RESEARCH:MEDICAL

 

Medical Image Compounding, Restoration and Enhancement

 

    Hongcheng Wang and Yunqiang Chen and Tong Fang and Jason Tyan and Narendra Ahuja

 

 

INTRODUCTION


The rapid growing field of biomedical imaging has composed a great challenge and opportunity to image processing tasks. It is usually very difficult to obtain accurate prior models for many biomedical imaging modalities, such as X-ray, ultrasound, magnetic resonance images (MRI), CT and others. For example, the ultrasound speckle noise is non-stationary and changes according to ultrasound attenuation and the sub-resolution scatterers in the tissue, and for CT, several adjustable protocol factors, such as tube current-time product and slice thickness, have effects on image noise. With the understanding of physical and image properties of medical imaging,
we developed computational algorithms from two aspects: (1) Gradient adaptive image restoration and enhancement: The method and system we developed based on adaptive gradient modification suppress noise, preserve image structures, and enhance
small structures (e.g. thin vessels) [3]; and (2) Multi-image compounding: We developed a novel mutual information regularized Bayesian framework which effectively utilizes the traditional generative signal/noise models but is much more robust to various model errors, to explicitly enforce the noise independence for multi-image restoration [1,2].

 

 

PUBLICATIONS


1. Yunqiang Chen, Hongcheng Wang, Tong Fang, and Jason Tyan, Mutual Information Regularized Bayesian Framework for Multiple Image Restoration, in Proc. IEEE International Conference on Computer Vision (ICCV), 2005

Abstract: Bayesian methods have been extensively used in various applications. However, there are two intrinsic issues rarely addressed, namely generalization and validity. In the context of multiple image restoration, we show that traditional Bayesian methods are sensitive to model errors and cannot guarantee valid results satisfying the underlying prior knowledge, e.g. independent noise property. To improve the Bayesian framework's generalization, we propose to explicitly enforce the validity of the result. Independent noise prior is very important but largely under-utilized in previous literature. In this paper, we use mutual information (MI) to explicitly enforce the independence. Efficient approximations based on Taylor expansion are proposed to adapt MI into standard energy forms to regularize the Bayesian methods. The new regularized Bayesian framework effectively utilizes the traditional generative signal/noise models but is much more robust to various model errors, as demonstrated in experiments on some demanding imaging applications.

Full Text:   PDF (314KB)  

BibTex:

@inproceedings{1097647,
 author = {Yunqiang Chen and Hongcheng Wang and Tong Fang and Jason Tyan},
 title = {Mutual Information Regularized Bayesian Framework for Multiple Image Restoration},
 booktitle = {ICCV '05: Proceedings of the Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1},
 year = {2005},
 isbn = {0-7695-2334-X-01},
 pages = {190--197},
 doi = {http://dx.doi.org/10.1109/ICCV.2005.164},
 publisher = {IEEE Computer Society},
 address = {Washington, DC, USA},
 }

2. Yunqiang Chen, Hongcheng Wang, Tong Fang and Jason Tyan, Image Compounding based on Independent Noise Constraint, in Proc. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2005

Abstract: Image restoration has been extensively studied in the past. But multi-image based restoration/compounding is still surprisingly primitive. It usually starts with weighted averaging of the multiple images followed by single-image based restoration methods, which discards the abundant information hinted in the multiple images that can help the restoration process. In this paper, we utilize the fact that the images are corrupted by independent noise and design a new independence measurement based on the properties of independent random variables. The new independence measurement can be efficiently evaluated and imposed as an energy term into the traditional Maximum a Posteriori (MAP) framework, compensating to the generative models of signal and noise. It can effectively prevent the signal from being smoothed out as noise and hence dramatically improve the restoration quality and robustness, especially when accurate noise/signal models are difficult to obtain. Experiments on real medical images show very promising results.

Full Text:   PDF (199KB)  

 

3. Hongcheng Wang, Yunqiang Chen, Tong Fang, Jason Tyan and Narendra Ahuja, Gradient Adaptive Image Restoration and Enhancement, submitted, 2006

Abstract: (available soon!)

Full Text:   PDF (available soon!)  

 

ILLUSTRATION


 
 

 (c)

(a)

(b)

   
Figure (a) and (b) are two of the three input noisy images captured using different frequency, and Figure (c) is the output using our mutual information regularized Bayesian framework described in [1].
     


Updated: Jan.1, 2006