which allows faster acquisitions. The aim of this study was to compare the image quality of standard and accelerated T2 weighted turbo-spin-echo (TSE) images of the prostate reconstructed with and without DLR and to find associations between perceived image quality and calculated image characteristics.

Methods

In a cohort of 47 prospectively enrolled consecutive patients referred for bi-parametric prostate magnetic resonance imaging (MRI), two T2-TSE acquisitions in the transverse plane were acquired on a 3T scanner-a standard T2-TSE sequence and a short sequence accelerated by a factor of two using compressed sensing (CS). The images were reconstructed with and without DLR in super-resolution mode. The image quality was rated in six domains. Signal-to-noise ratio (SNR), and image sharpness were measured.

Results

The mean acquisition time was 281±23 s for the standard and 140±12 s for the short acquisition (P<0.0001). DLR images had higher sharpness compared to non-DLR (P<0.001). Short and short-DLR had lower SNR than the standard and standard-DLR (P<0.001). The perceived image quality of short-DLR was rated better in all categories compared to the standard sequence (P<0.001 to P=0.004). All domains of subjective evaluation were correlated with measured image sharpness (P<0.001).

Conclusions

T2-TSE acquisition of the prostate accelerated using CS combined with DLR reconstruction provides images with increased sharpness that have a superior quality as perceived by human readers compared to standard T2-TSE. The perceived image quality is correlated with measured image contrast.

Pregnancy induces retinal microvascular changes indicating cardio-metabolic stress

We performed longitudinal examinations of the arterial retinal microvasculature using Adaptive Optics Retinal Imaging in a 30-year-old healthy woman with twin pregnancy from the 23rd week of gestation (wog) to three days postpartum. Two blinded graders recorded the average wall-to-lumen ratio (WLR) of the examined retinal artery. There was a significant increase in the mean WLR over the course of pregnancy followed by a decreasing WLR from the 37th wog. The demonstrated changes in WLR may be an expression of vascular remodeling and adaptation to volume load which indicates that pregnancy can be viewed as a cardiovascular stress test.

Assessment of retinal vein pulsation through video-ophthalmoscopy and simultaneous biosignals acquisition

The phenomenon of retinal vein pulsation is still not a deeply understood topic in retinal hemodynamics. In this paper, we present a novel hardware solution for recording retinal video sequences and physiological signals using synchronized acquisition, we apply the photoplethysmographic principle for the semi-automatic processing of retinal video sequences and we analyse the timing of the vein collapse within the cardiac cycle using of an electrocardiographic signal (ECG). We measured the left eyes of healthy subjects and determined the phases of vein collapse within the cardiac cycle using a principle of photoplethysmography and a semi-automatic image processing approach. We found that the time to vein collapse (T_vc) is between 60 ms and 220 ms after the R-wave of the ECG signal, which corresponds to 6% to 28% of the cardiac cycle. We found no correlation between T_vc and the duration of the cardiac cycle and only a weak correlation between T_vc and age (0.37, p = 0.20), and T_vc and systolic blood pressure (-0.33, p = 0.25). The Tvc values are comparable to those of previously published papers and can contribute to the studies that analyze vein pulsations.

Using deep learning for gene detection and classification in raw nanopore signals

Recently, nanopore sequencing has come to the fore as library preparation is rapid and simple, sequencing can be done almost anywhere, and longer reads are obtained than with next-generation sequencing. The main bottleneck still lies in data postprocessing which consists of basecalling, genome assembly, and localizing significant sequences, which is time consuming and computationally demanding, thus prolonging delivery of crucial results for clinical practice. Here, we present a neural network-based method capable of detecting and classifying specific genomic regions already in raw nanopore signals—squiggles. Therefore, the basecalling process can be omitted entirely as the raw signals of significant genes, or intergenic regions can be directly analyzed, or if the nucleotide sequences are required, the identified squiggles can be basecalled, preferably to others. The proposed neural network could be included directly in the sequencing run, allowing real-time squiggle processing.

Self-supervised pretraining for transferable quantitative phase image cell segmentation

In this paper, a novel U-Net-based method for robust adherent cell segmentation for quantitative phase microscopy image is designed and optimised. We designed and evaluated four specific post-processing pipelines. To increase the transferability to different cell types, non-deep learning transfer with adjustable parameters is used in the post-processing step. Additionally, we proposed a self-supervised pretraining technique using nonlabelled data, which is trained to reconstruct multiple image distortions and improved the segmentation performance from 0.67 to 0.70 of object-wise intersection over union. Moreover, we publish a new dataset of manually labelled images suitable for this task together with the unlabelled data for self-supervised pretraining.

VerSe: A Vertebrae labelling and segmentation benchmark for multi-detector CT images

Vertebral labelling and segmentation are two fundamental tasks in an automated spine processing pipeline. Reliable and accurate processing of spine images is expected to benefit clinical decision support systems for diagnosis, surgery planning, and population-based analysis of spine and bone health. However, designing automated algorithms for spine processing is challenging predominantly due to considerable variations in anatomy and acquisition protocols and due to a severe shortage of publicly available data. Addressing these limitations, the Large Scale Vertebrae Segmentation Challenge (VerSe) was organised in conjunction with the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) in 2019 and 2020, with a call for algorithms tackling the labelling and segmentation of vertebrae. Two datasets containing a total of 374 multi-detector CT scans from 355 patients were prepared and 4505 vertebrae have individually been annotated at voxel level by a human-machine hybrid algorithm (https://osf.io/nqjyw/, https://osf.io/t98fz/). A total of 25 algorithms were benchmarked on these datasets. In this work, we present the results of this evaluation and further investigate the performance variation at the vertebra level, scan level, and different fields of view. We also evaluate the generalisability of the approaches to an implicit domain shift in data by evaluating the top-performing algorithms of one challenge iteration on data from the other iteration. The principal takeaway from VerSe: the performance of an algorithm in labelling and segmenting a spine scan hinges on its ability to correctly identify vertebrae in cases of rare anatomical variations. The VerSe content and code can be accessed at: https://github.com/anjany/verse.

