Proceedings IEEE International Symposium on Biomedical Imaging

We provide a unified framework for smoothing noisy brain image data along attributes of choice derived from diffusion tensor imaging. The framework is based on a variational segmentation functional approach that outputs smoothed regions within the white matter that are relatively homogeneous with respect to specific diffusion tensor image properties. The smoothed tensor fields and the associated edge fields are recovered in a number of ways, thus illustrating the applicability of the proposed unified framework for smoothing and feature extraction in support of the anatomic identification of white matter fiber systems in the human brain.

In 2D multislice MRI, the resolution in the slice direction is often worse than the in-plane resolution. For certain diagnostic imaging applications, isotropic resolution is necessary but true 3D acquisition methods are not practical. In this case, if the imaging volume is acquired two or more times, with small spatial shifts between acquisitions, combination of the data sets using an iterative superresolution algorithm gives improved resolution and better edge definition in the slice-select direction.

We present a new fast backprojection algorithm for CT fan-beam reconstruction. The new algorithm operates directly on fan-beam data without prior rebinning to parallel-beam projections. The algorithm reduces the computational complexity from O(N/sup 3/) for the traditional fan-beam algorithm to O(N/sup 2/ log N). Simulations demonstrate speedups of greater than 50-fold for a 512 /spl times/ 512 image, with no perceivable degradation in accuracy. The algorithm also applies to multi-slice helical 3D reconstruction, and extends to 3D cone-beam reconstruction.

Fusion of intravascular ultrasound (IVUS) and biplane X-ray angiography leads to a geometrically correct representation of coronary vessels. A comparison of three volume quantification methods - polytope, Watanabe, and Simpson-based - is reported. The three methods allow local estimation of plaque accumulation. To determine volumetric indices, space between the lumen and adventitia surfaces is subdivided and each of the volume elements considered separately to achieve volume quantification. Polyhedral volume elements are employed and their volumes estimated by each of the proposed approaches. The volume quantification methods were validated in 92 computer-generated shapes and tested on routine patient data. All three methods were found highly accurate. By comparing the volume quantification errors, the polytope and Watanabe methods were found statistically significantly more accurate than the Simpson-based approach.

Trong thập kỷ qua, đã có những tiến bộ nhanh chóng trong việc phát triển các kỹ thuật hình ảnh cộng hưởng từ (MRI) cung cấp thông tin giải phẫu, chức năng và phân tử cụ thể. Song song với những phát triển này, có sự tiến bộ to lớn trong di truyền phân tử, đã tạo ra các công cụ mạnh mẽ cho việc hiểu và thao tác cơ sở di truyền của chức năng tế bào và mô. Hiện nay, đang có sự quan tâm ngày càng tăng trong việc kết hợp những tiến bộ trong MRI với tiến bộ trong di truyền phân tử để phát triển các phương pháp hiệu quả nhằm xác định các thay đổi di truyền cụ thể trên sinh lý phức tạp của mô nguyên vẹn. Một số ví dụ về các phát triển mới trong MRI được đưa ra với ứng dụng cho việc phân loại các mô hình chuột chuyển gen và knockout cụ thể. Một số giả thuyết được đưa ra về việc tích hợp MRI vào di truyền chức năng.

We present a new method for estimating heart motion from two-dimensional (2D) echocardiographic sequences. It is inspired by the Lucas-Kanade algorithm for optical flow which estimates motion parameters over a sliding window. However, instead of assuming that the motion is constant within the analysis window, we consider a model that is locally affine and can account for typical heart motions such as dilation/contraction and shear. Another refinement is spatial adaptivity which is achieved by estimating displacement vectors at multiple scales and selecting the most promising fit. The affine parameters are estimated in the least squares sense using a separable spatial (resp., spatio-temporal) B-spline window. This particular choice is motivated by the fact that the B-splines are nearly isotropic (Gaussian-like) and that they satisfy a two-scale equation. We use this latter property to derive a wavelet-like algorithm that leads to a fast computation of B-spline-weighted inner products and moments at dyadic scales, which speeds up our method considerably. We test the algorithm on synthetic and real ultrasound sequences and show that it compares favorably with other methods, such as Lucas-Kanade and Horn-Schunk.

The imaging process in a transmission electron microscope (TEM) produces a number of artifacts including the contrast transfer function (CTF) and envelope functions. In addition, when electron cryomicroscopy is performed for purposes of single particle reconstruction, signal to noise ratios are very low, generally peaking between 0.1 and 1.0, and averaging 0.01 to 0.1. We apply a semi-empirical model of the microscope artifacts to optimally correct data used for 3D single particle reconstructions.

Heating can enhance gene transfection/expression. We developed a novel endovascular microwave heating source using an MR imaging-guidewire (MRIG) to deliver external microwave energy into the target vessels. We investigated (i) microwave power loss along differently-sized MRIGs; (ii) microwave power distribution around MRIGs; and (iii) temperature increases versus input microwave powers with MRIGs. Then, we evaluated, in vivo, the possibility of using the MR imaging/heating guidewire to generate high-resolution MR images and deliver the controlled therapeutic heat energy into the target vessel wall with histological confirmation. The results showed that the MRIG had no thermal damage to the vessel wall and can be used as an intravascular local heating device. This study has established the groundwork for further efficient management of cardiovascular atherosclerotic diseases using MRI-based vascular gene therapy.

Diffraction Enhanced Imaging is a phase sensitive imaging modality, which recently has been demonstrated to be a powerful tool for improving the visibility of tiny low absorbing structures. Especially in medical imaging the capability to detect small details, not visible in a conventional radiograph is of relevant importance. An analyzer crystal is used as angular filter to select or reject X-rays that are deviated when traversing a sample if a gradient of the real part of the refractive index is encountered. At the medical imaging beamline of the synchrotron radiation facility Elettra (Trieste) experiments were carried out to assess the applicability of this technique to materials where a combined refraction from a large amount of small structures, not detectable separately, gives rise to an overall effect of multiple scattering. Tuning the analyzer crystal in different angular positions the variation of the contrast was studied. Strong contrast enhancement was measured in comparison to the conventional absorption image.

Fourier methods are a natural choice for reconstructing magnetic resonance images (MRI). Unfortunately, however, due to the many different tissues normally present in each scan, the Gibbs ringing artifact often hinders accurate reconstruction. These effects are exacerbated in the presence of random noise, which is inherent in MRI spectral data. Recently, numerical edge detection and reconstruction methods have been developed that effectively reduce the Gibbs oscillations while maintaining high resolution accuracy at the edges. This paper addresses the issue of noise in MRI reconstruction and its effects on the ability to recover the image. The numerical method we apply here not only recovers the images with very high accuracy, but it is also robust in the presence of noise.