Hidefumi Watanabe

A conditional statistical shape model with integrated error estimation of the conditions; Application to liver segmentation in non-contrast CT images

This paper presents a novel conditional statistical shape model in which the condition can be relaxed instead of being treated as a hard constraint. The major contribution of this paper is the integration of an error model that estimates the reliability of the observed conditional features and subsequently relaxes the conditional statistical shape model accordingly. A three-step pipeline consisting of (1) conditional feature extraction from a maximum a posteriori estimation, (2) shape prior estimation through the novel level set based conditional statistical shape model with integrated error model and (3) subsequent graph cuts segmentation based on the estimated shape prior is applied to automatic liver segmentation from non-contrast abdominal CT volumes. Comparison with three other state of the art methods shows the superior performance of the proposed algorithm.

Radiographic imaging below a volcanic crater floor with cosmic-ray muons

We present a novel application of cosmic-ray muon radiography to image the shallow density structure beneath Asama Volcano, Japan. We use a single detector (emulsion cloud chamber) set up in an underground vault at an elevation of 2250 m on the eastern flank of Asama, 310 m below the summit of the edifice and 1 km away from the crater. The results point to two high-density anomalies located between the original pre-2004 eruption crater floor and post-2004 eruption crater profile. A third low-density anomaly is imaged immediately below the pre-2004 eruption crater floor. The spatial extent of each density anomaly is about 100 to 200 m. To know if this method, applied to other volcanoes, would produce contrasting results, we performed the measurement in 1944 Usu lava dome. We confirmed a bulbous shape measuring approximately 300 m in diameter and narrowing downwards. The diameter of the uppermost part of the conduit is estimated at 100 ± 15 m at an elevation of 260 m a.s.l. and 50 ± 15 m at an elevation of 217 m a.s.l., demonstrating a resolution that is significantly better than that typically achieved with seismic tomography based on picks of first arrival times from earthquakes or artificial sources.

Evidence for subsurface magma bodies beneath Izu-Oshima volcano inferred from a seismic scattering analysis and possible interpretation of the magma plumbing system of the 1986 eruptive activity

Abstract Observed seismic waveforms were analyzed to find the distribution of seismic scatterers beneath Izu-Oshima volcano in an island located about 100 km south of Tokyo, Japan, using seismic waveforms generated by earthquakes, whose epicentral distances are from several tens to about a thousand kilometers. The method employed in the study was developed based on a prestack migration of shot gathers and is composed of the following procedure: (1) static correction, (2) the estimation of waveforms incident into a volume of interest, i.e., subsurface structure of the volcano, and (3) an inverse procedure using the developed migration scheme. The location of scatterers and the order of scattering became clear with the present analysis. The petrological implications and results from geoelectromagnetic studies by other researchers agree with the proposed model in which the existence of main and submagma reservoirs beneath the volcano were indicated. These results are interpreted in terms of a model on the magma migration path during the 1986 eruptive activity of the volcano. Seismic scattering analysis played a significant and indispensable role in this study to give a direct view of a possible underground plumbing system, and is considered a promising method for revealing the unknown structure of volcanoes in the future.

A method for simultaneous velocity and density inversion and its application to exploration of subsurface structure beneath Izu-Oshima volcano, Japan

We have developed a method for three-dimensional simultaneous velocity and density inversion using traveltimes of local earthquakes and gravity data. The purpose of this method is to constrain the velocity inversion and increase the spatial resolution of shallow velocity structures by introducing additional gravity data. The gravity data contributes to the P- and S-wave velocity models by imposing constraints between seismic velocities and density. The constraint curve is constructed so as to fit the data for porous rock samples, and deviations from the curve are taken into account in the inversion. The constraint is imposed at only the first layer, because density structure is well resolved at shallower parts and it is difficult to determine uniquely at greater depths. Synthetic inversion tests indicate that gravity data can improve the resolution of the velocity models for this layer. The method is applied to investigate the subsurface structure of Izu-Oshima volcano, Japan and velocity structures with high spatial resolution are obtained. The additional gravity data contribute primarily to improvement of the S-wave velocity model. At 0.25 km depth, a high velocity anomaly due to caldera-filling lava flows is observed. At 1.25 and 2.5 km depths, high velocity intrusive bodies are detected. A NW-SE trending high velocity belt at 1.25 km depth is interpreted as being caused by repeated intrusion of dikes.

