A Quantitative and Semi-Quantitative Analysis of Computed Tomography Indices in Patients With Pneumonia and Acute Respiratory Failure in the Intensive Care Unit
A Quantitative and Semi-Quantitative Analysis of Computed Tomography Indices in Patients With Pneumonia and Acute Respiratory Failure in the Intensive Care Unit
Acute respiratory failure is a primary cause of intensive care unit admissions. It occurs when the lungs fail to adequately oxygenate arterial blood and/or prevent carbon dioxide retention. While the definition does not include absolute values, arterial PaO2 below 60 mmHg and arterial PaCO2 above 50 mmHg are generally accepted as indicators. However, these values should be interpreted in the context of individual patient characteristics. Pneumonia remains the most common etiology of acute respiratory failure. Typically of infectious origin, pneumonia alters respiratory mechanics, disrupting the lung's gas exchange function, ventilation-perfusion balance, and volumetric spirometric parameters.
Mortality and morbidity rates increase significantly when pneumonia patients require invasive mechanical ventilation. Recent advancements in quantitative CT technology enable clinicians to assess the volumetric state of the lungs without performing spirometric tests. Volumetric lung measurements aid in diagnosing lung diseases, assessing severity, planning treatment strategies, and predicting prognosis. Numerous studies have demonstrated promising correlations between quantitative CT data and physiological measurements in monitoring various pulmonary conditions, including Interstitial Lung Disease, Chronic Obstructive Pulmonary Disease, Small Airway Diseases, and COVID-19 Pneumonia.
CT scans are routinely performed on patients presenting with acute respiratory failure due to pneumonia. While imaging primarily evaluates lung parenchyma, additional tests such as spirometry are typically required to assess functional volumetric changes in the lungs. However, performing spirometric tests on critically ill patients is extremely challenging and often impractical, though theoretically possible.
Previous research has successfully demonstrated correlations between quantitative CT measurements and disease prognosis, particularly in chronic lung diseases. These measurements have also been utilized in acute conditions such as COVID-19. In critically ill patients with acute respiratory failure, additional lung information can assist clinicians in prognostic prediction and facilitate earlier intervention.
This study employs quantitative CT (qCT) measurements derived from CT attenuation histograms to examine the relationship between these parameters and disease prognosis in pneumonia patients with acute respiratory failure on invasive mechanical ventilation. These measurements include mean lung attenuation (MLA) and threshold-based volumetric measurements [low-density volume (LDV), medium-density volume (MDV), high-density volume (HDV), the ratio of MDV to total lung volume (MDV/TLV), and the ratio of HDV to total lung volume (HDV/TLV)].
Statistical analyses were planned to be conducted using IBM SPSS for Windows version 29.0 (IBM Corp., Armonk, NY, USA). The Kolmogorov-Smirnov test was intended to be employed to assess the normality assumption. Continuous variables were planned to be presented as median and interquartile range (IQR) in cases where the normality assumption was not met. Categorical variables were to be summarized as frequencies and percentages. Inter-group comparisons were planned to be performed using the Mann-Whitney U test. The relationships between categorical variables were intended to be examined using the Chi-square test.
Acute respiratory failure is a leading cause of intensive care unit admissions. It occurs when the lungs fail to adequately oxygenate arterial blood and/or prevent carbon dioxide retention. Although the definition does not include absolute values, arterial PaO2 below 60 mmHg and arterial PaCO2 above 50 mmHg are generally accepted as indicators. However, these values should be considered in the context of individual patients. Pneumonia is the most common cause of acute respiratory failure. Pneumonia is often infectious in origin. It alters respiratory mechanics, disrupting the lung's gas exchange function, ventilation-perfusion balance, and volumetric spirometric values.
Mortality and morbidity increase when pneumonia patients require invasive mechanical ventilation. Recent technological advancements in quantitative CT measurements allow clinicians to assess the volumetric state of the lungs without performing spirometric tests. Volumetric lung measurements assist clinicians in diagnosing lung diseases, assessing severity, planning treatment, and predicting prognosis. Numerous studies have shown promising correlations between quantitative CT data and physiological measurements in monitoring various lung diseases, including Interstitial Lung Disease, Chronic Obstructive Pulmonary Disease, Small Airway Diseases, and COVID-19 Pneumonia.
CT scans are routinely performed on patients with acute respiratory failure due to pneumonia upon initial presentation. While imaging primarily evaluates lung parenchyma, additional tests like spirometry are needed to assess functional volumetric changes in the lungs. However, performing spirometric tests on critically ill patients is extremely challenging and often impractical, though theoretically possible.
Previous studies have successfully demonstrated correlations between quantitative CT measurements and disease prognosis, particularly in chronic lung diseases. These measurements have also been utilized in acute conditions such as COVID-19. In critically ill patients with acute respiratory failure, additional lung information can aid clinicians in prognostic prediction and enable earlier intervention.
