Improvement of the image quality of MSCT of the pelvis with a raw data-based, multidimensional filter

Purpose: To evaluate the potential of raw data-based multidimensional adaptive filtering (MAF) by determining its effects on image noise and image quality in multi-slice spiral CT (MSCT) of the pelvis. Materials and Methods: Fifty patients with rectal and bladder cancer were examined with MSCT using a high-resolution protocol. Reconstructions were performed with dedicated reconstruction software and a standard algorithm, both without and with MAF, with four different modification fractions selected from 5% to 20%. The noise was measured at six anatomic sites of the pelvis. Image quality and image noise were rated on a 5-point-scale. Results: Modification fractions of 15% (15% MAF) and 20% (20% MAF) significantly reduced the noise level at all measurement points in comparison with lower modification fractions (p < 0.05). Overall quality of the reconstructed images was rated better with 15% MAF and 20% MAF than with other modification fractions (p < 0.05). No further improvement of the image quality was observed by changing from 15% MAF to 20% MAF (p > 0.05). 15% MAF achieved a mean noise reduction of 33%. All examinations showed an improved visualization of the rectal wall and perirectal lymph nodes. Conclusions: MAF improves the image quality by reducing the noise level and by removing noise structures. This technique offers new perspectives to radiation dose reduction in CT.