Drain‐back process of basaltic magma in the summit conduit detected by microgravity observation at Izu‐Oshima Volcano, Japan

Basalt magma sometimes moves through volcanic conduits causing only minor deformation. In that case, we may detect magma movements directly by microgravity observations. After the 1986 eruption of Izu-Oshima volcano, we observed anomalous gravity variations localized at the summit. Based on a vertical cylindrical conduit model, we estimate the time variations of the head of magma in the summit conduit and clarify the magma drain-back process after the 1986 eruption.

Ground deformation associated with volcanic tremor at Izu-Oshima Volcano

Izu-Oshima volcano had summit and fissure eruptions in November, 1986 after 12 years of dormancy, and three small eruptions followed these events within one year. Episodic and continuous volcanic tremors were observed for the period containing these eruptions. It is a remarkable discovery that the episodic volcanic tremor was accompanied by a small but sharp ground deformation, the polarity of which was variable. The distribution of tilt vectors reveals that the source of ground deformation was always located beneath the northwestern flank of the volcano, where a magma reservoir was predicted by other studies. On the other hand, the seismologically detected tremor source was determined to be at a shallow depth below the central pit crater, a few kilometers away from the predicted magma reservoir. It is thus inferred that the tremor source near the crater generated pressure increases or decreases that were simultaneously transmitted through the vent to the magma reservoir and lead to its expansion or contraction.

Relaxed conditional statistical shape models and their application to non-contrast liver segmentation

This paper proposes a novel conditional statistical shape model (SSM) that allows a relaxed conditional term. The method is based on the selection formula and allows a seamless transition between the non-conditional SSM and the conventional conditional SSM. Unlike a conventional conditional SSM, the relaxed conditional SSM can take the reliability of the condition into account. Organ shapes estimated by the proposed SSM were used as shape priors for Graph Cut based segmentation. Results for liver shape estimation and subsequent liver segmentation show the benefit of the proposed model over conventional conditional SSMs.

Square deformed map with simultaneous expression of close and distant view

We propose a new type of deformed map where both the close and the distant view are expressible.

A Statistical Shape Model for Multiple Organs Based on Synthesized-Based Learning

This paper presents a statistical shape model for multiple abdominal organs using synthesized-based learning to compensate the lack of a large manually labeled training data set. Experiments on 23 non-contrast CT volumes showed that a model trained on both true and synthesized data, outperforms conventional shape models, in terms of generalization, specificity and overlap of neighboring organs.

Relaxed Conditional Hierarchical Statistical Shape Model of Multiple Organs

This paper proposes a relaxed conditional hierarchical statistical shape model (SSM) of multiple organs. After extracting shape and pose parameters from the training label dataset of multiple organs, the shape model and the pose model of each organ are constructed by principal component analysis (PCA). Subsequently, the principal scores of all organs are concatenated into a vector, and the vectors computed from the training dataset are forwarded to the PCA-based statistical modeling of the multiple organs under conditions of their neighboring organs. A relaxation scheme is introduced, to take into account errors in the conditions. This study focuses on modeling of a spleen and a gallbladder given a liver as a conditional organ. The performance of the model was evaluated with the measures of generalization and specificity, which were computed by three-fold cross-validation using labels of 27 abdominal CT volumes with the size of 170 × 170 × 110 voxels and a resolution of 1.8809 mm/voxel. Compared with a hierarchical SSM without conditions, generalization and specificity were improved from 0.488 to 0.506 and from 0.319 to 0.328 on average, respectively. In addition, the proposed relaxed conditional hierarchical SSM outperformed a hierarchical SSM with hard conditions. The performance indices were improved by 0.040 and 0.010 for generalization and specificity, respectively.