Intracranial calcifications, particularly within the falx cerebri, serve as crucial diagnostic markers ranging from benign accumulations to signs of severe pathologies. The falx cerebri, a dural fold that separates the cerebral hemispheres, presents challenges in visualization due to its low contrast in standard imaging techniques. Recent advancements in artificial intelligence (AI), particularly in machine learning and deep learning, have significantly transformed radiological diagnostics. This study aims to explore the application of AI in the segmentation and detection of falx cerebri calcifications using Cone-Beam Computed Tomography (CBCT) images through a comprehensive literature review and a detailed case report. The case report presents a 59-year-old patient diagnosed with falx cerebri calcifications whose CBCT images were analyzed using a cloud-based AI platform, demonstrating effectiveness in segmenting these calcifications, although challenges persist in distinguishing these from other cranial structures. A specific search strategy was employed to search electronic databases, yielding four studies exploring AI-based segmentation of the falx cerebri. The review detailed various AI models and their accuracy across different imaging modalities in identifying and segmenting falx cerebri calcifications, also highlighting the gap in publications in this area. In conclusion, further research is needed to improve AI-driven methods for accurately identifying and measuring intracranial calcifications. Advancing AI applications in radiology, particularly for detecting falx cerebri calcifications, could significantly enhance diagnostic precision, support disease monitoring, and inform treatment planning.
This retrospective study aims to investigate the impact of cone-beam computed tomography (CBCT) viewing parameters such as contrast, slice thickness, and sharpness on the identification of the inferior alveolar nerve (IAC). A total of 25 CBCT scans, resulting in 50 IACs, were assessed by two investigators using a three-score system (good, average, and poor) on cross-sectional images. Slice thicknesses of 0.25 mm, 0.5 mm, and 1 mm were tested, along with varying sharpness (0, 6, 8, and 10) and contrast (0, 400, 800, and 1200) settings. The results were statistically analyzed to determine the optimal slice thickness for improved visibility of IAC, followed by evaluating the influence of sharpness and contrast using the optimal thickness. The identified parameters were then validated by performing semi-automated segmentation of the IACs and structure overlapping to evaluate the mean distance. Inter-rater and intra-rater reliability were assessed using Kappa statistics, and inferential statistics used Pearson's Chi-square test. Inter-rater and intra-rater reliability for all parameters were significant, ranging from 69% to 83%. A slice thickness of 0.25 mm showed consistently "good" visibility (80%). Sharpness values of zero and contrast values of 1200 also demonstrated high frequencies of "good" visibility. Overlap analysis resulted in an average mean distance of 0.295 mm and a standard deviation of 0.307 mm across all patients' sides. The study revealed that a slice thickness of 0.25 mm, zero sharpness value, and higher contrast value of 1200 improved the visibility and accuracy of IAC segmentation in CBCT scans. The individual patient's characteristics, such as anatomical variations, decreased bone density, and absence of canal walls cortication, should be considered when using these parameters.
- Publication type
- Journal Article MeSH
