Summary
Objectives:
Pleural thickenings as biomarker of exposure to asbestos may evolve into malignant
pleural mesothelioma. Foritsearly stage, pleurectomy with perioperative treatment
can reduce morbidity and mortality. The diagnosis is based on a visual investigation
of CT images, which is a time-consuming and subjective procedure. Our aim is to develop
an automatic image processing approach to detect and quantitatively assess pleural
thickenings.
Methods:
We first segment the lung areas, and identify the pleural contours. A convexity model
is then used together with a Hounsfield unit threshold to detect pleural thickenings.
The assessment of the detected pleural thickenings is based on a spline-based model
of the healthy pleura.
Results:
Tests were carried out on 14 data sets from three patients. In all cases, pleural
contours were reliably identified, and pleural thickenings detected. PC-based Computation
times were 85 min for a data set of 716 slices, 35 min for 401 slices, and 4 min for
75 slices, resulting in an average computation time of about 5.2 s per slice. Visualizations
of pleurae and detected thickeningswere provided.
Conclusion:
Results obtained so far indicate that our approach is able to assist physicians in
the tedious task of finding and quantifying pleural thickenings in CT data. In the
next step, our system will undergo an evaluation in a clinical test setting using
routine CT data to quantifyits performance.
Keywords
Malignant pleural mesothelioma - pleural thickening - thoracic spiral computed tomography
- computerassisted diagnosis - automatic image processing algorithms
