Search

Oversight in medical images is a crucial problem, and timely reporting of medical images is desired. Therefore, an all-purpose anomaly detection method that can detect virtually all types of lesions/diseases in a given image is strongly desired. However, few commercially available and versatile anomaly detection methods for medical images have been provided so far. Recently, anomaly detection methods built upon deep learning methods have been rapidly growing in popularity, and these methods seem to provide reasonable solutions to the problem. However, the workload to label the images necessary for training in deep learning remains heavy. In this study, we present an anomaly detection method based on two trained flow-based generative models. With this method, the posterior probability can be computed as a normality metric for any given image. The training of the generative models requires two sets of images: a set containing only normal images and another set containing both normal and abnormal images without any labels. In the latter set, each sample does not have to be labeled as normal or abnormal; therefore, any mixture of images (e.g., all cases in a hospital) can be used as the dataset without cumbersome manual labeling. The method was validated with two types of medical images: chest X-ray radiographs (CXRs) and brain computed tomographies (BCTs). The areas under the receiver operating characteristic curves for logarithm posterior probabilities of CXRs (0.868 for pneumonia-like opacities) and BCTs (0.904 for infarction) were comparable to those in previous studies with other anomaly detection methods. This result showed the versatility of our method.

We present the new basis functions to investigate the 't Hooft equation, the lowest order mesonic Light-Front Tamm-Dancoff equation for $\rm SU(N_C)$ gauge theories. We find the wave function can be well approximated by new basis functions and obtain an analytic formula for the mass of the lightest bound state. Its value is consistent with the precedent results., Comment: 6 pages, 2 figures. Prepared for Fifth Workshop on QCD (QCD2000), Villefranche-Sur-Mer, France, 3-7 January 2000

We investigate the behavior of small perturbations around the Kaluza-Klein monopole in the five dimensional space-time. We find that the even parity gravitational wave does not propagate in the five dimensional space-time with Kaluza-Klein monopole provided that the gravitational wave is constant in the fifth direction. We conclude that a gravitational wave and a U(1) magnetic monopole do not coexist in five dimensional Kaluza-Klein spacetime., Comment: 10 pages, LaTeX. To appear in Modern Physics Letters A

Two dimensional QCD is quantized on the light front coordinate. We solve the Einstein-Schr\"odinger equation by the use of Tamm-Dancoff truncation and find that the simplest wavefunction produces the $M/g$ versus $m/g$ relation in agreement with other calculations, where $M$ and $m$ are the masses of the ground state and quarks, respectively., Comment: 16 pages, LaTeX, 3 PostScript formatted figures