GPU Accelerated Real Time Rotation, Scale and Translation Invariant Image Registration Method

Brain shift is an important obstacle for the application of image guidance during neurosurgical interventions. There has been a growing interest in intra-operative imaging systems to update the image-guided surgery systems with real-time data. However, due to the innate limitations of the current imaging modalities, accurate and real-time brain shift compensation remains as a challenging problem. In this study, application of the intra-operative photoacoustic (PA) imaging and registration of the intra-operative PA images with pre-operative brain MR images is proposed to compensate brain deformation during surgery. Finding a satisfactory multimodal image registration method is a challenging problem due to complicated and unpredictable nature of brain deformation. In this study, the co-sparse analysis model is proposed for PA-MR image registration which can capture the interdependency of two modalities. The proposed algorithm works based on the minimization of mapping transform by using a pair of analysis operators. These operators are learned by the alternating direction method of multipliers. The method was evaluated using experimental phantom and ex-vivo data obtained from mouse brain. The results of phantom data show about 60% and 63% improvement in root mean square error (RMSE) and target registration error (TRE) in comparison with commonly used normalized mutual information registration method. In addition, the results of mouse brain and phantom data shown more accurate performance for PA versus ultrasound imaging for brain shift calculation. Finally, by using the proposed registration method, the intra-operative PA images could become a promising tool when the brain shift invalidated pre-operative MRI.

Download Full-text

An Underwater Image Real-Time Registration Approach Based on SURF

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.437.888 ◽

2013 ◽

Vol 437 ◽

pp. 888-893 ◽

Cited By ~ 1

Author(s):

Chao Li ◽

Yong Jie Pang ◽

Ming Wei Sheng ◽

Hai Huang

Keyword(s):

Image Registration ◽

Real Time ◽

Median Filter ◽

Filter Method ◽

Interest Points ◽

Registration Method ◽

Speeded Up Robust Features ◽

Time Performance ◽

Underwater Image ◽

Image Contrast Enhancement

In order to meet the demands of real-time performance and robustness for underwater image registration, a novel image registration method based on the SURF (Speeded-Up Robust Features) algorithm is proposed. During the image acquisition process, noise was generated inevitably because of many influencing factors such as atmospheric turbulence, camera defocus during image capturing or relative motion between the camera and the object. Firstly, median filter method was involved during the image preprocessing for underwater image contrast enhancement. Secondly, the SURF algorithm was used to obtain the interest points of the reference and registering images, and the nearest neighbor method was applied to search for coarse matching points. To obtain the precise matching points, the dominant orientations of the coarse matching points were used to eliminate the mismatching points. Finally, the precise matching points were adapted to calculate the mapping relationship between the registering and reference images, the bilinear interpolation method was applied to resample the registering image, and then the registered image was obtained. Experimental results indicated that the proposed preprocessing methods obviously enhanced the image quality, and the introduced image registration approach effectively improved the real-time performance and guaranteed the robustness at the same time.

Download Full-text

Semi-Supervised Deep Learning-Based Image Registration Method with Volume Penalty for Real-Time Breast Tumor Bed Localization

Sensors ◽

10.3390/s21124085 ◽

2021 ◽

Vol 21 (12) ◽

pp. 4085

Author(s):

Marek Wodzinski ◽

Izabela Ciepiela ◽

Tomasz Kuszewski ◽

Piotr Kedzierawski ◽

Andrzej Skalski

Keyword(s):

Deep Learning ◽

Image Registration ◽

Real Time ◽

Breast Tumor ◽

Tumor Resection ◽

Registration Method ◽

Registration Error ◽

Target Registration Error ◽

Tumor Bed ◽

Registration Process

Breast-conserving surgery requires supportive radiotherapy to prevent cancer recurrence. However, the task of localizing the tumor bed to be irradiated is not trivial. The automatic image registration could significantly aid the tumor bed localization and lower the radiation dose delivered to the surrounding healthy tissues. This study proposes a novel image registration method dedicated to breast tumor bed localization addressing the problem of missing data due to tumor resection that may be applied to real-time radiotherapy planning. We propose a deep learning-based nonrigid image registration method based on a modified U-Net architecture. The algorithm works simultaneously on several image resolutions to handle large deformations. Moreover, we propose a dedicated volume penalty that introduces the medical knowledge about tumor resection into the registration process. The proposed method may be useful for improving real-time radiation therapy planning after the tumor resection and, thus, lower the surrounding healthy tissues’ irradiation. The data used in this study consist of 30 computed tomography scans acquired in patients with diagnosed breast cancer, before and after tumor surgery. The method is evaluated using the target registration error between manually annotated landmarks, the ratio of tumor volume, and the subjective visual assessment. We compare the proposed method to several other approaches and show that both the multilevel approach and the volume regularization improve the registration results. The mean target registration error is below 6.5 mm, and the relative volume ratio is close to zero. The registration time below 1 s enables the real-time processing. These results show improvements compared to the classical, iterative methods or other learning-based approaches that do not introduce the knowledge about tumor resection into the registration process. In future research, we plan to propose a method dedicated to automatic localization of missing regions that may be used to automatically segment tumors in the source image and scars in the target image.

Download Full-text