The aim of the current study is to propose an algorithm for the automated calculation of water-equivalent diameter (Dw) and size-specific dose estimate (SSDE) from clinical computed tomography (CT) images in head protocol context, a python-based algorithm was developed in order to execute an interface to read DICOM images and compute the mentioned metrics. All CT datasets were retrospectively acquired by the Siemens emotion 16 C T scanner. The algorithm developed in this study presents a novel approach for preprocessing and analyzing DICOM images obtained from head CT scans. Exploiting a custom Python script, the preprocessing pipeline begins with the conversion of pixel values to Hounsfield Units (HU) to ensure quantitative analysis accuracy. Subsequently, Gaussian filtering reduced noise while preserving essential features, followed by thresholding using Otsu's method for head region segmentation. Morphological operations refined the binary mask, which enhanced the method for contour detection via active contours. The outcome of this procedure is an accurate measurement of the head size metrics, including lateral (LAT) and anteroposterior (AP) dimensions, effective diameter (Deff), and water equivalent diameter (Dw). Statistical analyses are performed to derive mean statistics and compute SSDE using a conversion factor derived from Dw. Validation of the proposed algorithm used images of patients who had undergone head examination in a major pediatric university hospital. The proposed algorithm was compared to IndoseCT v20 b for Dw and SSDE calculations. The Dw values from HeadCTDosi were strongly correlated to those of IndoseCT by R2 = 0.97 and the SSDE values also presented high correlation of R2 = 0.93. This study presents an improved algorithm for calculating Dw and SSDE in the context of head CT, which notably can outperform the manual, subjective, and time-consuming process of contouring the head region, positioning it as a potentially essential tool for routine clinical dose monitoring.
Read full abstract