Enhancing Fetal Classification Accuracy through Computer-Aided Impainting Techniques

• Main Authors: Zanbouaa, Asmae, Zrira, Nabila, Slimani, Saad, Allak, Anass, Benmiloud, Ibtissam, Bouyakhf, El Houssine
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Artificial intelligence; Buildings; Classification; Data models; Deep learning; Fetal; Hospitals; Inpainting; Medical diagnostic imaging; Ultrasonic imaging; Ultrasound
• ISSN: 2832-8337
• DOI: 10.1109/ISIVC61350.2024.10577871

To improve fetal classification accuracy, computer aided techniques have been created to assist doctors in classifying various planes of fetal scans, thereby eliminating errors associated with traditional diagnostic procedures that rely primarily on medical skills. The effectiveness of these systems is critical to achieving high-quality categorization results. Even though various state-of-the-art approaches have been undertaken in this area, their effectiveness remains limited due to a primary focus on model building while neglecting fundamental data preparation. To address these issues, this research provides an ultrasound image-based classification method for fetal planes (i.e., cephalic, abdominal, and femur) that uses artificial intelligence applied to inpainted images, emphasizing the importance of this approach. Furthermore, our studies have been carried out using a carefully annotated dataset gathered from different hospitals in Morocco.