Head-to-head comparison of lung perfusion with dual-energy CT and SPECT-CT.

PURPOSE: To compare the quantitative and qualitative lung perfusion data acquired with dual energy CT (DECT) to that acquired with a large field-of-view cadmium-zinc-telluride camera single-photon emission CT coupled to a CT system (SPECT-CT). MATERIALS AND METHODS: A total of 53 patients who underwent both dual-layer DECT angiography and perfusion SPECT-CT for pulmonary hypertension or pre-operative lobar resection surgery were retrospectively included. There were 30 men and 23 women with a mean age of 65.4±17.5 (SD)years (range: 18-88years). Relative lobar perfusion was calculated by dividing the amount (of radiotracer or iodinated contrast agent) per lobe by the total amount in both lungs. Linear regression, Bland-Altman analysis, and Pearson's correlation coefficient were also calculated. Kappa test was used to test agreements in morphology and severity of perfusion defects assessed on SPECT-CT and on DECT iodine maps with a one-month interval. Wilcoxon rank sum test was used to compare the sharpness of perfusion defects and radiation dose among modalities. RESULTS: Strong correlations for relative lobar perfusion using linear regression analysis and Pearson's correlation coefficient (r=0.93) were found. Bland-Altman analysis revealed a -0.10 bias, with limits of agreement between [-6.01; 5.81]. With respect to SPECT- CT as standard of reference, the sensitivity, specificity, PPV, NPV, accuracy for lobar perfusion defects were 89.4% (95% CI: 82.6-93.4%), 96.5% (95% CI: 92.1-98.5%), 95.6% (95% CI: 90.9-97.8%), 91.4% (95% CI: 85.6-94.9%) and 93.0% (95% CI: 87.6-96.1%) respectively. High level of agreement was found for morphology and severity of perfusion defects between modalities (Kappa=0.84 and 0.86 respectively) and on DECT images among readers (Kappa=0.94 and 0.89 respectively). A significantly sharper delineation of perfusion defects was found on DECT images (P<0.0001) using a significantly lower equivalent dose of 4.1±2.3 (SD) mSv (range: 1.9-11.85mSv) compared to an equivalent dose of 5.3±1.1 (SD) mSv (range: 2.8-7.3mSv) for SPECT-CT, corresponding to a 21.2% dose reduction (P=0.0004). CONCLUSION: DECT imaging shows strong quantitative correlations and qualitative agreements with SPECT-CT for the evaluation of lung perfusion.

Convolutional neural network evaluation of over-scanning in lung computed tomography.

INTRODUCTION: The purpose of this study was to develop a convolutional neural network (CNN) to determine the extent of over-scanning in the Z-direction associated with lung computed tomography (CT) examinations. MATERIALS AND METHODS: The CT examinations of 250 patients were used to train the machine learning software and 100 were used to validate the results. Each lung CT examination was divided into cervical, lung, and abdominal areas by the CNN and 2 independent radiologists, and the length of each area was measured. Every part above or below the lung marks was labeled as over-scanning. The accuracy of the CNN was calculated after the training phase and agreement between CNN and radiologists was assessed using kappa statistics during the validation phase. After validation the software was used to estimate the length of each of the three areas and the total over-scanning in further 1000 patients. RESULTS: An accuracy of 0.99 was found for the testing dataset and a very good agreement (kappa=0.98) between the CNN and the radiologists' evaluation was found for the validation dataset. Over-scanning was 22.8% with the CNN and 22.2% with the radiologists. The degree of over-scanning was 22.6% in 1000 lung CT examinations. CONCLUSION: Our study shows a substantial over estimation of the length of the area to be scanned during lung CT and thus an unnecessary patient's over-exposure to ionizing radiation. This over-scanning can be assessed easily, reliably and quickly using CNN.