Optimized image reconstruction for detection of deep venous thrombosis at multidetector-row CT venography

The aims of this study were to optimize image quality for indirect CT venography (sequential versus spiral), and to evaluate different image reconstruction parameters for patients with suspected deep venous thrombosis (DVT). Fifty-one patients (26/25 with/without DVT) were prospectively evaluated for pulmonary embolism (PE) with standard multidetector-row computed tomography (MDCT) protocols. Retrospective image reconstruction was done with different slice thicknesses and reconstruction increments in sequential and spiral modes. All reconstructions were read for depiction of DVT and to evaluate best reconstruction parameters in comparison with the thinnest reconstruction ("gold standard"). Image noise and venous enhancement were measured as objective criteria for image quality. Subjective image quality was rated on a four-point scale. Effective dose was estimated for all reconstructions. In sequential 10/50 reconstruction DVT was completely detected in 13/26 cases, partially in 10/26 cases and was not detected at all in 3/26 cases, and 15/26, 9/26 and 2/26 cases for the 10/20 reconstruction, respectively. DVT was completely detected in all spiral reconstructions. Image noise ranged between 14.8-29.1 HU. Median image quality was 2. Estimated effective dose ranged between 2.3 mSv and 11.8 mSv. Gaps in sequential protocols may lead to false negative results. Therefore, spiral scanning protocols for complete depiction of DVT are mandatory.