Masaru Tsudagawa

Detection of Ground Glass Opacities in Lung CT Images Using Gabor Filters and Neural Networks

This paper aims at developing a CAD system used for the detection of Ground Glass Opacity (GGO) nodules in chest CT images. In our scheme, we apply Gabor filter on the CT image in order to enhance the detection process. After this we perform some morphological operations including threshold process and labeling to extract the objects having high intensity values. Then, some feature analysis is used to examine these objects to decide which of them are likely to be cancer candidates. Following the feature analysis, a template matching between the potential cancer candidates and some Gaussian reference models is performed to determine the similarity between them. The algorithm was applied on 715 slices containing 25 GGO nodules and achieved detection sensitivity of 92% with False Positive (FP) rate of 0.76 FP/slice. Finally, we used an Artificial Neural Network (ANN) to reduce the number of FP findings. After using ANN, we were able to reduce the FP rate to 0.25 FP/slice but at the expense of decreasing the detection sensitivity to 84%

A New CAD System for Detecting Localized Ground Glass Opacity Nodules in Lung CT images Using Cross and Coronary Section Images

This study is intended to propose a novel Computer Aided Diagnosis (CAD) scheme for automatic detection of localized Ground Glass Opacity (GGO) nodules in chest Computed Tomography (CT) images. The main idea of our method is to confirm the existence of the GGO candidates in the cross sectional CT images by examining the coronary sectional CT images based on the fact that nodular candidates tend to appear circular in both sections. Our detection scheme begins with a preprocessing stage to the cross sectional CT image to extract the lung region and enhance the intensity values of the nodular regions. We then filter the resulting image with Gabor filter followed by thresholding and labeling to assign the suspected regions and match them with some predefined reference Gaussian templates. Thereafter, some characteristic morphological and gray level features are used to discriminate the potential GGO candidates. Finally, we confirm the existence of the GGO nodules by examining the coronary sectional CT images. Receiver Operating Characteristic (ROC) curve for our CAD scheme was constructed to evaluate its performance. The proposed scheme had an area under the ROC curve of 0.94, which proves its potential effectiveness in GGO nodule detection

An estimation method of the central frequency for optical velocity sensors

The authors propose a signal processing method for measuring the velocity of moving objects (or moving scenes) using 2D spatial filters. The central (peak) frequency of the bell-shaped spectrum that contains the velocity information on the moving objects is estimated by fitting the sensor measurement data to AR(2) (second-order autoregressive) models so that the estimated coefficients directly yield the estimated peak of the spectrum. A computer simulation experiment was carried out and showed that the method yields a good estimate of the peak frequency corresponding to the velocity of the moving object.<<ETX>>

Image velocity measurement by using spatial filter method

This paper presents an approach of image velocity measurement by using both of spatial filter and correlation method. The goal is to achieve the accurate measurement of image velocity, which in fact presented difficulties in the existence of background. We show high accuracy by using the approached method as result.

Automatic liver segmentation method featuring a novel filter for multiphase multidetector-row helical computed tomography.

Purpose: To introduce an automatic liver segmentation method that includes a novel filter for multiphase multidetector-row helical computed tomography. Materials and Methods: We acquired 3-phase multidetector-row computed tomographic scans that included unenhanced, arterial, and portal phases. The liver was segmented using our novel adaptive linear prediction filter designed to reduce the difference between filter input and output values in the liver region and to increase these values outside the liver region. Results: The segmentation algorithm produced a mean dice similarity coefficient (DSC) value of 91.4%. Conclusion: The application of our adaptive linear prediction filter was effective in automatically extracting liver regions.

An optimum cut-off frequency of temporal filter and comparison of SN ratio obtained by two types of aperture in optical seeker

This paper is described on optical aperture and electrical filter for the optical seeker. Two kinds of optical aperture are compared, that is, single and polarized double slits. In addition, the decision procedure of low-cut-off frequency of used electricity band-pass filter is shown. As a result, the double slit excels the about 1.25 to 2.33 times single slit on the SN ratio.

A new 3-D vision system with spatial filters

A novel 3-D vision system using a spatial filter for unmanned carriers such as a mobile robot or an automated vehicle is proposed. The authors present the fundamental principle of the system, its signal processing method, and experimental results. To confirm the applicability of the system, simulation experiments have been carried out showing that obstacles placed at 40 cm and 70 cm can be discriminated. Simultaneous measurements of their ranges, azimuth angles, and sizes were carried out successfully.<<ETX>>