Segmentation of lung radiography a preterm infant with respiratory distress syndrome

Abstract

The aim of this work is to analyze the visualisation of respiratory distress in a preterm infant by segmenting the radiographs image of the lungs. In the study, the radiograph of a preterm infant at the 27th week of gestation was processed, who was diagnosed with respiratory distress syndrome grade IV according to Bomsell. Visualisation of respiratory distress of a preterm infant was preformed by segmentation of the radiographic image of the lungs within the MATLAB programmable environment. Before the segmentation procedure, the radiographic image was pre-processed (noise removal, edge enhancement). Extraction of the expanded part of the lung parenchyma through the analysis of the parameters of the confusion matrix and the Dice coefficient. The results show the visualization of the unaffected (expanded) lung region, as well as the automatic identification of the region. Performance evaluation parameter show satisfactory accuracy and are above 0.85. Analysis of the segmentation of radiographic images of a preterm infant with respiratory distress syndrome can have a significant contribution in the process of diagnosis of respiratory distress syndrome as well as in the evaluation of patients after the therapy.