Takahiro Iwamoto

Moment-based texture segmentation of luminal contour in intravascular ultrasound images

PurposeA system for luminal contour segmentation in intravascular ultrasound images is proposed.MethodsMoment-based texture features are used for clustering of the pixels in the input image. After the clustering, morphological smoothing and a boundary detection process are applied and the final image is obtained.ResultsThe proposed method was applied to 15 images from different patients, and a correlation coefficient of 0.86 was obtained between the areas of lumen automatically and manually defined.ConclusionMoment-based texture features together with the radial feature are powerful tools for identification of the lumen region in intravascular ultrasound images. Morphological filtering was useful for improving the segmentation results.

High frequency ultrasound imaging of surface and subsurface structures of fingerprints

High frequency ultrasound is suitable for non-invasive evaluation of skin because it can obtain both morphological and biomechanical information. A specially developed acoustic microscope system with the central frequency of 100 MHz was developed. The system was capable of (1) conventional C-mode acoustic microscope imaging of thinly sliced tissue, (2) ultrasound impedance imaging of the surface of in vivo thick tissue and (3) 3D ultrasound imaging of inside of the in vivo tissue. In the present study, ultrasound impedance imaging and 3D ultrasound imaging of in vivo fingerprints were obtained. The impedance image showed pores of the sweat glands in the surface of fingerprint and 3D ultrasound imaging showed glands of the rear surface of fingerprint. Both findings were not visualized by normal optical imaging, thus the system can be applied to pathological diagnosis of skin lesions and assessment of aging of the skin in cosmetic point of view.

High frequency ultrasound characterization of artificial skin

Regenerated skin with 3-dimensional structure is desired for the treatment of large burn and for the plastic surgery. High frequency ultrasound is suitable for non-destructive testing of the skin model because it provides information on morphology and mechanical properties. In this paper, spectral parameters of ultrasound radio-frequency signal from a specially developed high-frequency ultrasound imaging system were evaluated for tissue characterization of artificial skin. Results suggest that spectral parameters are useful for classification of epidermis and dermis in the artificial skin model. The system is also a useful tool for the noninvasive and nondestructive evaluation of skin.

An adaptive fuzzy segmentation of intravascular ultrasound images

A system for automatic definition of region of interest (ROI) in intravascular ultrasound images is proposed. This system performs this task in two phases: first a preprocessing based on morphological smoothing and fuzzy enhancement is done and then a fuzzy clustering algorithm is applied to the preprocessed image. The preliminary results showed that this technique is a feasible method for finding calcified regions that can be explored by classifiers.

Coronary plaque classification through intravascular ultrasound radiofrequency data analysis using self-organizing map

Intravascular ultrasound (IVUS) is an important clinical tool in the assessment of atherosclerotic plaque in coronary artery diseases. Using IVUS, we can obtain high resolution echo image of cross-sections of the coronary artery. However, it is difficult to accurately classify plaques by using the echogram only. We propose a method of IVUS Radiofrequency (RF) signal classification using self-organizing map (SOM). Characteristic ROIs (region of interest) of the IVUS echogram of patients with coronary lesions were selected by an expert medical doctor, and the SOM learned from these ROIs. The SOM could classify the RF signals with accuracies of 95.9% for fibrous plaque, 99.5% for blood, 96.2% for calcified plaque and 16.3% for media regions. This result suggests that the proposed technique is useful for automatic characterization of plaque in coronary artery.

P1D-3 A System for Tissue Characterization and Quantification of Calcium Regions in Intravascular Ultrasound

This paper presents a combination of signal processing and image processing techniques for automatic segmentation and characterization of intravascular ultrasound images. This system is comprised of two modules, the tissue characterization module and the calcium quantification module. The tissue characterization module is based on classification of RF signal performed by a self-organizing map (SOM) previously trained. The calcium quantification module, using the image generated by the envelop of the RF signal, performs an adaptive thresholding based on the Otsus method. The thresholding process is followed by the analysis of the acoustic shadow regime of the input image which permits to distinguish calcification regions from other small bright regions that, usually, still remain in the image after the thresholding processing

Three-dimensional ultrasound imaging of regenerated skin with high frequency ultrasound

Regenerated skin with complete organ structure is desired for the treatment of large burn and for the plastic surgery. High frequency ultrasound is suitable for nondestructive testing of such skin models because it can obtain both morphological and biomechanical information in non-invasive manner. A specially developed acoustic microscope system with the central frequency of 100 MHz was developed for evaluation of regenerated skin. The system was capable of (1) conventional C-mode acoustic microscope imaging of thinly sliced tissue, (2) ultrasound impedance in vivo imaging of the surface of tissue and (3) 3D ultrasound imaging of in vivo tissue. Sound speed in C- mode showed higher values at the area of dense fibroblasts in collagen sponge. The system can be used as the in vitro nondestructive evaluation tool in the process of regenerating skin and as the in vivo diagnostic system after implantation of the skin.

Radio Frequency Signal Analysis for Tissue Characterization of Coronary Artery: In Vivo Intravascular Ultrasound Study

Intravascular ultrasound (IVUS) is an important clinical tool that provides high resolution cross-sectional image of coronary artery. However, it is difficult to accurately classify plaque composition by conventional IVUS images. In the present study, we apply self-organizing map (SOM) of radiofrequency (RF) signal spectra for automatic plaque classification in IVUS diagnosis. IVUS data were acquired with a commercially available IVUS system with the central frequency of 40 MHz. We used double SOM classifier. The 1st classifier is supervised-SOM, learned four structures (blood, catheter, shadow, and outer lumen) based on spectral parameters. The 2nd classifier is unsupervised-SOM, used for classifying remained data, which were not classified the 1st classifier. We defined categories on the 2nd SOM by using K-means clustering method. Finally, color codes were assigned to the plaque component values, and the tissue color coded maps were reconstructed. Results suggest that the proposed technique is useful for automatic characterization of plaque components in IVUS image.

Recent progress of acoustic microscopy for medicine and biology

Dramatic shortening of the calculation time provided by the recent progress of the computer technology has made biomedical researchers possible to assess nonlinear acoustic phenomena in soft materials which had been assumed as acting linearly. Besides, the spread of Windows‐based personal computer and peripheral devices enabled easier and cheaper configuration of the whole acoustic microscope system such as pulse generation, analogue/digital conversion, mechanical scanning and image processing. According to these progresses, conventional acoustic microscopy with only C‐mode imaging has widened its data acquisition mode to B‐mode, C‐mode, surface acoustic impedance mode, and three‐dimensional (3D) imaging. The base of our acoustic microscope system was consisted of (1) PVDF transducer with the central frequency of 100 MHz, (2) ultrasonic pulser made of high speed semiconductor switching, (3) mechanical scanner using two linear servo motors, (4) high speed PCI card digitizer with the sampling frequency of 2...