This study was conducted to evaluate the lesion detection capabilities of anatomically based maximum a posteriori (MAP) image reconstruction methods in emission computed tomography using the computer observer model. In lesion detection tasks, conventional anatomically based MAP reconstruction methods cannot preserve lesions not present in the anatomical image with high contrast and at the same time suppress noise in the background regions. We previously proposed a new anatomically based MAP reconstruction method called the SOS-MAP method, which is based on the spots-on-smooth image model in which the image is modeled by the sum of the smooth background image and the sparse spot image, and showed that the SOS-MAP method can overcome the above-mentioned drawback of conventional anatomically based MAP methods. However, the lesion detection capabilities of the SOS-MAP method remained to be clarified. In the present study, the computer observer model was used to evaluate the lesion detection capabilities of the SOS-MAP method, and it was found that the SOS-MAP method is superior to conventional anatomically based MAP methods for the detection of lesions.

View full abstract

The recent development of small semiconductor photodetectors such as multi-pixel photon counters (MPPCs) that can replace conventional photomultiplier tubes is leading researchers to create new PET detectors. The National Institute of Radiological Sciences is therefore developing a next-generation PET detector known as "X'tal Cube" in collaboration with Chiba University, Tokyo University, and Hamamatsu Photonics K.K. One of the most challenging aspects of this detector is implementing the effective detection of scintillation photons by optically covering all 6 surfaces of a segmented crystal block with MPPCs. From the viewpoint of positional determination, however, a redundant signal from the scintillation photons is acquired for the X'tal Cube. In this report, we propose a new method for improving the accuracy of positional determination by selecting data containing a large amount of positional information. The proposed method is combined with the conventional Anger-type calculation, which is known as a center of gravity operation method for positional estimation. The results of Monte Carlo simulations showed that selecting the data of only 4 effective faces resulted in superior determination performance as compared to using the data of all 6 faces.

View full abstract

In current standard radiotherapy of lung cancer, an excessive volume of normal tissues may be irradiated due to respiratory displacement. If the motion of the lungs could be predicted by computer simulation, radiation exposure to normal tissues could be reduced. Lung deformation is modeled as an elastic material. The lung model is created automatically from computed tomography (CT) images using binarization, mathematical morphology, and region growing. The effects of abdominal breathing and pleural sliding are incorporated into the boundary conditions by template matching between CT images acquired during the inhalation and exhalation phases. Lung motion is then predicted using the moving particle simulation (MPS) method. Since CT images are required for radiotherapy planning, we propose a generation method for simulation-based 4DCT equivalent to actual 4DCT using CT images acquired during the inhalation and exhalation phases and particle simulation results obtained using the image warping technique.

View full abstract

This report describes the improvements of a naïve Bayes model that infers the diagnosis of pulmonary nodules in chest CT images based on the findings obtained when a radiologist interprets the CT images. We have previously introduced an inference model using a naïve Bayes classifier and have reported its clinical value based on evaluation using clinical data. In the present report, we introduce the following improvements to the original inference model: (1) the selection of findings based on correlations and the generation of a model using only these findings, and (2) the introduction of classifiers that integrate several simple classifiers each of which is specialized for specific diagnosis. These improvements were found to increase the inference accuracy by 10.4% (p < .01) as compared to the original model in 100 cases (222 nodules) based on leave-one-out evaluation.

View full abstract

This report presents a new method for automatic scan prescription planning of both the scan planes and Navigator Tracker locations in MRI liver scanning. A 3D dataset acquired using a fast T1 sequence is preprocessed and converted to 2D projection images to avoid the need for complicated and time-consuming 3D segmentation. A 2D active shape model (ASM) is applied to the 2D coronal projection data, and the external shape of the liver dataset is initially extracted based on a rough estimate of the inferior edge of the liver from the 2D projection dataset. The scan plane locations are determined from the inferior and superior edges of the shape model. Navigator Tracker, which is used for motion-compensated MR abdominal MR scanning, establishes the location by selecting one of the landmarks in the ASM. Datasets obtained from 38 volunteers were tested. Good results were obtained, suggesting that this method may prove to be useful for clinical diagnosis.

