Affiliations: Mediso Medical Imaging Systems, Budapest,
Hungary | Department of Image Processing and Computer Graphics,
University of Szeged, Szeged, Hungary | Clinical Center Vojvodina, University of Novi Sad,
Novi Sad, Serbia | Research Unit for Digital Information and Image
Processing, University Clinic for Radiology, Medical University Graz, Graz,
Austria
Note: [] Corresponding author: Laszlo Papp, Mediso Medical Imaging
Systems, Baross Str. 91–94, Budapest, H-1047 Hungary, Tel.: +3613993034;
E-mail: [email protected]
Abstract: Funnel chest (Pectus excavatum) is the most common deformity of the
anterior chest in children. Present paper describes a method to process and
classify CT slices representing funnel chest deformities. A manually chosen CT
slice was processed to detect the inner curvature of the chest for
characterization. Normalized data from the detected inner curvature was gained
and saved next to a manually-given deformity type for further classification
rule determinations. Based on the multiple correlations of the values gained
from the inner curvature, a hierarchical classification was performed on 199
patient data. Results have shown that the calculated values gained from the
inner curvature can accurately characterize the deformity type of the chest.
Since minimal user interaction was necessary to detect and characterize the
inner curvature, our method is considered to be an effective automated
procedure for funnel chest deformity classifications.
