Yuichiro Hayashi

Multi-organ Segmentation Based on Spatially-Divided Probabilistic Atlas from 3D Abdominal CT Images

This paper presents an automated multi-organ segmentation method for 3D abdominal CT images based on a spatially-divided probabilistic atlases. Most previous abdominal organ segmentation methods are ineffective to deal with the large differences among patients in organ shape and position in local areas. In this paper, we propose an automated multi-organ segmentation method based on a spatially-divided probabilistic atlas, and solve this problem by introducing a scale hierarchical probabilistic atlas. The algorithm consists of image-space division and a multi-scale weighting scheme. The generated spatial-divided probabilistic atlas efficiently reduces the inter-subject variance in organ shape and position either in global or local regions. Our proposed method was evaluated using 100 abdominal CT volumes with manually traced ground truth data. Experimental results showed that it can segment the liver, spleen, pancreas, and kidneys with Dice similarity indices of 95.1%, 91.4%, 69.1%, and 90.1%, respectively.

Intraoperative subcortical mapping of a language-associated deep frontal tract connecting the superior frontal gyrus to Broca's area in the dominant hemisphere of patients with glioma

OBJECT The deep frontal pathway connecting the superior frontal gyrus to Brocas area, recently named the frontal aslant tract (FAT), is assumed to be associated with language functions, especially speech initiation and spontaneity. Injury to the deep frontal lobe is known to cause aphasia that mimics the aphasia caused by damage to the supplementary motor area. Although fiber dissection and tractography have revealed the existence of the tract, little is known about its function. The aim of this study was to determine the function of the FAT via electrical stimulation in patients with glioma who underwent awake surgery. METHODS The authors analyzed the data from subcortical mapping with electrical stimulation in 5 consecutive cases (3 males and 2 females, age range 40-54 years) with gliomas in the left frontal lobe. Diffusion tensor imaging (DTI) and tractography of the FAT were performed in all cases. A navigation system and intraoperative MRI were used in all cases. During the awake phase of the surgery, cortical mapping was performed to find the precentral gyrus and Brocas area, followed by tumor resection. After the cortical layer was removed, subcortical mapping was performed to assess language-associated fibers in the white matter. RESULTS In all 5 cases, positive responses were obtained at the stimulation sites in the subcortical area adjacent to the FAT, which was visualized by the navigation system. Speech arrest was observed in 4 cases, and remarkably slow speech and conversation was observed in 1 case. The location of these sites was also determined on intraoperative MR images and estimated on preoperative MR images with DTI tractography, confirming the spatial relationships among the stimulation sites and white matter tracts. Tumor removal was successfully performed without damage to this tract, and language function did not deteriorate in any of the cases postoperatively. CONCLUSIONS The authors identified the left FAT and confirmed that it was associated with language functions. This tract should be recognized by clinicians to preserve language function during brain tumor surgery, especially for tumors located in the deep frontal lobe on the language-dominant side.

Neurosurgical robotic system for brain tumor removal

PurposeBrain tumor (e.g., glioma) resection surgery, representing the first step for many treatments, is often difficult and time-consuming for neurosurgeons. Thus, intelligent neurosurgical instruments have been developed to improve tumor removal.MethodsThe concept and robotic structure of intelligent neurosurgical instruments were introduced. These instruments consist of a surgical robot, a master device and operating software. The robot incorporates a surgical motion base and tool manipulator, including a volume control suction tool. Open Core Control software was developed for connecting intelligent neurosurgical instruments through a network connection and integrating the instruments into a system.ResultsMechanical evaluation tests on the components and a preliminary system evaluation were performed. A phantom model was fixed on a head frame, and a tumor-removal procedure was successfully performed using prototype intelligent neurosurgical instruments.ConclusionIntelligent neurosurgical instruments are feasible and suitable for on-going evaluation in practical tasks, including in-vivo animal testing.