View full abstract

The objective of the present study was to establish a histology-based 3D reconstruction procedure and thus to improve our understanding of the tumor microenvironment. Based on a pathodiagnostic survey, tissue specimens (3 mm in diameter) were dissected out from the invasion front of tongue squamous cell carcinoma (SCC) using a paraffin tissue microarray. The experimental protocol comprised step-wise immunolabeling of serially cut sections (4 μm thick) and image acquisition using a Virtual Microscopy system (Nano-Zoomer, Hamamatsu) that certifies the quality of the digitized images, particularly their spatial and RGB color resolution. Image registration and segmentation for 3D reconstruction were automated with the aid of Ratoc TRI-SRF2 software to avoid operator-dependent subjectivity. Visualization and quantification of 3D constructs facilitated validation of the continuity or segregation of carcinoma cells in the circumscribing stroma and assessment of the frequency of spatial contact between cancer cells and blood/lymph vessels. These 3D parameters are of great value for the prediction and diagnosis of cancer aggressiveness and malignancy in the coherent microenvironment. The results of this study demonstrate the advantages and suggest the wide range of potential applications of histology-based 3D reconstruction for evaluating biological tissue architecture and the interaction of multiple constituents.

View full abstract

The signal obtained in the phase-scrambling Fourier transform (PSFT) imaging technique can be transformed to the signal described by the Fresnel transform of the objects, in which the amplitude of the PSFT presents some kind of blurred image of the objects. Therefore, the signal can be considered to exist in the object domain as well as the Fourier domain of the object. This notable feature makes it possible to assign weights to the reconstructed images by applying a weighting function to the PSFT signal after data acquisition, and as a result, pseudo-parallel image reconstruction using these aliased image data with different weights on the images is feasible. In this study, the improvements in image quality with such pseudo-parallel imaging were examined and demonstrated. The weighting function of the PSFT signal that provides a given weight on the image is estimated using the obtained image data and is iteratively updated after SENSE-based image reconstruction. Simulation studies showed that reconstruction errors were dramatically reduced and that the spatial resolution was also improved in almost all image spaces. The proposed method was applied to signals synthesized from MR image data with phase variations to verify its effectiveness. It was found that the image quality was improved and that images almost entirely free of aliasing artifacts could be obtained.

View full abstract

We are investigating an integrated medical network system with an electronic letter of introduction function and a 3D image support function operating in the Internet environment. However, the problems with current C/S (client/server)-type systems are inadequate security countermeasures and insufficient transmission availability. In this report, we propose a medical information cooperation system architecture that employs a P2P (peer-to-peer)-type communication method rather than a C/S-type method, which helps to prevent a reduction in processing speed when large amounts of data (such as 3D images) are transferred. In addition, a virtual clinic was created and a feasibility study was conducted to evaluate the P2P-type system. The results showed that efficiency was improved by about 77% in real-time transmission, suggesting that this system may be suitable for practical application.

View full abstract

A new motion tracking method for local myocardial tissue using M-mode ultrasound signals is proposed. This method is applicable to the quantitative assessment of myocardial performance in clinics. M-mode echocardiograms are widely used in clinics for myocardial diagnosis. However, physicians are required to employ tedious and time-consuming diagnostic protocols. In the proposed method, the myocardial motion in M-mode echocardiograms is automatically tracked based on dynamic programming (DP) optimization. The method consists of two main stages. In the first stage, the correlation weighted mean algorithm estimates the velocity field in the myocardial wall. In the second stage, the instantaneous displacement of a set of tracking points is calculated from the estimated velocity field, and the DP tracking method tracks the motion of the tracking points. The experimental results obtained using clinical ultrasound data demonstrate that the proposed DP tracking method can provide higher performance in myocardial motion tracking as compared to other methods.

View full abstract

It is extremely important to develop an effective method for the registration of ultrasound images acquired at different stages of diagnosis and treatment. Reliable image registration is very useful for monitoring tumors over time and for evaluating therapeutic effectiveness, which are essential for determining the most appropriate clinical management of the patient. Intensive research has led to considerable progress in image registration for diagnostic modalities such as CT, MR, and PET. However, due to various factors peculiar to ultrasound examinations, there have been significantly fewer research studies on image registration focusing on ultrasound than on other diagnostic modalities. Diagnostic ultrasound systems that can acquire and record three-dimensional volume data continuously have recently been developed. In ultrasound diagnosis, it is extremely important to be able to reliably match the positions of a series of images obtained at various stages of the diagnostic and therapeutic processes. This has led to strong demand for a reliable image registration technique. The use of ultrasound-guided radiofrequency ablation (RFA) in the treatment of liver cancer has now become widespread. As an application developed for RFA, the present study introduces a volume similarity index that uses two or more small regions. The results of this study demonstrate that effective image registration can be achieved for ultrasound images.

View full abstract