Mohammad Javad Shafiee

Efficient Bayesian inference using fully connected conditional random fields with stochastic cliques

Conditional random fields (CRFs) are one of the most powerful frameworks in image modeling. However practical CRFs typically have edges only between nearby nodes; using more interactions and expressive relations among nodes make these methods impractical for large-scale applications, due to the high computational complexity. Recent work has shown that fully connected CRFs can be tractable by defining specific potential functions. In this paper, we present a novel framework to tackle the computational complexity of a fully connected graph without requiring specific potential functions. Instead, inspired by random graph theory and sampling methods, we propose a new clique structure called stochastic cliques. The stochastically fully connected CRF (SFCRF) is a marriage between random graphs and random fields, benefiting from the advantages of fully connected graphs while maintaining computational tractability. The effectiveness of SFCRF was examined by binary image labeling of highly noisy images. The results show that the proposed framework outperforms an adjacency CRF and a CRF with a large neighborhood size.

Apparent Ultra-High

A promising, recently explored, alternative to ultra-high b-value diffusion weighted imaging (UHB-DWI) is apparent ultra-high b-value diffusion-weighted image reconstruction (AUHB-DWR), where a computational model is used to assist in the reconstruction of apparent DW images at ultra-high b-values. Firstly, we present a novel approach to AUHB-DWR that aims to improve image quality. We formulate the reconstruction of an apparent DW image as a hidden conditional random field (HCRF) in which tissue model diffusion parameters act as hidden states in this random field. The second contribution of this paper is a new generation of fully connected conditional random fields, called the hidden stochastically fully connected conditional random fields (HSFCRF) that allows for efficient inference with significantly reduced computational complexity via stochastic clique structures. The proposed AUHB-DWR algorithms, HCRF and HSFCRF, are evaluated quantitatively in nine different patient cases using Fishers criteria, probability of error, and coefficient of variation metrics to validate its effectiveness for the purpose of improving intensity delineation between expert identified suspected cancerous and healthy tissue within the prostate gland. The proposed methods are also examined using a prostate phantom, where the apparent ultra-high b-value DW images reconstructed using the tested AUHB-DWR methods are compared with real captured UHB-DWI. The results illustrate that the proposed AUHB-DWR methods has improved reconstruction quality and improved intensity delineation compared with existing AUHB-DWR approaches.

b

Deep neural networks are a branch in machine learning that has seen a meteoric rise in popularity due to its powerful abilities to represent and model high-level abstractions in highly complex data. One area in deep neural networks that are ripe for exploration is neural connectivity formation. A pivotal study on the brain tissue of rats found that synaptic formation for specific functional connectivity in neocortical neural microcircuits can be surprisingly well modeled and predicted as a random formation. Motivated by this intriguing finding, we introduce the concept of StochasticNet where deep neural networks are formed via stochastic connectivity between neurons. As a result, any type of deep neural networks can be formed as a StochasticNet by allowing the neuron connectivity to be stochastic. Stochastic synaptic formations in a deep neural network architecture can allow for efficient utilization of neurons for performing specific tasks. To evaluate the feasibility of such a deep neural network architecture, we train a StochasticNet using four different image datasets (CIFAR-10, MNIST, SVHN, and STL-10). Experimental results show that a StochasticNet using less than half the number of neural connections as a conventional deep neural network achieves comparable accuracy and reduces overfitting on the CIFAR-10, MNIST, and SVHN data sets. Interestingly, StochasticNet with less than half the number of neural connections, achieved a higher accuracy (relative improvement in test error rate of ~6% compared to ConvNet) on the STL-10 data set than a conventional deep neural network. Finally, the StochasticNets have faster operational speeds while achieving better or similar accuracy performances.

-Value Diffusion-Weighted Image Reconstruction via Hidden Conditional Random Fields

Multiparametric MRI has shown considerable promise as a diagnostic tool for prostate cancer grading. Diffusion-weighted MRI (DWI) has shown particularly strong potential for improving the delineation between cancerous and healthy tissue in the prostate gland. Current automated diagnostic methods using multiparametric MRI, however, tend to either use low-level features, which are difficult to interpret by radiologists and clinicians, or use highly subjective heuristic methods. We propose a novel strategy comprising a tumor candidate identification scheme and a hybrid textural-morphological feature model for delineating between cancerous and non-cancerous tumor candidates in the prostate gland via multiparametric MRI. Experimental results using clinical multiparametric MRI datasets show that the proposed strategy has strong potential as a diagnostic tool to aid radiologists and clinicians identify and detect prostate cancer more efficiently and effectively.

StochasticNet: Forming Deep Neural Networks via Stochastic Connectivity

The prevalence of compressive sensing is continually growing in all facets of imaging science. Com- pressive sensing allows for the capture and reconstruction of an entire signal from a sparse (under- sampled), yet sufficient, set of measurements that is representative of the target being observed. This compressive sensing strategy reduces the duration of the data capture, the size of the acquired data, and the cost of the imaging hardware as well as complexity while preserving the necessary underlying information. Compressive sensing systems require the accompaniment of advanced re- construction algorithms to reconstruct complete signals from the sparse measurements made. Here, a new reconstruction algorithm is introduced specifically for the reconstruction of compressive multispectral (MS) sensing data that allows for high-quality reconstruction from acquisitions at sub-Nyquist rates. We propose a multilayered conditional random field (MCRF) model, which extends upon the CRF model by incorporating two joint layers of certainty and estimated states. The proposed algorithm treats the reconstruction of each spectral channel as a MCRF given the sparse MS measurements. Since the observations are incomplete, the MCRF incorporates an extra layer determining the certainty of the measurements. The proposed MCRF approach was evaluated using simulated compressive MS data acquisitions, and is shown to enable fast acquisition of MS sensing data with reduced imaging hardware cost and complexity.

