Region-adaptive magnetic resonance image enhancement for improving CNN-based segmentation of the prostate and prostatic zones.

Automatic segmentation of the prostate of and the prostatic zones on MRI remains one of the most compelling research areas. While different image enhancement techniques are emerging as powerful tools for improving the performance of segmentation algorithms, their application still lacks consensus due to contrasting evidence regarding performance improvement and cross-model stability, further hampered by the inability to explain models' predictions. Particularly, for prostate segmentation, the effectiveness of image enhancement on different Convolutional Neural Networks (CNN) remains largely unexplored. The present work introduces a novel image enhancement method, named RACLAHE, to enhance the performance of CNN models for segmenting the prostate's gland and the prostatic zones. The improvement in performance and consistency across five CNN models (U-Net, U-Net++, U-Net3+, ResU-net and USE-NET) is compared against four popular image enhancement methods. Additionally, a methodology is proposed to explain, both quantitatively and qualitatively, the relation between saliency maps and ground truth probability maps. Overall, RACLAHE was the most consistent image enhancement algorithm in terms of performance improvement across CNN models with the mean increase in Dice Score ranging from 3 to 9% for the different prostatic regions, while achieving minimal inter-model variability. The integration of a feature driven methodology to explain the predictions after applying image enhancement methods, enables the development of a concrete, trustworthy automated pipeline for prostate segmentation on MR images.

Variational Gaussian Mixture Models with robust Dirichlet concentration priors for virtual population generation in hypertrophic cardiomyopathy: a comparison study.

Nowadays, there is a growing need for the development of computationally efficient virtual population generators for large-scale in-silico clinical trials. In this work, we utilize the Gaussian Mixture Models (GMM) with variational Bayesian inference (BGMM) using robust estimations of Dirichlet concentration priors for the generation of virtual populations. The estimations were based on an exponential transformation of the number of Gaussian components. The proposed method was compared against state-of-the-art virtual data generators, such as, the Bayesian networks, the supervised tree ensembles (STE), the unsupervised tree ensembles (UTE), and the artificial neural networks (ANN) towards the generation of 20000 virtual patients with hypertrophic cardiomyopathy (HCM). Our results suggest that the proposed BGMM can yield virtual distributions with small inter- and intra-correlation difference (0.013 and 0.012), in lower execution time (4.321 sec) than STE which achieved the second-best performance.

Generation of virtual patient data for in-silico cardiomyopathies drug development using tree ensembles: a comparative study.

In-silico clinical platforms have been recently used as a new revolutionary path for virtual patients (VP) generation and further analysis, such as, drug development. Advanced individualized models have been developed to enhance flexibility and reliability of the virtual patient cohorts. This study focuses on the implementation and comparison of three different methodologies for generating virtual data for in-silico clinical trials. Towards this direction, three computational methods, namely: (i) the multivariate log-normal distribution (log- MVND), (ii) the supervised tree ensembles, and (iii) the unsupervised tree ensembles are deployed and evaluated against their performance towards the generation of high-quality virtual data using the goodness of fit (gof) and the dataset correlation matrix as performance evaluation measures. Our results reveal the dominance of the tree ensembles towards the generation of virtual data with similar distributions (gof values less than 0.2) and correlation patterns (average difference less than 0.03).

Material properties of human lumbar intervertebral discs across strain rates.

BACKGROUND CONTEXT: The use of finite element (FE) methods to study the biomechanics of the intervertebral disc (IVD) has increased over recent decades due to their ability to quantify internal stresses and strains throughout the tissue. Their accuracy is dependent upon realistic, strain-rate dependent material properties, which are challenging to acquire. PURPOSE: The aim of this study was to use the inverse FE technique to characterize the material properties of human lumbar IVDs across strain rates. STUDY DESIGN: A human cadaveric experimental study coupled with an inverse finite element study. METHODS: To predict the structural response of the IVD accurately, the material response of the constituent structures was required. Therefore, compressive experiments were conducted on 16 lumbar IVDs (39±19 years) to obtain the structural response. An FE model of each of these experiments was developed and then run through an inverse FE algorithm to obtain subject-specific constituent material properties, such that the structural response was accurate. RESULTS: Experimentally, a log-linear relationship between IVD stiffness and strain rate was observed. The material properties obtained through the subject-specific inverse FE optimization of the annulus fibrosus (AF) fiber and AF fiber ground matrix allowed a good match between the experimental and FE response. This resulted in a Young modulus of AF fibers (-MPa) to strain rate (ε˙, /s) relationship of YMAF=31.5ln(ε˙)+435.5, and the C10 parameter of the Neo-Hookean material model of the AF ground matrix was found to be strain-rate independent with an average value of 0.68 MPa. CONCLUSIONS: These material properties can be used to improve the accuracy, and therefore predictive ability of FE models of the spine that are used in a wide range of research areas and clinical applications. CLINICAL SIGNIFICANCE: Finite element models can be used for many applications including investigating low back pain, spinal deformities, injury biomechanics, implant design, design of protective systems, and degenerative disc disease. The accurate material properties obtained in this study will improve the predictive ability, and therefore clinical significance of these models.

The contribution of the glenoid labrum to glenohumeral stability under physiological joint loading using finite element analysis.

The labrum contributes to passive glenohumeral joint stability. Cadaveric studies have demonstrated that this has position and load dependency, which has not been quantified under physiological loads. This study aims to validate subject-specific finite element (FE) models against in vitro measurements of joint stability and to utilise the FE models to predict joint stability under physiological loads. The predicted stability values were within ± one standard deviation of experimental data and the FE models showed a reduction in stability of 10-15% with high, physiological, loads. The developed regression equations provide the first representation of passive glenohumeral stability and will aid surgical decision-making.

Material properties of bovine intervertebral discs across strain rates.

The intervertebral disc (IVD) is a complex structure responsible for distributing compressive loading to adjacent vertebrae and allowing the vertebral column to bend and twist. To study the mechanical behaviour of individual components of the IVD, it is common for specimens to be dissected away from their surrounding tissues for mechanical testing. However, disrupting the continuity of the IVD to obtain material properties of each component separately may result in erroneous values. In this study, an inverse finite element (FE) modelling optimisation algorithm has been used to obtain material properties of the IVD across strain rates, therefore bypassing the need to harvest individual samples of each component. Uniaxial compression was applied to ten fresh-frozen bovine intervertebral discs at strain rates of 10-3-1/s. The experimental data were fed into the inverse FE optimisation algorithm and each experiment was simulated using the subject specific FE model of the respective specimen. A sensitivity analysis revealed that the IVD's response was most dependent upon the Young's modulus (YM) of the fibre bundles and therefore this was chosen to be the parameter to optimise. Based on the obtained YM values for each test corresponding to a different strain rate (ε̇), the following relationship was derived:YM=35.5lnε̇+527.5. These properties can be used in finite element models of the IVD that aim to simulate spinal biomechanics across loading rates.