E1094. Ultra-High-Resolution Photon-Counting CT of the Lung: Optimization of Reconstruction Kernel and Quantum Iterative Reconstruction
  1. Adrienn Tóth; Medical University of South Carolina
  2. Jordan Chamberlin; Medical University of South Carolina
  3. Gregory Puthoff; Medical University of South Carolina
  4. Jim O'Doherty; Medical University of South Carolina; Siemens Medical Solutions
  5. Dhruw Maisuria; Medical University of South Carolina
  6. Aaron McGuire; Medical University of South Carolina
  7. Ismail Kabakus; Medical University of South Carolina
CT has long been the preferred imaging modality for the evaluation of the lung parenchyma. Previous studies have shown that with photon-counting (PC) CT, higher-order bronchi, small pulmonary vessels, and small lung nodules can be detected, in comparison with conventional energy-integrating-detector (EID) CT. The ultra-high-resolution (UHR) mode of PCCT enables imaging at 0.2-mm slice thickness. This can be utilized for better visualization of small anatomic structures of the lung parenchyma and subtle changes associated with lung pathologies. Detailed visualization of subtle lung abnormalities and morphologic changes can improve the reader’s confidence in evaluating interstitial pneumonia, pulmonary emphysema, and lung nodule borders, and in classifying interstitial lung diseases (ILDs). The purpose of our study was to evaluate the optimal reconstruction kernel and the optimal strength level of quantum iterative reconstruction (QIR) for UHR PCCT of the lung.

Materials and Methods:
In our first study cohort, 25 patients who underwent unenhanced chest CT were retrospectively included. Images were reconstructed with four different kernels: Bl64 (least sharp), Br76, Br84, and Br96 (sharpest). Our second study cohort included 24 patients who underwent unenhanced chest CT. Images were reconstructed with all strength levels of QIR (1 - 4) and QIR-Off (QIR-0). All examinations were performed at the PCCT scanner in UHR mode (0.2-mm slice thickness). The objective analysis included measurements of CT attenuation, noise, signal-to-noise Ratio (SNR), and contrast-to-noise Ratio (CNR). Two independent readers rated the images regarding image sharpness, airway details, subjective image noise, and overall image quality on a 5-point Likert scale.

In the first cohort, the Br76 kernel consistently outperformed all other reconstruction kernels in every category of the quantitative and qualitative analysis (all <em>p</em> < .001). In the second cohort, noise was reduced by 66% from QIR-Off to QIR-4 (<em>p</em> < .001). CNR was 2.8-fold higher from QIR-Off to QIR-4 (<em>p</em> < .001). Subjective image noise was best for QIR-4 (<em>p</em> < .001), and image sharpness and airway details were rated best at intermediate QIR levels.

UHR PCCT provides a detailed evaluation of subtle lung abnormalities and morphological changes, anticipating promising clinical benefits in the imaging of various chest disorders. Optimizing the reconstruction parameters can help to exploit the potential of the UHR mode. Our recent two studies provided a detailed evaluation of the image quality using different reconstruction kernels and strength levels of QIR. Considering all quantitative and qualitative image quality metrics, our results indicate that the Br76 kernel and QIR-3 provide the best image quality for UHR PCCT of the lung. The results of these studies may guide protocol optimization for UHR PCCT of the lung.