Kazuki Matsuzaki

Development of a 3D Brain PET Scanner Using CdTe Semiconductor Detectors and Its First Clinical Application

Targeting improved spatial resolution, a three-dimensional positron-emission-tomography (PET) scanner employing CdTe semiconductor detectors and using depth-of-interaction (DOI) information was developed, and its physical performance was evaluated. This PET scanner is the first to use semiconductor detectors dedicated to the human brain and head-and-neck region. Imaging performance of the scanner used for 18F -fluorodeoxy glucose (FDG) scans of phantoms and human brains was evaluated. The gantry of the scanner has a 35.0-cm-diameter patient port, the trans-axial field of view (FOV) is 31.0 cm, and the axial FOV is 24.6 cm. The energy resolution averaged over all detector channels and timing resolution were 4.1% and 6.8 ns (each in FWHM), respectively. Spatial resolution measured at the center of FOV was 2.3-mm FWHM-which is one of the best resolutions achieved by human PET scanners. Noise-equivalent count ratio (NEC2R) has a maximum in the energy window of 390 to 540 keV and is 36 kcps/Bq/cm3 at 3.7 kBq/cm3 . The sensitivity of the system according to NEMA 1994 was 25.9 cps/Bq/cm3. Scatter fraction of the scanner is 37% for the energy window of 390 to 540 keV and 23% for 450 to 540 keV. Images of a hot-rod phantom and images of brain glucose metabolism show that the structural accuracy of the images obtained with the semiconductor PET scanner is higher than that possible with a conventional Bismuth Germanium Oxide (BGO) PET scanner. In addition, the developed scanner permits better delineation of the head-and-neck cancer. These results show that the semiconductor PET scanner will play a major role in the upcoming era of personalized medicine.

Development of a Prototype 3D PET Scanner Using Semiconductor Detectors and Depth of Interaction Information

A prototype brain positron emission tomography (PET) scanner using semiconductor detectors and depth of inleraclion (DOI) information has been developed to achieve high spatial resolulion and reduced scalier fraction. At the first step of the development, we created a two-dimensional prololype PET scanner composed of a single-slice full-ring detector unit to confirm the feasibility of the basic technologies that are necessary lo realize a semiconductor PET scanner. Through phantom and small-animal studies, the feasibility of the semiconductor PET was confirmed and the results showed that the semiconductor PET could produce quantitative imaging with high spatial resolution. Based on these achievements, a prototype brain PET scanner was developed to demonstrate the high spatial resolution and quantitative imaging capability required in human imaging.

Evaluation of delineation of image details in semiconductor PET utilizing the normalized mutual information technique.

ObjectivePET using semiconductor detectors provides high-quality images of the human brain because of its high spatial resolution. To quantitatively evaluate the delineation of image details in clinical PET images, we used normalized mutual information (NMI) to quantify the similarity with images obtained through MRI. NMI is used to evaluate image quality by determining similarity with a reference image. The aim of this study was to evaluate quantitatively the delineation of image details provided by semiconductor PET. Materials and methodsTo quantitatively evaluate anatomical delineation in clinical PET images, MRI scans of patients were used as T1-weighted images. [18F]-fluorodeoxyglucose (18F-FDG) PET brain images were obtained from six patients using (a) a Hitachi semiconductor PET scanner and (b) a ECAT HR+ scintillator PET scanner. The NMI calculated from the semiconductor PET and MRI was denoted by NMIsemic, whereas the NMI calculated from conventional scintillator PET and MRI was denoted by NMIconve. The higher the value of NMI, the greater the similarity to MRI. ResultsNMIsemic ranged from 1.22 to 1.29, whereas NMIconve ranged from 1.13 to 1.18 (P<0.05). Furthermore, all the NMI values of the semiconductor PET were higher than those of the conventional scintillator PET. ConclusionUtilizing NMI, we quantitatively evaluated the delineation of image details in clinical PET images. The results reveal that semiconductor PET has superior anatomical delineation and physical performance compared with conventional scintillator PET. This improved delineation of image details makes semiconductor PET promising for clinical applications.

