High-resolution chest CT using 1024-matrix reconstruction: phantom and clinical evaluation of image quality and post-processing capability

To evaluate the image quality and clinical utility of high-resolution computed tomography (CT) images reconstructed using post-processing with a 1024 × 1024 matrix. Four CT image types were reconstructed: type A (512 × 512 matrix, 1-mm slice thickness), type B (1024 × 1024 matrix, 1-mm slice thickness), type C (1024 × 1024 matrix, 0.5-mm slice thickness), and type D (retargeted zoomed images; 512 × 512 matrix, 1-mm slice thickness). Type D images served as the reference standard with a field of view (FOV) half that of the other image types. Phantom image quality was objectively assessed using the task transfer function (TTF), noise power spectrum (NPS), and attenuation profile analysis. In the clinical study, two radiologists independently evaluated image quality in 93 patients, including abnormal lung findings, normal structures, and subjective image noise. In the phantom study, post-processing 1024 × 1024 matrix reconstruction derived from the same sinogram as conventional 512 × 512 reconstruction improved spatial resolution while maintaining comparable noise characteristics at the same slice thickness. Clinically, image quality of normal structures and abnormal lesions was significantly higher in the order of type C > type B > type A (p < 0.0001). Subjective image noise followed the same order. Image quality and noise levels of type B images were comparable to those of type D images. Post-processing 1024 × 1024 matrix reconstruction improves spatial resolution and provides clinically useful high-resolution CT images comparable to retargeted zoomed reconstruction without the limitation of a reduced FOV.