Deriving Hounsfield units using grey levels in cone beam CT: a clinical application

Standard computed tomography (CT) uses Hounsfield units (HU) to quantify tissue density. Cone beam CT (CBCT) lacks direct HU derivation, limiting its use in quantitative analysis. Prior research has shown that grey levels in CBCT correlate with linear attenuation coefficients, but no consistent method exists for converting these to HUs. This gap motivated the need for a reproducible technique. Researchers have explored regression models to link grey levels with known materials. However, no prior work had resolved how to translate these into HUs with clinical accuracy. The variability in photon energies and scanner models adds uncertainty. This study aims to address these limitations by testing a regression-based approach. The goal is to enable clinical use of CBCT for quantitative HU estimation.

This study aimed to develop and test a method for calculating Hounsfield units (HUs) from grey levels in cone beam CT (CBCT) scans. The specific problem addressed is the lack of a reliable way to derive HUs from CBCT data. The motivation comes from the need to use CBCT for quantitative tissue analysis in clinical settings. Current methods rely on standard CT, which is less accessible in dental and maxillofacial imaging. The study tested whether linear regression could link grey levels to HUs effectively. It also evaluated how scanner-specific factors affect the accuracy of HU calculations. The researchers sought to determine if a single regression model could be applied across different CBCT systems. This would allow for more consistent and clinically useful HU values in CBCT imaging.

The study used an acrylic intraoral reference object containing materials with known attenuation properties. These included aluminum, cortical bone equivalent, trabecular bone equivalent, polymethylmethacrylate, and water equivalent material. Thirty-one scans were performed on the Asahi Alphard 3030 and thirty on the Planmeca ProMax 3D CBCT systems. Grey levels of the reference materials were recorded for each scan. Linear regression was applied between grey levels and linear attenuation coefficients at various photon energies. The energy with the highest regression coefficient was selected as the effective energy. Hounsfield units were calculated using the standard HU equation based on the effective energy. The results were compared to HU values derived from grey levels using the regression equation.

The study found a satisfactory linear relationship between grey levels and linear attenuation coefficients across the tested materials. This allowed Hounsfield units to be calculated from grey levels using regression equations. The highest regression coefficients were observed at specific photon energies, which were used as effective energies. However, uncertainty in determining effective energies led to unrealistic values and variability in HU calculations. Direct regression from grey levels to Hounsfield units at fixed energies produced more consistent results. The method was tested on two different CBCT systems, showing variability in performance. The calculated HU values were compared to those from standard CT equations, showing reasonable agreement. These findings suggest that grey levels in CBCT can be used to estimate Hounsfield units with acceptable accuracy.

The study demonstrated that Hounsfield units can be derived from grey levels in cone beam CT using linear regression. The method relies on identifying an effective energy that maximizes the correlation between grey levels and attenuation coefficients. While the approach showed promise, uncertainty in determining effective energies introduced variability. Direct regression from grey levels to Hounsfield units at fixed energies provided more consistent results. The method was tested on two different CBCT systems, showing that scanner-specific factors influence accuracy. The findings suggest that this technique could be used clinically to estimate HU values from CBCT scans. However, the variability in effective energies highlights the need for further refinement. The study supports the feasibility of using CBCT for quantitative tissue analysis in clinical settings.