A computer aided diagnostic system for radiotherapy planning.

Planning for radiation therapy intervention implies the definition of treatment volumes as well as a clear delimitation of normal tissue. This paper presents a Computer Aided Diagnostic system for the automatic CT image analysis. Two important problems are solved: the spinal cord segmentation and the detection of lung metastases. Some subordinate problems are also solved: the detection of spinal canal, lamina, lungs, and ribs, as well as the identification of thorax contour. The developed methodologies use a knowledge-driven image processing based on Anatomical Structures Maps and task-oriented architecture. Experiments were performed on CT images from La Chaux de Fonds Hospital (Switzerland). Evaluations were performed using a visual inspection of the contours projected on the CT image slices. The radiologist decided whether each of the contours obtained with our system was acceptable or not. The accuracy of the method was defined as the fraction of CT slices in which the particular contour was correctly located. In the case of spinal cord segmentation, the procedure was tested on 23 patients (1051 images), resulting in an accuracy of 91%. In the case of lung tumors detection, the method showed an accuracy of > 90%, with testing performed on 20 patients for a total of 988 images. The experiments performed show that the method is reliable, with possible future application in an oncology department.

[1]  D. T. Lee,et al.  Medial Axis Transformation of a Planar Shape , 1982, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Terry E. Weymouth,et al.  Multiple organ definition in CT using a Bayesian approach for 3D model fitting , 1995, Optics & Photonics.

[3]  A Fenster,et al.  An automated segmentation method for three-dimensional carotid ultrasound images. , 2001, Physics in medicine and biology.

[4]  Demetri Terzopoulos,et al.  Snakes: Active contour models , 2004, International Journal of Computer Vision.

[5]  Milan Sonka,et al.  Knowledge-based segmentation of intrathoracic airways from multidimensional high-resolution CT images , 1994, Medical Imaging.

[6]  M van Herk,et al.  Radiation field edge detection in portal images. , 1991, Physics in medicine and biology.

[7]  J W Sayre,et al.  Knowledge-based segmentation of thoracic computed tomography images for assessment of split lung function. , 2000, Medical physics.

[8]  Michael Egmont-Petersen,et al.  A knowledge-based approach to automatic detection of the spinal cord in CT images , 2002, IEEE Transactions on Medical Imaging.

[9]  S Shiffman,et al.  A decision aid for diagnosis of liver lesions on MRI. , 1993, Proceedings. Symposium on Computer Applications in Medical Care.

[10]  Sebastiano B. Serpico,et al.  A knowledge-based system for biomedical image processing and recognition , 1987 .

[11]  Hongyi Li,et al.  Object recognition in brain CT-scans: knowledge-based fusion of data from multiple feature extractors , 1995, IEEE Trans. Medical Imaging.

[12]  V. F Kumar,et al.  Image Interpretation Using Bayesian Networks , 1996, IEEE Trans. Pattern Anal. Mach. Intell..

[13]  Heinrich Niemann,et al.  Control and explanation in a signal understanding environment , 1993, Signal Process..

[14]  E. Rogers,et al.  VIA-RAD: a blackboard-based system for diagnostic radiology. Visual Interaction Assistant for Radiology , 1995, Artif. Intell. Medicine.

[15]  Neculai Archip,et al.  Lung metastasis detection and visualization on CT images: a knowledge-based method , 2002, Comput. Animat. Virtual Worlds.

[16]  M van Herk,et al.  Automatic three-dimensional inspection of patient setup in radiation therapy using portal images, simulator images, and computed tomography data. , 1996, Medical physics.