TY - JOUR ID - TI - Stereo Photogrammetry vs Computed Tomography for 3D Medical Measurements AU - Doaa Awda Hussien PY - 2019 VL - 5 IS - 4 SP - 202 EP - 212 JO - Karbala International Journal of Modern Science مجلة كربلاء العالمية للعلوم الحديثة SN - 2405609X 24056103 AB - The acquisition of 3D body measurements by computer vision-based remote sensing is becoming extremely substantial in clinical studies nowadays. Thus, accurate 3D models of human anatomical surfaces are required in many clinical routines like disease diagnosis, patient follow-up, surgical planning, computer assisted surgery and different biomechanical applications. These models can be generated from different imaging techniques such as computed tomography (CT). However, 3D conventional medical imaging like CT scan have serious limitations in application such as exposing the patient to ionizing radiation due to repeated scans, highly cost imaging process, failure to provide color texture information, and also lacking to apply the process in standing position. Recently 3D model reconstruction using stereo-photogrammetry have been taken into consideration as a reliable alternative technique to CT scan. This is particularly true because photogrammetry can achieve accurate 3D models with low cost and high accuracy results and further considered a non-invasiveness and non-irradiating technique if compared to CT imaging. In this paper, 3D digital models generated by close range photogrammetry (CRP), is investigated for validation against CT scan models (Stereo-Lithography (STL)) which expressed as gold standard models. This is particularly applied to propose using CRP as an alternative non-irradiated solution to X-ray radiated imaging to provide accurate 3D measurements for plastic surgery treatment in certain medical cases. Two 3D plastic models used as test objects in this research and therefore scanned by CT scan device and hand-held digital camera. The accuracy assessment was carried out using cloud-to-mesh (C2M) function followed by co-registration process between each pair of 3D corresponding surfaces. The C2M deviation distances computed and analyzed for all tested models. The RMSE values for deviation distances between individual pairs of 3D corresponding surfaces were found to be several parts from millimeter in both tested objects. Statistical analysis was also applied which shows no significant differences between both compared generated models. Validation measurements of specific regions within the models were carried out by computing the relative difference between both techniques which found to be clinically adequate and medically accepted for application.

ER -