What is it about?

We propose, in this work, a hybrid approach based on a pairing of semantic and visual attributes of images. We used two complementary methods: The first method is based on semantic attributes. In this method the indexing as well as the image search are based on specific keywords. The 2nd method is based on low level attributes namely: color, shape and texture. In this case the images are represented by vectors characterizing the digital content of the image (color, texture and shape). An appropriate combination of these two subsystems gives rise to a global system called: multimodal (hybrid) system. The results obtained are very encouraging, and prove the effectiveness of our approach.

Featured Image

Why is it important?

Multimodal learning system dedicated to public health practitioners. which can help radiologist experts in clinical decision-making during the diagnosis and treatment of certain diseases (facilitate the analysis and interpretation of radiological examinations).


The improved version of this approach, based on other techniques (e.g. segmentation by machine Learning or U-Net approaches) will generate a new system that will serve as a powerful CAD tool.

Noureddine BOURKACHE
Universite Mouloud Mammeri de Tizi Ouzou

Read the Original

This page is a summary of: Images indexing and matched assessment of semantics and visuals similarities applied to a medical learning X-ray image base, Journal of X-Ray Science and Technology, September 2022, IOS Press,
DOI: 10.3233/xst-221180.
You can read the full text:



The following have contributed to this page