-
1.
Multi-label classification of retinal lesions in diabetic retinopathy for automatic analysis of fundus fluorescein angiography based on deep learning.
Pan, X, Jin, K, Cao, J, Liu, Z, Wu, J, You, K, Lu, Y, Xu, Y, Su, Z, Jiang, J, et al
Graefe's archive for clinical and experimental ophthalmology = Albrecht von Graefes Archiv fur klinische und experimentelle Ophthalmologie. 2020;(4):779-785
Abstract
PURPOSE To automatically detect and classify the lesions of diabetic retinopathy (DR) in fundus fluorescein angiography (FFA) images using deep learning algorithm through comparing 3 convolutional neural networks (CNNs). METHODS A total of 4067 FFA images from Eye Center at the Second Affiliated Hospital of Zhejiang University School of Medicine were annotated with 4 kinds of lesions of DR, including non-perfusion regions (NP), microaneurysms, leakages, and laser scars. Three CNNs including DenseNet, ResNet50, and VGG16 were trained to achieve multi-label classification, which means the algorithms could identify 4 retinal lesions above at the same time. RESULTS The area under the curve (AUC) of DenseNet reached 0.8703, 0.9435, 0.9647, and 0.9653 for detecting NP, microaneurysms, leakages, and laser scars, respectively. For ResNet50, AUC was 0.8140 for NP, 0.9097 for microaneurysms, 0.9585 for leakages, and 0.9115 for laser scars. And for VGG16, AUC was 0.7125 for NP, 0.5569 for microaneurysms, 0.9177 for leakages, and 0.8537 for laser scars. CONCLUSIONS Experimental results demonstrate that DenseNet is a suitable model to automatically detect and distinguish retinal lesions in the FFA images with multi-label classification, which lies the foundation of automatic analysis for FFA images and comprehensive diagnosis and treatment decision-making for DR.
-
2.
Validation of a Deep Learning System for the Full Automation of Bite and Meal Duration Analysis of Experimental Meal Videos.
Konstantinidis, D, Dimitropoulos, K, Langlet, B, Daras, P, Ioakimidis, I
Nutrients. 2020;(1)
Abstract
Eating behavior can have an important effect on, and be correlated with, obesity and eating disorders. Eating behavior is usually estimated through self-reporting measures, despite their limitations in reliability, based on ease of collection and analysis. A better and widely used alternative is the objective analysis of eating during meals based on human annotations of in-meal behavioral events (e.g., bites). However, this methodology is time-consuming and often affected by human error, limiting its scalability and cost-effectiveness for large-scale research. To remedy the latter, a novel "Rapid Automatic Bite Detection" (RABiD) algorithm that extracts and processes skeletal features from videos was trained in a video meal dataset (59 individuals; 85 meals; three different foods) to automatically measure meal duration and bites. In these settings, RABiD achieved near perfect agreement between algorithmic and human annotations (Cohen's kappa κ = 0.894; F1-score: 0.948). Moreover, RABiD was used to analyze an independent eating behavior experiment (18 female participants; 45 meals; three different foods) and results showed excellent correlation between algorithmic and human annotations. The analyses revealed that, despite the changes in food (hash vs. meatballs), the total meal duration remained the same, while the number of bites were significantly reduced. Finally, a descriptive meal-progress analysis revealed that different types of food affect bite frequency, although overall bite patterns remain similar (the outcomes were the same for RABiD and manual). Subjects took bites more frequently at the beginning and the end of meals but were slower in-between. On a methodological level, RABiD offers a valid, fully automatic alternative to human meal-video annotations for the experimental analysis of human eating behavior, at a fraction of the cost and the required time, without any loss of information and data fidelity.
-
3.
Applications of radiomics and machine learning for radiotherapy of malignant brain tumors.
Kocher, M, Ruge, MI, Galldiks, N, Lohmann, P
Strahlentherapie und Onkologie : Organ der Deutschen Rontgengesellschaft ... [et al]. 2020;(10):856-867
-
-
Free full text
-
Abstract
BACKGROUND Magnetic resonance imaging (MRI) and amino acid positron-emission tomography (PET) of the brain contain a vast amount of structural and functional information that can be analyzed by machine learning algorithms and radiomics for the use of radiotherapy in patients with malignant brain tumors. METHODS This study is based on comprehensive literature research on machine learning and radiomics analyses in neuroimaging and their potential application for radiotherapy in patients with malignant glioma or brain metastases. RESULTS Feature-based radiomics and deep learning-based machine learning methods can be used to improve brain tumor diagnostics and automate various steps of radiotherapy planning. In glioma patients, important applications are the determination of WHO grade and molecular markers for integrated diagnosis in patients not eligible for biopsy or resection, automatic image segmentation for target volume planning, prediction of the location of tumor recurrence, and differentiation of pseudoprogression from actual tumor progression. In patients with brain metastases, radiomics is applied for additional detection of smaller brain metastases, accurate segmentation of multiple larger metastases, prediction of local response after radiosurgery, and differentiation of radiation injury from local brain metastasis relapse. Importantly, high diagnostic accuracies of 80-90% can be achieved by most approaches, despite a large variety in terms of applied imaging techniques and computational methods. CONCLUSION Clinical application of automated image analyses based on radiomics and artificial intelligence has a great potential for improving radiotherapy in patients with malignant brain tumors. However, a common problem associated with these techniques is the large variability and the lack of standardization of the methods applied.