This study employs quantitative CT (qCT) measurements obtained from CT attenuation histograms to examine the relationship between these parameters and disease prognosis in pneumonia patients with acute respiratory failure on invasive mechanical ventilation. These measurements include mean lung attenuation (MLA) and threshold-based volumetric measurements [low-density volume (LDV), medium-density volume (MDV), high-density volume (HDV), the ratio of MDV to total lung volume (MDV/TLV), and the ratio of HDV to total lung volume (HDV/TLV)].
Method Patient Selection A retrospective review of 88 patients diagnosed with acute respiratory failure due to pneumonia, admitted to the general intensive care unit of Kocaeli University Medical Faculty between 2017 and 2024, was planned. Patients with acute respiratory failure due to pneumonia, who had CT imaging upon admission and were not intubated at admission, were to be included in the study. CT scans performed for diagnostic purposes in the emergency department or hospital ward were defined as screening CTs.
Exclusion criteria were planned to include patients under 18 years of age, those with non-infectious etiology of acute respiratory failure, COVID-19 patients, those intubated during transfer to intensive care, lung cancer patients, those with known chronic lung disease, patients intubated during the first ICU examination, and those whose intubation reason within the first 10 days was not acute respiratory failure.
The study was approved by the Ethics Committee of Kocaeli University Faculty of Medicine, Kocaeli, Turkey. The study number is KU GOKAEK-2024/07.45 (Project number: 2024/216).
Image Quantification and Quantitative CT (qCT) Measurements Baseline CT scans of patients admitted to intensive care with pneumonia and acute respiratory failure between 2017 and 2024 were planned to be retrieved from the radiology archive. All scans were performed using a 16-slice multi-detector CT scanner (Aquilion TSX-101A, Toshiba Medical Systems Corporation, Tokyo, Japan) with the following scan parameters: 250 mAs, 120 kV, pitch 1, 16 × 1 mm collimation, 1 mm reconstruction interval, and 1 mm reconstruction slice thickness. Scans of patients in supine position and at full inspiration during imaging were planned to be included in the study. Lung parenchyma was to be analyzed with a window width of 1,600 Hounsfield Units (HU) and a level of -500 HU.
Scans were analyzed using Vitrea® Advanced Visualization software (Canon Group, Minnetonka, MN) for image quantification. Images that had automatically detected main airways, blood vessels, airway branching points, lungs, and lobes were included in the image analysis. Blood vessels and airways up to the subsegmental levels were automatically excluded from quantification. This process was intended to prevent the analysis of any anatomical structures that could lead to errors in assessing lung parenchymal density.
Following this process, MLA was planned to be automatically calculated based on the analysis of lung density histograms using digital image processing. Using a density mask technique that highlights voxels within a specific density range to aid quantitation, three different volumes based on the volume of voxels in three different density ranges were planned to be obtained. Thresholds were to be automatically set at -1020 to -920 HU for low-density volume (LDV), -920 to -720 HU for medium-density volume (MDV), and -720 to 0 HU for high-density volume (HDV). Total lung volume (TLV) was to be calculated as equivalent to the total volume of lung voxels in the range of -1020 to 0 HU, i.e., the sum of LDV + MDV + HDV for each patient.
The Low Density Index (LDI%) was planned to be defined as the ratio between the lung volume below the low-density threshold for lung density and the lung parenchymal volume containing lung voxels below the upper density threshold. LDI% is the 3D equivalent of the 2D relative area index (RA%) used in early lung density studies. The LD (Low Density) Index (%) was to be recorded as Low volume / (Low volume + Medium volume).
Following [Parr2009] and [Stoel2004] studies, the 15th percentile point (Perc15) was defined as the cut-off value at which the 15% of lung voxels with the lowest density are distributed and was to be expressed in Hounsfield Units (HU). The 15% point (HU) index is converted to Percentile Density PD15% by adding 1000 (HU), assuming air density is 0 g/l and air CT attenuation is -1000 HU. For example, a 15th percentile point index of -883 HU corresponds to a percentile density PD15%, which is (-883 + 1000) = 117 g/l. PD15 (g/l) was planned to be expressed in g/l by adding 1000(HU) to the HU value at which voxels with a specific percentage (15%) in the frequency distribution histogram have a lower density.
Statistical analyses were planned to be conducted using IBM SPSS for Windows version 29.0 (IBM Corp., Armonk, NY, USA). The Kolmogorov-Smirnov test was intended to be employed to assess the normality assumption. Continuous variables were planned to be presented as median and interquartile range (IQR) in cases where the normality assumption was not met. Categorical variables were to be summarized as frequencies and percentages. Inter-group comparisons were planned to be performed using the Mann-Whitney U test. The relationships between categorical variables were intended to be examined using the Chi-square test.
Inclusion Criteria:
Exclusion Criteria: