Development of a kinetic analysis technique for PACS management and a screening examination in dynamic radiography.

Abstract

The purpose of this study was to develop a method of kinetic analysis for picture archiving and communication system (PACS) management and computer-aided diagnostic application in dynamic chest radiography. The main analytical technique used in this study was a new algorithm that converts dynamic radiographs into a color-static image. The algorithm is a visualization technique for kinetic information that uses the intensity-density transformation and the direction classification in optical flow. The image made by the new algorithm was defined as a "kinetic map," and, by analysis using the kinetic map, a patient collation system and nodule detection system were constructed. By analysis that used an artificial neural network of certain feature vectors as kinetic map similarity, the collation system obtained good identification performance. Temporal subtraction processing between a current-status map with simulated nodule and previous-status map detected the region of abnormality as the simulated nodule. It is expected that our method of analysis will be useful as a screening examination for risk management and computer-aided diagnostic application in dynamic chest radiography.