Objectives

i) To define a quantitative framework, based on AI, for bone micro-cracks automatic description and classification;

ii) to define morphological indices to classify healthy and pathological bones;

iii) to compare the approach based on artificial intelligence with the topological one; 

iv) to validate the micro-scale morphological indices.

Topic in Brief

The aim is to implement an approach based on AI in the form of CNN to automatically detect cracks and morphological indices in order to manage the large amount of data obtained from Synchrotron and SEM images. This automatic tool will allow the analysis of a large number of samples to statically quantify the indices characterizing healthy and pathological bones. The AI-tool permits to classify the physio-pathological conditions, as an aid in the clinical diagnosis.

Enrolment &

Planned Secondments

Enrolment: NTNU

Secondments:

1) Prof. Dlotko (IMPAN): to discuss and compare the different approaches to automatically detect cracks from images   

2) Prof. Carradò (CNRS): to measure micro-scale morphological indices and cracks from SEM 3D images

Expected Results

i) A quantitative framework, based on convolutional neural networks, able to detect the micro-cracks from the images acquired at the synchrotron. The ground truth is the manual segmentation of synchrotron slices;

ii) validation of the framework by the comparison with experimental measurements;

iii) the definition of micro-scale indices to evaluate crack morphological characteristics;

iv) validation of micro-scale morphological indices;

v) classification of the micro-morphologies and damages of healthy and pathological bones.