-
4.
CEST MR-Fingerprinting: Practical considerations and insights for acquisition schedule design and improved reconstruction.
Perlman, O, Herz, K, Zaiss, M, Cohen, O, Rosen, MS, Farrar, CT
Magnetic resonance in medicine. 2020;(2):462-478
-
-
Free full text
-
Abstract
PURPOSE To understand the influence of various acquisition parameters on the ability of CEST MR-Fingerprinting (MRF) to discriminate different chemical exchange parameters and to provide tools for optimal acquisition schedule design and parameter map reconstruction. METHODS Numerical simulations were conducted using a parallel computing implementation of the Bloch-McConnell equations, examining the effect of TR, TE, flip-angle, water T1 and T2 , saturation-pulse duration, power, and frequency on the discrimination ability of CEST-MRF. A modified Euclidean distance matching metric was evaluated and compared to traditional dot product matching. L-Arginine phantoms of various concentrations and pH were scanned at 4.7T and the results compared to numerical findings. RESULTS Simulations for dot product matching demonstrated that the optimal flip-angle and saturation times are 30∘ and 1100 ms, respectively. The optimal maximal saturation power was 3.4 μT for concentrated solutes with a slow exchange rate, and 5.2 μT for dilute solutes with medium-to-fast exchange rates. Using the Euclidean distance matching metric, much lower maximum saturation powers were required (1.6 and 2.4 μT, respectively), with a slightly longer saturation time (1500 ms) and 90∘ flip-angle. For both matching metrics, the discrimination ability increased with the repetition time. The experimental results were in agreement with simulations, demonstrating that more than a 50% reduction in scan-time can be achieved by Euclidean distance-based matching. CONCLUSIONS Optimization of the CEST-MRF acquisition schedule is critical for obtaining the best exchange parameter accuracy. The use of Euclidean distance-based matching of signal trajectories simultaneously improved the discrimination ability and reduced the scan time and maximal saturation power required.
-
5.
Multiclass retinal disease classification and lesion segmentation in OCT B-scan images using cascaded convolutional networks.
Zhong, P, Wang, J, Guo, Y, Fu, X, Wang, R
Applied optics. 2020;(33):10312-10320
Abstract
Disease classification and lesion segmentation of retinal optical coherence tomography images play important roles in ophthalmic computer-aided diagnosis. However, existing methods achieve the two tasks separately, which is insufficient for clinical application and ignores the internal relation of disease and lesion features. In this paper, a framework of cascaded convolutional networks is proposed to jointly classify retinal diseases and segment lesions. First, we adopt an auxiliary binary classification network to identify normal and abnormal images. Then a novel, to the best of our knowledge, U-shaped multi-task network, BDA-Net, combined with a bidirectional decoder and self-attention mechanism, is used to further analyze abnormal images. Experimental results show that the proposed method reaches an accuracy of 0.9913 in classification and achieves an improvement of around 3% in Dice compared to the baseline U-shaped model in segmentation.
-
6.
Inherent and unpredictable bias in multi-component DESPOT myelin water fraction estimation.
West, DJ, Teixeira, RPAG, Wood, TC, Hajnal, JV, Tournier, JD, Malik, SJ
NeuroImage. 2019;:78-88
Abstract
Multicomponent driven equilibrium steady-state observation of T1 and T2 (mcDESPOT) aims to quantify the Myelin Water Fraction (MWF) using a two-pool microstructural model. The MWF has been used to track neurodevelopment and neurodegeneration and has been histologically correlated to myelin content. mcDESPOT has a clinically feasible acquisition time and high signal-to-noise ratio (SNR) relative to other MWF techniques. However, disagreement exists in the literature between experimental studies that show MWF maps with plausible grey matter-white matter (GM-WM) contrast and theoretical work that questions the accuracy and precision of mcDESPOT. We demonstrate that mcDESPOT parameter estimation is inaccurate and imprecise if intercompartmental exchange is included in the microstructural model, but that significant bias results if exchange is neglected. The source of apparent MWF contrast is likely due to the complex convergence behaviour of the Stochastic Region Contraction (SRC) method commonly used to fit the mcDESPOT model. mcDESPOT-derived parameter estimates are hence not directly relatable to the underlying microstructural model and are only comparable to others using similar acquisition schemes and fitting constraints.
-
7.
Cardiorenal sodium MRI at 7.0 Tesla using a 4/4 channel 1 H/23 Na radiofrequency antenna array.
