Manning Wang

Classification and analysis of the errors in neuronavigation.

There are many different types of errors in neuronavigation, and the reasons and results of these errors are complex. For a neurosurgeon using the neuronavigation system, it is important to have a clear understanding of when an error may occur, what the magnitude of it is, and how to avoid it or reduce its influence on the final application accuracy. In this article, we classify all the errors into 2 groups according to the working principle of neuronavigation systems. The first group contains the errors caused by the differences between the anatomic structures in the images and that of the real patient, and the second group contains the errors occurring in transforming the position of surgical tools from the patient space to the image space. Each group is further divided into 2 subgroups. We discuss 16 types of errors and classify each of them into one of the subgroups. The classification and analysis of these errors should help neurosurgeons understand the power and limits of neuronavigation systems and use them more properly.

Anatomical landmarks for point-matching registration in image-guided neurosurgery.

Accurate patient to image registration is the core for successful image‐guided neurosurgery. While skin adhesive markers (SMs) are widely used in point‐matching registration, a proper implementation of anatomical landmarks (ALs) may overcome the inconvenience brought by the use of SMs.

A Brain-Deformation Framework Based on a Linear Elastic Model and Evaluation Using Clinical Data

In image-guided neurosurgery, brain tissue displacement and deformation during neurosurgical procedures are a major source of error. In this paper, we implement and evaluate a linear-elastic-model-based framework for correction of brain shift using clinical data from five brain tumor patients. The framework uses a linear elastic model to simulate brain-shift behavior. The model is driven by cortical surface deformations, which are tracked using a surface-tracking algorithm combined with a laser-range scanner. The framework performance was evaluated using displacements of anatomical landmarks, tumor contours and self-defined evaluation parameters. The results show that tumor deformations predicted by the present framework agreed well with the ones observed intraoperatively, especially in the parts of the larger deformations. On average, a brain shift of 3.9 mm and a tumor margin shift of 4.2 mm were corrected to 1.2 and 1.3 mm, respectively. The entire correction process was performed in less than 5 min. The data from this study suggest that the technique is a suitable candidate for intraoperative brain-deformation correction.

Distribution templates of the fiducial points in image-guided neurosurgery.

BACKGROUND Point-pair registration is widely used in an image-guided neurosurgery system. Poor distribution of the fiducial points leads to an increase in the target registration error (TRE). OBJECTIVE This study aimed to provide templates consisting of optimized positioning of the fiducial points to reduce the TRE in image-guided neurosurgery. METHODS We divided the head into 6 regions and provided distribution templates of the fiducial points for each of them. A variable termed TREM(r) was used to express the approximate expected square of the TRE at the target point with a specified distribution of fiducial points. We randomly selected 85 patients from 5 hospitals who underwent image-guided neurosurgery and compared the TREM(r) of the real fiducial points with that of the templates. RESULTS We grouped the patients by hospitals and regions. The mean TREM(r)s of the templates were much smaller than those of the real fiducial points. In each group, the range of the TREM(r) values of the templates was much smaller than that of the real fiducial points. CONCLUSION This study provides an easy method to implement a good distribution of the fiducial points to help reduce TRE in image-guided neurosurgery. The templates are simple and exact and can be easily integrated into current workflow.BACKGROUND Point-pair registration is widely used in an image-guided neurosurgery system. Poor distribution of the fiducial points leads to an increase in the target registration error (TRE). OBJECTIVE This study aimed to provide templates consisting of optimized positioning of the fiducial points to reduce the TRE in image-guided neurosurgery. METHODS We divided the head into 6 regions and provided distribution templates of the fiducial points for each of them. A variable termed TREM(r) was used to express the approximate expected square of the TRE at the target point with a specified distribution of fiducial points. We randomly selected 85 patients from 5 hospitals who underwent image-guided neurosurgery and compared the TREM(r) of the real fiducial points with that of the templates. RESULTS We grouped the patients by hospitals and regions. The mean TREM(r)s of the templates were much smaller than those of the real fiducial points. In each group, the range of the TREM(r) values of the templates was much smaller than that of the real fiducial points. CONCLUSION This study provides an easy method to implement a good distribution of the fiducial points to help reduce TRE in image-guided neurosurgery. The templates are simple and exact and can be easily integrated into current workflow.

Alteration of the Intra- and Cross- Hemisphere Posterior Default Mode Network in Frontal Lobe Glioma Patients.

