Registration of Mri and Ct Scan Images With the Use of Mutual Information



Mehridehnavi A¹ ; Hatefwahied R² Show Affiliations
Source: Journal of Isfahan Medical School Published:2012

Abstract

Background: In medicine decision making in respect to prognosis diagnosis, and treatment is a major challenge. Therefore, in this respect supportive and auxiliary modalities are critical and desirable. Cancerous tumor, edema and lesion detection in deep layers of body tissues are impossible without depth imaging modalities such as CT scan and MRI (magnetic resonance image). Since these modalities have limitations, it is recommended to use them in a supportive manner. One way to compare these modality images is bilateral adaptation by the use of maximum mutual information (MI). Methods: In this research the two modalities of medical images of CT scan and MRI have been studied in mutual information consideration. One of the images is assumed as a reference and fixed in parameters and the other is transformed as float. To ease the comparison the pixel intensity in both modalities are normalized into 0-255. With the application of the Kullback-Leibler rule of probability, the researcher tried to maximize mutual information. Float images are transformed by their coordinate in x and y direction and rotation angle th. The software should change the float image parameters and try to find maximum mutual information. Findings: In the procedure of the same modality image registration, CT scan with CT scan or MRI with MRI, for six available images there was 180 trials. For similar modalities of variation of ± 11 mm and ± 30 degrees, mean value displacement for maximum mutual information was between 8.2 to 9.1 mm with the standard deviation of ± 1 to ± 2.6 while for different modalities it was 4.4 to 11.3 mm with standard deviation of ±4 to ± 6.7. The difference between these values shows a mismatch of CT scan modalities with MRI modalities. Conclusion: This mismatch could be due to modality constitution. In MRI imaging intensity and contrast are based on H1 content T1 and T2 relaxation times, while in CT scan imaging intensity and contrast are determined with X-ray absorbance of tissues based on HVL absorbance of different tissues. Therefore, registration of CT scan modality with MRI modality on the basis of intensity and contrast is not satisfactory. We suggest the use of other configurations such as morphological characteristics of these modalities.