Boehmert, L, Kuehne, A, Waiczies, H, Wenz, D, Eigentler, TW, Funk, S, von Knobelsdorff-Brenkenhoff, F, Schulz-Menger, J, Nagel, AM, Seeliger, E, et al
Magnetic resonance in medicine. 2019;(6):2343-2356
Abstract
PURPOSE Cardiorenal syndrome describes disorders of the heart and the kidneys in which a dysfunction of 1 organ induces a dysfunction in the other. This work describes the design, evaluation, and application of a 4/4-channel hydrogen-1/sodium (1 H/23 Na) RF array tailored for cardiorenal MRI at 7.0 Tesla (T) for a better physiometabolic understanding of cardiorenal syndrome. METHODS The dual-frequency RF array is composed of a planar posterior section and a modestly curved anterior section, each section consisting of 2 loop elements tailored for 23 Na MR and 2 loopole-type elements customized for 1 H MR. Numerical electromagnetic field and specific absorption rate simulations were carried out. Transmission field ( B1+ ) uniformity was optimized and benchmarked against electromagnetic field simulations. An in vivo feasibility study was performed. RESULTS The proposed array exhibits sufficient RF characteristics, B1+ homogeneity, and penetration depth to perform 23 Na MRI of the heart and kidney at 7.0 T. The mean B1+ field for sodium in the heart is 7.7 ± 0.8 µT/√kW and in the kidney is 6.9 ± 2.3 µT/√kW. The suitability of the RF array for 23 Na MRI was demonstrated in healthy subjects (acquisition time for 23 Na MRI: 18 min; nominal isotropic spatial resolution: 5 mm [kidney] and 6 mm [heart]). CONCLUSION This work provides encouragement for further explorations into densely packed multichannel transceiver arrays tailored for 23 Na MRI of the heart and kidney. Equipped with this technology, the ability to probe sodium concentration in the heart and kidney in vivo using 23 Na MRI stands to make a critical contribution to deciphering the complex interactions between both organs.
-
8.
Amide proton transfer imaging of tumors: theory, clinical applications, pitfalls, and future directions.
Kamimura, K, Nakajo, M, Yoneyama, T, Takumi, K, Kumagae, Y, Fukukura, Y, Yoshiura, T
Japanese journal of radiology. 2019;(2):109-116
Abstract
Amide proton transfer (APT) imaging is an emerging molecular magnetic resonance imaging technique based on chemical exchange saturation transfer (CEST). APT imaging has shown promise in oncologic imaging, especially in the imaging of brain tumors. This review article illustrates the theory of CEST/APT imaging and describes the clinical utility, pitfalls, and potential for future development of APT imaging.
-
9.
Automatic segmentation of the foveal avascular zone in ophthalmological OCT-A images.
Díaz, M, Novo, J, Cutrín, P, Gómez-Ulla, F, Penedo, MG, Ortega, M
PloS one. 2019;(2):e0212364
Abstract
Angiography by Optical Coherence Tomography (OCT-A) is a non-invasive retinal imaging modality of recent appearance that allows the visualization of the vascular structure at predefined depths based on the detection of the blood movement through the retinal vasculature. In this way, OCT-A images constitute a suitable scenario to analyze the retinal vascular properties of regions of interest as is the case of the macular area, measuring the characteristics of the foveal vascular and avascular zones. Extracted parameters of this region can be used as prognostic factors that determine if the patient suffers from certain pathologies (such as diabetic retinopathy or retinal vein occlusion, among others), indicating the associated pathological degree. The manual extraction of these biomedical parameters is a long, tedious and subjective process, introducing a significant intra and inter-expert variability, which penalizes the utility of the measurements. In addition, the absence of tools that automatically facilitate these calculations encourages the creation of computer-aided diagnosis frameworks that ease the doctor's work, increasing their productivity and making viable the use of this type of vascular biomarkers. In this work we propose a fully automatic system that identifies and precisely segments the region of the foveal avascular zone (FAZ) using a novel ophthalmological image modality as is OCT-A. The system combines different image processing techniques to firstly identify the region where the FAZ is contained and, secondly, proceed with the extraction of its precise contour. The system was validated using a representative set of 213 healthy and diabetic OCT-A images, providing accurate results with the best correlation with the manual measurements of two experts clinician of 0.93 as well as a Jaccard's index of 0.82 of the best experimental case in the experiments with healthy OCT-A images. The method also provided satisfactory results in diabetic OCT-A images, with a best correlation coefficient with the manual labeling of an expert clinician of 0.93 and a Jaccard's index of 0.83. This tool provides an accurate FAZ measurement with the desired objectivity and reproducibility, being very useful for the analysis of relevant vascular diseases through the study of the retinal micro-circulation.
-
10.
Pitfalls in liver MRI: Technical approach to avoiding misdiagnosis and improving image quality.
Yacoub, JH, Elsayes, KM, Fowler, KJ, Hecht, EM, Mitchell, DG, Santillan, C, Szklaruk, J
Journal of magnetic resonance imaging : JMRI. 2019;(1):41-58
Abstract
The following is an illustrative review of common pitfalls in liver MRI that may challenge interpretation. This article reviews common technical and diagnostic challenges encountered when interpreting dynamic multiphasic T1 -weighted imaging, hepatobiliary phase imaging, and diffusion-weighted imaging of the liver. Additionally, each section includes suggestions for avoiding diagnostic and technical errors. Level of Evidence: 5 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:41-58.