Multiparametric MRI prostate cancer analysis via a hybrid morphological-textural model.

We present Temporal Conditional Random Fields, a probabilistic framework for modeling object motion. The state-of-the-art discriminative approach for tracking is known as dynamic conditional random fields. This method models an event based on spatial and temporal relation between pixels in an image sequence without any prediction. To facilitate such a powerful graphical model with prediction and come up with a CRF-based predictor, we propose a set of new temporal relations for object tracking, with feature functions such as optical flow (calculated among consequent frames) and line filed features. We validate our proposed method with real data sequences and will show that the TCRF prediction is nearly equivalent with result of template matching. Experimental results indicate that our TCRF can predict future state of any maneuvering target with nearly zero error during its constant motion. Not only the proposed TCRF has a simple and easy to implement structure, but also it outperforms the state-of-the-art predictors such as Kalman filter.

Reconstruction of compressive multispectral sensing data using a multilayered conditional random field approach

Recent studies have shown that deep neural networks (DNNs) can outperform state-of-the-art algorithms for a multitude of computer vision tasks. However, the ability to leverage DNNs for near real-time performance on embedded systems have been all but impossible so far without requiring specialized processors or GPUs. In this paper, we present a new motion detection algorithm that leverages the power of DNNs while maintaining low computational complexity needed for near real-time embedded performance without specialized hardware. The proposed Neural Response Mixture (NeRM) model leverages rich deep features extracted from the neural responses of an efficient, stochastically-formed deep neural network (StochasticNet) for constructing Gaussian mixture models to detect motion in a scene. NeRM was implemented embedded on an Axis surveillance camera, and results demonstrated that the proposed NeRM approach can achieve strong motion detection accuracy while operating at near real-time performance.

Model-based tracking: Temporal conditional random fields

A new supervised algorithm for document image binarization is proposed.The proposed method uses a discriminative graphical model for binarization.The results are compared with the participants of a famous contest. Binarization is one of the key initial steps in image analysis and system understanding. Different types of document degradations make the binarization a very challenging task. This paper proposes a statistical framework for binarizing degraded document images based on the concept of conditional random fields (CRFs). The CRFs are discriminative graphical models which model conditional distribution and are used in structural classifications. The distribution of binarized images given the degraded ones is modelled with respect to a set of informative features extracted for all sites of the document image. The recent marginal based learning method 5 is used for the estimation of parameters of the model. The proposed graphical framework enables the depending labelling of all the sites of image despite the independent pixel-by-pixel binarization of other methods. The performance of our system is evaluated on different document image datasets and is compared with several well-known binarization methods. Experimental results show comparable performance with respect to other state-of-the-art methods.

Embedded Motion Detection via Neural Response Mixture Background Modeling

The study of oxygen uptake (V̇o2) dynamics during walking exercise transitions adds valuable information regarding fitness. However, direct V̇o2 measurements are not practical for general population under realistic settings. Devices to measure V̇o2 are associated with elevated cost, uncomfortable use of a mask, need of trained technicians, and impossibility of long-term data collection. The objective of this study was to predict the V̇o2 dynamics from heart rate and inputs from the treadmill ergometer by a novel artificial neural network approach. To accomplish this, 10 healthy young participants performed one incremental and three moderate constant work rate treadmill walking exercises. The speed and grade used for the moderate-intensity protocol was related to 80% of the V̇o2 response at the gas exchange threshold estimated during the incremental exercise. The measured V̇o2 was used to train an artificial neural network to create an algorithm able to predict the V̇o2 based on easy-to-obtain inputs. The dynamics of the V̇o2 response during exercise transition were evaluated by exponential modeling. Within each participant, the predicted V̇o2 was strongly correlated to the measured V̇o2 ( = 0.97 ± 0.0) and presented a low bias (~0.2%), enabling the characterization of the V̇o2 dynamics during treadmill walking exercise. The proposed algorithm could be incorporated into smart devices and fitness equipment, making them suitable for tracking changes in aerobic fitness and physical health beyond the infrequent monitoring of patients during clinical interventions and rehabilitation programs.

Document image binarization using a discriminative structural classifier

We present an illumination-compensation method to enable fast and reliable background subtraction under sudden, local illumination changes in wide area surveillance videos. We use Probabilistic Illumination Range Modeling (PIRM) to model the conditional probability distribution of current frame intensity given background intensity. With this model, we can identify a continuous range of current frame intensities that map to the same background intensity, and scale all pixels within that range in the current frame appropriately to enable illumination-compensated background subtraction. Experimental results using a standard academic dataset as well as very challenging industry videos show that PIRM can achieve improvements in compensating for sudden, local illumination changes.