Evaluation of medical image registration by using 3d SIFT and phase-only correlation

An effective method for quantitatively evaluating rigid and non-rigid image registration without any manual assessment is proposed. This evaluation method is based on feature point detection in reference images and corresponding point localization in registered floating images. For feature point detection, a 3D SIFT keypoint detector is applied to determine evaluation reference points in liver vessel regions of reference images. For corresponding point localization, a 3D phase-only correlation approach is applied to match reference points and their corresponding points. Distance between the reference points and the correspondences can be used to estimate image registration errors. With the proposed method, users can evaluate different registration algorithms using their own image data automatically.

False-positive reduction of liver tumor detection using ensemble learning method

We proposed a novel ensemble learning method which can be applied to false-positive reduction of liver tumor detection. In many cases of the liver tumor detection, training data has some issues due to characteristics of liver tumors, and the conventional ensemble learning methods such as Bagging and AdaBoost tend to degrade sensitivity. The proposed method generates various weak classifiers based on adaptive sampling in order to enhance an ensemble effect against such issues, and can achieve accuracy satisfying requirements of liver tumor detection. We applied the method to 48 CT images and evaluated the accuracy. Results showed that the proposed method succeeded in reducing false positives greatly (from 3.96 to 1.10/image) while maintaining the required sensitivity.

Metastatic liver tumor detection from 3D CT images using a level set algorithm with liver-edge term

We developed a metastatic liver tumor detection method using a level set algorithm with a liver-edge term. The level set algorithm is suitable for detection that requires an automated and accurate technique to reduce the time it takes to interpret the results. The conventional detection method, which is based on shape analysis using the Hessian matrix, tends to miss tumors on the edge of liver parenchyma because such tumors have a different shape than those in the center: on the edge they are blob-like and in the center they are step-like. The proposed method, which we call the liver-edge term, improves the accuracy of detection on the edge of liver parenchyma by recognizing step-like shapes on an intensity distribution. We applied the method to five 3-D CT images and evaluated the accuracy. Results showed that the proposed method had an average sensitivity of 92% compared to the 88% of the conventional method.

Size-adaptive hepatocellular carcinoma detection from 3D CT images based on the level set method

Automatic detection of hepatocellular carcinoma (HCC) from 3D CT images effectively reduces interpretation work. Several detection methods have been proposed. However, there still remains a tough problem of adaptation detection methods to a wide range of tumor sizes, especially to small nodules, since it is difficult to distinguish tumors from other structures, including noise. Although the level set method (LS) is a powerful tool for detecting objects with arbitrary topology, it is still poor at detecting small nodules due to low contrast. To detect small nodules, early phase images are useful since low contrast in the late phase causes miss-detection of some small nodules. Nevertheless, conventional methods using early phase images face two problems: one is failure to extract small nodules due to low contrast even in early phase images, and the other is false-positive (FP) detection of vessels adjacent to tumors. In this paper, a new robust detection method adapted to the wide range of tumor sizes has been proposed that uses only early phase images. To overcome these two problems, our method consists of two techniques. One is regularizing surface evolution used in LS by applying a new HCC filter that can enhance both small nodules and large tumors. The other is regularizing the surface evolution by applying a Hessian-matrix-based filter that can enhance the vessel structures. Experimental results showed that the proposed method improves sensitivity by over 15% and decreases FP by over 20%, demonstrating that the proposed method is useful for detecting HCC accurately.

Efficient registration method of medical images using GPU

Registration of medical images is an important task; however, automatic image-based registration is computationally expensive. Given this task, the authors propose an efficient rigid registration method, which is based on mutual information and uses a graphics processing unit (GPU). Mutual-information-based registration methods require joint-histogram computation. Although a GPU can provide high performance computing, a joint histogram has a large number of bins, and the computation of such a histogram is not suitable for a GPU (whose shared memory is limited). Taking advantage of the fact that one image (the reference image) is not transformed during the registration process, the proposed method computes a joint histogram by computing multiple onedimensional histograms and combining them. The method can therefore be efficiently implemented on a GPU even with limited shared memory. Experimental results for 256 × 256 × 256 image registration show that the method is about 140 times faster than a standard implementation on a CPU and 2.6 times faster than previous methods using GPUs.