Development of a photogrammetric tool for cranial deformation analysis in infants

Resumen

Image-based and range-based solutions have great potential for accurate measurements in many different fields, including medicine. Cranial deformation is a problem with high prevalence among infants during their first months of life. Cranial deformation has different causes and associated risks. Most common deformations have positional causes and have long been considered to cause aesthetical problems only. Nowadays, authors are pointing at these deformations as a possible indicator of future developmental delays. Another type of cranial deformation is caused by the early closing of cranial sutures, that can lead to important health risks and is usually corrected by surgery. Early detection of the deformation leads to better development prospects, as cranium is especially malleable during the first months of life. Good monitoring also provides better data for evaluation of the different correction procedures, improving the outcome of the patients over time. Cranial deformation evaluation methodologies can be divided into three mean groups: 1) Low-cost measurements carried out by simple tools such as calliper and measuring tape and even visual assessment; 2) Radiological tests (Computed Tomography (CT) and Magnetic Resonance Imaging (MRI)), considered the gold standard, but costly and highly invasive; and 3) Image-based and range-based commercial solutions, non-invasive but too costly to be implemented as part of the clinical routine in many hospitals and clinics. In this doctoral thesis, the possibilities of photogrammetry as a method for cranial deformation assessment has been studied and a novel tool has been developed. The tool was required to be low-cost, objective and non-invasive, so no sedation would be required. The main challenge of the project was dealing with the movement of the infants during the consultation. In the first place, single-lens reflex (SLR) cameras were used to obtain head models of the patients. The methodology provided the required accuracy but the data acquisition was time-consuming and required an important setup. The data processing was also time-consuming, being totally manual and requiring expert users. The necessities to automatically obtain accurate 3D models from moving infants were studied and a novel tool was developed. The developed tool consists of a coded cap that works together with a smartphone app to guide the user during the data acquisition. Therefore, no expertise is required for data acquisition. The data is later processed on a server and the model and deformation parameters are made available to the user in less than 4 minutes with an accuracy of 2 mm. In order to make the model valuable for doctors, automatic measurements and deformation indexes are extracted from the model. Photogrammetry is a suitable technique to obtain high quality data and overcome the limitations. Together with ad-hoc developed software, it is the base of a fully automatic tool that is optimised for its use in clinics and hospitals. The developed tool is considered highly cost-effective and an important improvement as it provides similar data as existing costly setups, but it is also low-cost and portable.