Deep-learning-based fully automatic spine centerline detection in CT data

In this contribution, we present a fully automatic approach, that is based on two convolution neural networks (CNN) together with a spine tracing algorithm utilizing a population optimization algorithm. Based on the evaluation of 130 CT scans including heavily distorted and complicated cases, it turned out that this new combination enables fast and robust detection with almost 90% of correctly determined spinal centerlines with computing time of fewer than 20 seconds.

Iterative machine learning based rotational alignment of brain 3D CT data

The optimal rotational alignment of brain Computed Tomography (CT) images to a required standard position has a crucial importance for both automatic and manual diagnostic analysis. In this contribution, we present a novel two-step iterative approach for the automatic 3D rotational alignment of brain CT data. The angles of axial and coronal rotations are determined by an unsupervised by localisation of the Midsagittal Plane (MSP) method. This includes detection and pairing of medially symmetrical feature points. The sagittal rotation angle is subsequently estimated by regression convolutional neural network (CNN). The proposed methodology has been evaluated on a dataset of CT data manually aligned by radiologists. It has been shown that the algorithm achieved the low error of estimated rotations (≈1 degree) and in a significantly shorter time than the experts (≈2 minutes per case).

Learning-based vertebra localization and labeling in 3D CT data of possibly incomplete and pathological spines

BACKGROUND AND OBJECTIVE

We present a fully automatic system based on learning approaches, which aims to localization and identification (labeling) of vertebrae in 3D computed tomography (CT) scans of possibly incomplete spines in patients with bone metastases and vertebral compressions.

METHODS

The framework combines a set of 3D algorithms for i) spine detection using a convolution neural network (CNN) ii) spinal cord tracking based on combination of a CNN and a novel growing sphere method with a population optimization, iii) intervertebral discs localization using a novel approach of spatially variant filtering of intensity profiles and iv) vertebra labeling using a CNN-based classification combined with global dynamic optimization.

RESULTS

The proposed algorithm has been validated in testing databases, including also a publicly available dataset. The mean error of intervertebral discs localization is 4.4 mm, and for vertebra labeling, the average rate of correctly identified vertebrae is 87.1%, which can be considered a good result with respect to the large share of highly distorted spines and incomplete spine scans.

CONCLUSIONS

The proposed framework, which combines several advanced methods including also three CNNs, works fully automatically even with incomplete spine scans and with distorted pathological cases. The achieved results allow including the presented algorithms as the first phase to the fully automated computer-aided diagnosis (CAD) system for automatic spine-bone lesion analysis in oncological patients.

Deep convolutional neural network-based segmentation and classification of difficult to define metastatic spinal lesions in 3D CT data

This paper aims to address the segmentation and classification of lytic and sclerotic metastatic lesions that are difficult to define by using spinal 3D Computed Tomography (CT) images obtained from highly pathologically affected cases. As the lesions are ill-defined and consequently it is difficult to find relevant image features that would enable detection and classification of lesions by classical methods of texture and shape analysis, the problem is solved by automatic feature extraction provided by a deep Convolutional Neural Network (CNN). Our main contributions are: (i) individual CNN architecture, and pre-processing steps that are dependent on a patient data and a scan protocol - it enables work with different types of CT scans; (ii) medial axis transform (MAT) post-processing for shape simplification of segmented lesion candidates with Random Forest (RF) based meta-analysis; and (iii) usability of the proposed method on whole-spine CTs (cervical, thoracic, lumbar), which is not treated in other published methods (they work with thoracolumbar segments of spine only). Our proposed method has been tested on our own dataset annotated by two mutually independent radiologists and has been compared to other published methods. This work is part of the ongoing complex project dealing with spine analysis and spine lesion longitudinal studies.

Accurate micro-computed tomography imaging of pore spaces in collagen-based scaffold

In this work we have used X-ray micro-computed tomography (μCT) as a method to observe the morphology of 3D porous pure collagen and collagen-composite scaffolds useful in tissue engineering. Two aspects of visualizations were taken into consideration: improvement of the scan and investigation of its sensitivity to the scan parameters. Due to the low material density some parts of collagen scaffolds are invisible in a μCT scan. Therefore, here we present different contrast agents, which increase the contrast of the scanned biopolymeric sample for μCT visualization. The increase of contrast of collagenous scaffolds was performed with ceramic hydroxyapatite microparticles (HAp), silver ions (Ag(+)) and silver nanoparticles (Ag-NPs). Since a relatively small change in imaging parameters (e.g. in 3D volume rendering, threshold value and μCT acquisition conditions) leads to a completely different visualized pattern, we have optimized these parameters to obtain the most realistic picture for visual and qualitative evaluation of the biopolymeric scaffold. Moreover, scaffold images were stereoscopically visualized in order to better see the 3D biopolymer composite scaffold morphology. However, the optimized visualization has some discontinuities in zoomed view, which can be problematic for further analysis of interconnected pores by commonly used numerical methods. Therefore, we applied the locally adaptive method to solve discontinuities issue. The combination of contrast agent and imaging techniques presented in this paper help us to better understand the structure and morphology of the biopolymeric scaffold that is crucial in the design of new biomaterials useful in tissue engineering.

Combined bone lesion analysis in 3D CT data of vertebrae

Two novel statistically based methods for bone lesion detection and classification are presented. Together with the previously published MRF method [15], they form a triad of mutually complementary methods that promise, when fused, to enable higher reliability of bone lesion assessment.