Patients with frontal lobe gliomas often experience neurocognitive dysfunctions before surgery, which affects the default mode network (DMN) to different degrees. This study quantitatively analyzed this effect from the perspective of cerebral hemispheric functional connectivity (FC). We collected resting-state fMRI data from 20 frontal lobe glioma patients before treatment and 20 healthy controls. All of the patients and controls were right-handed. After pre-processing the images, FC maps were built from the seed defined in the left or right posterior cingulate cortex (PCC) to the target regions determined in the left or right temporal-parietal junction (TPJ), respectively. The intra- and cross-group statistical calculations of FC strength were compared. The conclusions were as follows: (1) the intra-hemisphere FC strength values between the PCC and TPJ on the left and right were decreased in patients compared with controls; and (2) the correlation coefficients between the FC pairs in the patients were increased compared with the corresponding controls. When all of the patients were grouped by their tumor’s hemispheric location, (3) the FC of the subgroups showed that the dominant hemisphere was vulnerable to glioma, and (4) the FC in the dominant hemisphere showed a significant correlation with WHO grade.

A new markerless patient‐to‐image registration method using a portable 3D scanner

PURPOSE Patient-to-image registration is critical to providing surgeons with reliable guidance information in the application of image-guided neurosurgery systems. The conventional point-matching registration method, which is based on skin markers, requires expensive and time-consuming logistic support. Surface-matching registration with facial surface scans is an alternative method, but the registration accuracy is unstable and the error in the more posterior parts of the head is usually large because the scan range is limited. This study proposes a new surface-matching method using a portable 3D scanner to acquire a point cloud of the entire head to perform the patient-to-image registration. METHODS A new method for transforming the scan points from the device space into the patient space without calibration and tracking was developed. Five positioning targets were attached on a reference star, and their coordinates in the patient space were measured prior. During registration, the authors moved the scanner around the head to scan its entire surface as well as the positioning targets, and the scanner generated a unique point cloud in the device space. The coordinates of the positioning targets in the device space were automatically detected by the scanner, and a spatial transformation from the device space to the patient space could be calculated by registering them to their coordinates in the patient space that had been measured prior. A three-step registration algorithm was then used to register the patient space to the image space. The authors evaluated their method on a rigid head phantom and an elastic head phantom to verify its practicality and to calculate the target registration error (TRE) in different regions of the head phantoms. The authors also conducted an experiment with a real patients data to test the feasibility of their method in the clinical environment. RESULTS In the phantom experiments, the mean fiducial registration error between the device space and the patient space, the mean surface registration error, and the mean TRE of 15 targets on the surface of each phantom were 0.34 ± 0.01 mm and 0.33 ± 0.02 mm, 1.17 ± 0.02 mm and 1.34 ± 0.10 mm, and 1.06 ± 0.11 mm and 1.48 ± 0.21 mm, respectively. When grouping the targets according to their positions on the head, high accuracy was achieved in all parts of the head, and the TREs were similar across different regions. The authors compared their method with the current surface registration methods that use only a part of the facial surface on the elastic phantom, and the mean TRE of 15 targets was 1.48 ± 0.21 mm and 1.98 ± 0.53 mm, respectively. In a clinical experiment, the mean TRE of seven targets on the patients head surface was 1.92 ± 0.18 mm, which was sufficient to meet clinical requirements. CONCLUSIONS The proposed surface-matching registration method provides sufficient registration accuracy even in the posterior area of the head. The 3D point cloud of the entire head, including the facial surface and the back of the head, can be easily acquired using a portable 3D scanner. The scanner does not need to be calibrated prior or tracked by the optical tracking system during scanning.

Fast and robust multimodal image registration using a local derivative pattern

Purpose: Deformable multimodal image registration, which can benefit radiotherapy and image guided surgery by providing complementary information, remains a challenging task in the medical image analysis field due to the difficulty of defining a proper similarity measure. This article presents a novel, robust and fast binary descriptor, the discriminative local derivative pattern (dLDP), which is able to encode images of different modalities into similar image representations. Methods: dLDP calculates a binary string for each voxel according to the pattern of intensity derivatives in its neighborhood. The descriptor similarity is evaluated using the Hamming distance, which can be efficiently computed, instead of conventional L1 or L2 norms. For the first time, we validated the effectiveness and feasibility of the local derivative pattern for multimodal deformable image registration with several multi‐modal registration applications. Results: dLDP was compared with three state‐of‐the‐art methods in artificial image and clinical settings. In the experiments of deformable registration between different magnetic resonance imaging (MRI) modalities from BrainWeb, between computed tomography and MRI images from patient data, and between MRI and ultrasound images from BITE database, we show our method outperforms localized mutual information and entropy images in terms of both accuracy and time efficiency. We have further validated dLDP for the deformable registration of preoperative MRI and three‐dimensional intraoperative ultrasound images. Our results indicate that dLDP reduces the average mean target registration error from 4.12 mm to 2.30 mm. This accuracy is statistically equivalent to the accuracy of the state‐of‐the‐art methods in the study; however, in terms of computational complexity, our method significantly outperforms other methods and is even comparable to the sum of the absolute difference. Conclusions: The results reveal that dLDP can achieve superior performance regarding both accuracy and time efficiency in general multimodal image registration. In addition, dLDP also indicates the potential for clinical ultrasound guided intervention.