A method for detecting undisplayed regions in virtual colonoscopy and its application to quantitative evaluation of fly-through methods1 ☆

Rationale and Objectives. When virtual endoscopy is used as a diagnostic tool, especially as a tool for detecting colon polyps, the user often performs automated fly-through based on automatically generated paths. In the case of automated fly-through in the colon, there are some blind areas at the backs of folds. The aim of this study is to detect undisplayed regions during fly-through and to perform quantitative evaluation. Materials and Methods. Undisplayed regions are detected by marking displayed triangles for surface rendering or displayed voxels for volume rendering. The voxels or triangles not having displayed marks are considered to be undisplayed triangles or voxels. Various kinds of automated fly-through paths generated from medial axes of the colon and flattened views of the colon from the viewpoint of the rate of undisplayed regions are evaluated. Results. The experiment results show that about 30% of colon regions are classified as undisplayed regions by the conventional automated fly-through along the medial axis and that the flattened view results in very few undisplayed regions. Conclusion. There is a possibility that the automated fly-through methods may cause many undisplayed regions.

Magnetically guided 3-dimensional virtual neuronavigation for neuroendoscopic surgery: technique and clinical experience.

OBJECTIVE The authors have developed a novel intraoperative neuronavigation with 3-dimensional (3D) virtual images, a 3D virtual navigation system, for neuroendoscopic surgery. The present study describes this technique and clinical experience with the system. METHODS Preoperative imaging data sets were transferred to a personal computer to construct virtual endoscopic views with image segmentation software. An electromagnetic tracker was used to acquire the position and orientation of the tip of the neuroendo-scope. Virtual endoscopic images were interlinked to an electromagnetic tracking system and demonstrated on the navigation display in real time. Accuracy and efficacy of the 3D virtual navigation system were evaluated in a phantom test and on 5 consecutive patients undergoing neuroendoscopic surgery. RESULTS Virtual navigation views were consistent with actual endoscopic views and trajectory in both phantom testing and clinical neuroendoscopic surgery. Anatomic structures that can affect surgical approaches were adequately predicted with the virtual navigation system. The virtual semitransparent view contributed to a clear understanding of spatial relationships between surgical targets and surrounding structures. Surgical procedures in all patients were performed while confirming with virtual navigation. In neurosurgery with a flexible neuroscope, virtual navigation also demonstrated anatomic structures in real time. CONCLUSION The interactive method of intraoperative visualization influenced the decision-making process during surgery and provided useful assistance in identifying safe approaches for neuroendoscopic surgery. The magnetically guided navigation system enabled navigation of surgical targets in both rigid and flexible endoscopic surgeries.

Automated anatomical labeling of abdominal arteries and hepatic portal system extracted from abdominal CT volumes

This paper proposes a method for automated anatomical labeling of abdominal arteries and a hepatic portal system. In abdominal surgeries, understanding blood vessel structure is critical since it is very complicated. The input of the proposed method is the blood vessel region extracted from the CT volume. The blood vessel region is expressed as a tree structure by applying a thinning process to it and compute the mapping from the branches in the tree structure to the anatomical names. First, several characteristic anatomical names are assigned by rule-based pre-processing. The branches assigned to these names are used as references. The remaining blood vessel names are assigned using a likelihood function trained by a machine-learning technique. Simple rule-based postprocessing can correct several blood vessel names. The output of the proposed method is a tree structure with anatomical names. In an experiment using 50 blood vessel regions manually extracted from abdominal CT volumes, the recall and precision rates of the abdominal arteries were 86.2% and 85.3%, and they were 86.5% and 79.5% for the hepatic portal system.

