The University of Osaka · Medicine
Professor Shoji Kido's research lab specializes in medical image analysis and computational diagnostics, focusing on the application of advanced image processing and machine learning techniques—particularly convolutional neural networks (CNN) and region-based CNN (R-CNN)—to improve the detection and diagnosis of lung abnormalities such as nodules and diffuse lung diseases. The lab also investigates fractal analysis of high-resolution CT (HRCT) images to characterize tumor textures and interfaces, enabling differentiation between benign and malignant pulmonary nodules, including subtypes of bronchoalveolar carcinoma. Additionally, the lab contributes to virological diagnostics through molecular techniques like PCR for detecting varicella-zoster virus (VZV), linking viral load to clinical outcomes such as ocular damage in herpes zoster ophthalmicus. The integration of radiological, pathological, and molecular data underscores the lab’s multidisciplinary approach to precision diagnosis in pulmonary and infectious diseases.
Figures are computed from collected data and may differ slightly.
Image-based computer-aided diagnosis (CADx) algorithm by use of convolutional neural network (CNN) does not necessarily require an image-feature extractor. Therefore, image-based CADx is powerful compared with feature-based CADx that requires the image-feature extractor for differential diagnosis of lung abnormalities such as lung nodules and diffuse lung diseases. We have also developed an image-based computer-aided detection (CADe) algorithm by use of regions with CNN features (R-CNN) for dete
A polymerase chain reaction system for the detection of varicella-zoster virus was established. Of 25 nucleotides, 4 oligonucleotide pairs (regions of thymidine kinase, thymidylate synthetase, glycoprotein I, and immediate early gene) were synthesized. The first three oligonucleotide pairs could be used as primers on the basis of specific DNA amplification. Varicella-zoster virus DNA was amplified by this polymerase chain reaction system in 20 of 20 vesicle samples, 5 of 6 crusts, and 12 of 13 t
Fractal dimensions reflect the characteristics of the lung-nodule interfaces of small peripheral pulmonary nodules. The FD(2D)s revealed the irregularities of the contours. On the other hand, FD(3D)s revealed the complexities of the heterogeneous textures. With use of FD(2D) and FD(3D), it may be possible to distinguish bronchogenic carcinomas from benign pulmonary nodules. Moreover, FD(3D) may make it possible to distinguish between adenocarcinomas and squamous cell carcinomas.
VZV viral load in the aqueous humor correlated significantly with damage to the iris (iris atrophy and pupil distortion) in patients with HZO and ZSH.
The textures of BACs that reveal ground-glass opacities are more complicated than those of nonBACs. The FDs can differentiate between small localized BACs, which have a good prognosis, and nonBACs, which have a poor prognosis. Fractal analysis is promising for characterization of small peripheral pulmonary bronchogenic carcinomas based on radiographic features of HRCT images.
The iterative noise-reduction algorithm is superior to conventional methods in detection of pulmonary nodules.
In computer-aided diagnosis systems for lung cancer, segmentation of lung nodules is important for analyzing image features of lung nodules on computed tomography (CT) images and distinguishing malignant nodules from benign ones. However, it is difficult to accurately and robustly segment lung nodules attached to the chest wall or with ground-glass opacities using conventional image processing methods. Therefore, this study aimed to develop a method for robust and accurate three-dimensional (3D)
To evaluate the reliability of storage phosphor radiography (SR) in diagnosis of subtle interstitial lung abnormalities, the differences among radiologists in interpreting conventional screen-film radiographs and full-size and minified SR images obtained in 80 patients were studied. Forty patients had subtle interstitial lung abnormalities and 40 had no lung abnormalities. Seven chest radiologists and seven residents evaluated the images by using a five-point presence of abnormality scale. Resul
A computerized method for analyzing interstitial lung abnormalities seen on chest radiographs was investigated. The method includes two main steps: (a) extraction of linear opacities on chest radiographs and (b) calculation of the fractal dimension. Extraction of linear opacities uses the processes of four-directional Laplacian-Gaussian filtering, binarization, and linear opacity judgment. The fractal dimensions in the processed images are then calculated by using the box-counting algorithm. The
The human visual system can interpret two-dimensional (2-D) line drawings like the Necker cube as three-dimensional (3-D) wire frames. On this human ability Thomas Marill presented two important papers. First one proposed the 3-D interpretation model based on the principle to minimize the standard deviation of the angles between line segments in 3-D wire frame (MSDA), and reported the results of simulation experiments. Second one proposed the principle to minimize the description length on the i
The single-exposure dual-energy subtraction method is superior to the conventional subtraction method in the detection of pulmonary nodules.
These results suggest that a 10:1 data compression ratio does not influence the detection of subtle interstitial lung abnormalities. However, information that is lost with a 20:1 data compression ratio might be essential for interpretation by experienced chest radiologists.
In this study, we present a self-supervised learning (SSL)-based model that enables anatomical structure-based unsupervised anomaly detection (UAD). The model employs an anatomy-aware pasting (AnatPaste) augmentation tool that uses a threshold-based lung segmentation pretext task to create anomalies in normal chest radiographs used for model pretraining. These anomalies are similar to real anomalies and help the model recognize them. We evaluate our model using three open-source chest radiograph
Open papers in the app to read, cite, and organize with AI.
