Dr. Sherif Fakhry Mohamed Abdelnaby, President of the Egyptian Russian University, announced that Dr. Amr Talaat, Minister of Communications and Information Technology, honored Dr. Mohamed Ali Salem, Lecturer in the Department of Artificial Intelligence at the Faculty of Artificial Intelligence at the university, by awarding him the third prize in the Arab Academic Research Competition for the Best Academic Research in Cybersecurity and Artificial Intelligence. The award ceremony took place during the closing event of the fourth edition of the International Conference on Information Security and Cybersecurity (CAISEC’25), organized by the Arab Organization for Information and Communication Technologies.

He added that the competition was under the patronage of Dr. Abdelmajid Bin Amara, Secretary General of the Federation of Arab Scientific Research Councils. The researcher’s participation was supported by Dr. Mohamed Kamal Mostafa, Chairman of the University’s Board of Trustees.

In the same context, Dr. Hesham Fathy, Dean of the Faculty of Artificial Intelligence at the Egyptian Russian University, explained that the research presented by Dr. Mohamed Ali Salem sets a realistic foundation for using artificial intelligence technologies in human medicine. It provides tools to help physicians in various health fields achieve more accurate diagnoses without the need for costly or long-term characterization. The next goal is to make this model ready for actual clinical use in hospitals. He pointed out that participating in prestigious competitions at both Arab and international levels aims to encourage faculty members to engage effectively with the international community, enhancing their skills and gaining experience in various technological fields.

Dr. Mohamed Ali Salem, Lecturer in the Department of Artificial Intelligence at the Faculty of Artificial Intelligence at the Egyptian Russian University, confirmed that the abstract of the research—ranked third among the best academic Arab research in cybersecurity and artificial intelligence—focuses on developing a new AI model to accurately define thyroid nodule boundaries in ultrasound images using “Weakly-Supervised Learning”.

The title of the research is:
“Weakly-supervised thyroid ultrasound segmentation: Leveraging multi-scale consistency, contextual features, and bounding box supervision for accurate target delineation.”

The lecturer added that one of the goals of the new model is to innovate a smart framework that uses only “Bounding Boxes,” which are easy and quick to create. The model integrates four advanced techniques to enhance accuracy:

– SAC (Scale-Aware Consistency): Ensures predictions fall within the correct range of the image.
– HPC (Hierarchical Prediction Consistency): Repeats predictions and compares results to improve distinction between the nodule and background.
– CFI (Contextual Feature Integration): Utilizes the surrounding context of the nodule for a better understanding of tissue.
– MPR (Multi-Path Refinement): Combines fine details and overall context by processing the image at multiple levels.

Dr. Mohamed Ali Salem stated that the new model achieved impressive accuracy on the TG3K and TN3K datasets, with results comparable to full supervision models, while significantly reducing the time and effort needed for annotation.

He summarized the research significance as follows:
– Reducing time and effort required for preparing AI training data.
– Providing high accuracy without the need for detailed labeling.
– Paving the way for practical and effective use of AI in hospitals.

He also emphasized that the model stands out due to its novelty and uniqueness compared to previous models, as it doesn’t require costly detailed data but offers performance close to full supervision. In terms of clinical relevance, the model helps doctors reduce diagnosis time and improve detection accuracy of nodules, especially in areas lacking human experts in annotation, such as rural hospitals or small clinics.