Surgical bedside master console for neurosurgical robotic system

PurposeWe are currently developing a neurosurgical robotic system that facilitates access to residual tumors and improves brain tumor removal surgical outcomes. The system combines conventional and robotic surgery allowing for a quick conversion between the procedures. This concept requires a new master console that can be positioned at the surgical bedside and be sterilized.MethodsThe master console was developed using new technologies, such as a parallel mechanism and pneumatic sensors. The parallel mechanism is a purely passive 5-DOF (degrees of freedom) joystick based on the author’s haptic research. The parallel mechanism enables motion input of conventional brain tumor removal surgery with a compact, intuitive interface that can be used in a conventional surgical environment. In addition, the pneumatic sensors implemented on the mechanism provide an intuitive interface and electrically isolate the tool parts from the mechanism so they can be easily sterilized.ResultsThe 5-DOF parallel mechanism is compact (17 cm width, 19cm depth, and 15cm height), provides a 505,050 mm and 90° workspace and is highly backdrivable (0.27N of resistance force representing the surgical motion). The evaluation tests revealed that the pneumatic sensors can properly measure the suction strength, grasping force, and hand contact. In addition, an installability test showed that the master console can be used in a conventional surgical environment.ConclusionThe proposed master console design was shown to be feasible for operative neurosurgery based on comprehensive testing. This master console is currently being tested for master-slave control with a surgical robotic system.

Multi-organ segmentation from 3D abdominal CT images using patient-specific weighted-probabilistic atlas

Organ segmentation of CT volumes is a basic function of computer-aided diagnosis and surgery-assistance systems. Many of these systems implement organ segmentation methods that are limited to specific organs and that are not robust in dealing with inter-subject differences of organ shape or position. In this paper, we propose an automated method for multi-organ segmentation of abdominal 3D CT volumes by using a patient-specific, weighted-probabilistic atlas for organ position. This is achieved in a two-step process. First, we prepare for segmentation by dividing an atlas database into multiple clusters. This is done using pairs of a training image and the corresponding manual segmentation data set. In the next step, we choose a cluster whose template image is the most similar to the target image. We then weight all of the atlases in the selected cluster by calculating the similarities between the atlases and the target image to dynamically generate a specific probabilistic atlas for each target image. We use the generated probabilistic atlas in MAP estimation to obtain a rough segmentation result and then refine it by using a graph-cut method. Our method can simultaneously segment four organs: the liver, spleen, pancreas and kidneys. Our weighting scheme greatly reduces segmentation error due to inter-subject differences. We applied our method to 100 cases of CT volumes and thus showed that it could segment the liver, spleen, pancreas and kidneys with Dice similarity coefficients of 95.2%, 89.7%, 69.6%, and 89.4%, respectively.

Development of a navigation-based CAD system for colon

We propose a navigation-based computer aided diagnosis (CAD) system for the colon. When diagnosing the colon using virtual colonoscopy (VC), a physician makes a diagnosis by navigating (flying-through) the colon. However, the viewpoints and the viewing directions must be changed many times because the colon is a very long and winding organ with many folds. This is a time-consuming task for physicians. We propose a new navigation-based CAD system for the colon providing virtual unfolded (VU) views, which enables physicians to observe a large area of the colonic wall at a glance. This system generates VU, VC, and CT slice views that are perfectly synchronized. Polyp candidates, which are detected automatically, are overlaid on them. We applied the system to abdominal CT images. The experimental results showed that the system effectively generates VU views for observing colon regions.

Development of a real-time tactile sensing system for brain tumor diagnosis

PurposeTactile sensing techniques may distinguish tumor from healthy tissue and have potential for intraoperative brain tumor diagnosis. The aim of this study is to develop a biocompatible real-time sensing system to measure tactile information such as softness and smoothness, and its application to brain tumor diagnosis.MethodsAn active tactile sensor is developed using balloon expansion. This compact system provides instantaneous tactile information and has potential for brain tumor diagnosis. Measurements are obtained on soft samples with different stiffness and surface condition with testing of boundary condition influence on thickness and area of the object. Then, measurements on white matter and gray matter of porcine ex vivo brain are done as the first step for brain tumor diagnosis.ResultsThe sensor can discriminate samples with different stiffness and surface condition subject to influence by boundary conditions. The sensor can evaluate an object relatively under the same boundary conditions but requires enough thickness and area to evaluate absolutely. Measurements on brain show that the sensor can discriminate between white matter and gray matter.ConclusionsAlthough the sensor has problems on absolute evaluation, results show that the sensor can evaluate tactile information, and it has potential for brain tumor diagnosis.