Quadratic Loss-Based Optimization for Multiclass Image Classification in Neo Uro Light Vision
DOI:
https://doi.org/10.35145/z5q4f997Keywords:
Quadratic Loss, Multiclass Image Classification, Medical Image Analysis, Class Imbalance, Artificial Intelligence, Neo Uro Light Vision, F1-Score OptimizationAbstract
Artificial intelligence (AI)-based image classification offers potential support for preliminary medical screening. However, multiclass classification systems may exhibit uneven performance across categories, particularly when the dataset contains an imbalanced class distribution. This study proposes a quadratic loss-based optimization framework for multiclass image classification in Neo Uro Light Vision, an AI-based system designed to support preliminary screening of external genital health conditions in children. The study employs a quantitative experimental approach, using precision, recall, and F1-score to evaluate classification performance across ten classes. The initial evaluation involved 179 images and achieved a weighted precision of 89.77%, weighted recall of 88.27%, and weighted F1-score of 88.41%. Class-wise results revealed variations in performance, with the F1-score ranging from 72.73% for Penile_strangulation to 93.33% for Fistula. The dataset also exhibited substantial class imbalance, with 84 images in the Hipospadia class compared with only four in the Penile_strangulation class. Based on these findings, a quadratic objective is formulated to quantify deviations between class-specific F1-scores and predefined performance targets. The proposed framework is intended to support the systematic comparison of model configurations and the evaluation of class-wise performance. However, the effectiveness of the optimization approach has not yet been empirically established and requires comparative experiments and independent validation. This study provides a mathematical foundation for further investigating class-sensitive optimization in AI-based preliminary screening while emphasizing the distinction between computational classification performance and clinical diagnostic validity.
References
Abbott, E. E., Rehman, T., Rosania, A., Lum, D. L., Taylor, T. B., Kirk, A. J., Taylor, R. A., Baker, E. F., Rabin, E., Padela, A., Genes, N., Srivastava, A., Sangal, R. B., & Apakama, D. (2026). Understanding and Addressing Bias in Artificial Intelligence Systems: A Primer for the Emergency Medicine Physician. JACEP Open, 7(1), 100311. https://doi.org/10.1016/J.ACEPJO.2025.100311
Ahsan, M. M., Ali, M. S., & Siddique, Z. (2024). Enhancing and improving the performance of imbalanced class data using novel GBO and SSG: A comparative analysis. Neural Networks, 173, 106157. https://doi.org/10.1016/J.NEUNET.2024.106157
Jena, B., Saxena, S., Nayak, G. K., Saba, L., Sharma, N., & Suri, J. S. (2021). Artificial intelligence-based hybrid deep learning models for image classification: The first narrative review. Computers in Biology and Medicine, 137, 104803. https://doi.org/10.1016/J.COMPBIOMED.2021.104803
Kanyana, R., Ounyesiga, L., & Uzorka, A. (2026). Artificial intelligence-assisted design and optimization of radiation shielding materials for medical applications. Journal of Radiation Research and Applied Sciences, 19(3), 102603. https://doi.org/10.1016/J.JRRAS.2026.102603
Kaur, R., & Singh, S. (2022). A comprehensive review of object detection with deep learning. Digital Signal Processing: A Review Journal, 132. https://doi.org/10.1016/j.dsp.2022.103812
Kautz, T., Eskofier, B. M., & Pasluosta, C. F. (2017). Generic performance measure for multiclass-classifiers. Pattern Recognition, 68, 111–125. https://doi.org/10.1016/J.PATCOG.2017.03.008
Kocak, B., Klontzas, M. E., Stanzione, A., Meddeb, A., Demircioğlu, A., Bluethgen, C., Bressem, K. K., Ugga, L., Mercaldo, N., Díaz, O., & Cuocolo, R. (2025). Evaluation metrics in medical imaging AI: fundamentals, pitfalls, misapplications, and recommendations. European Journal of Radiology Artificial Intelligence, 3, 100030. https://doi.org/10.1016/J.EJRAI.2025.100030
Liu, B., Pedersen, M., & Raja, K. (2026). Loss function assessment: Towards learning hard-to-learn classes. Neurocomputing, 686, 133638. https://doi.org/10.1016/J.NEUCOM.2026.133638
Mohammed, S., Budach, L., Feuerpfeil, M., Ihde, N., Nathansen, A., Noack, N., Patzlaff, H., Naumann, F., & Harmouch, H. (2025). The effects of data quality on machine learning performance on tabular data. Information Systems, 132, 102549. https://doi.org/10.1016/J.IS.2025.102549
Mollaie, F., Shakoor, M. H., Sepahvand, M., & Boostani, R. (2026). Medical data types in machine learning and their challenges: A comprehensive review. Intelligence-Based Medicine, 15, 100427. https://doi.org/10.1016/J.IBMED.2026.100427
Mukherjee, K. (2012). A Review of Robust Estimation under Conditional Heteroscedasticity. Handbook of Statistics, 30, 123–154. https://doi.org/10.1016/B978-0-444-53858-1.00006-5
Pitoyo, J., Jahrizal, Siregar, F. M., Nyoto, N., Putra, R., & Renaldo, N. (2026). Quantum Vision Ai For Pediatric Urological Screening Using Hybrid Quantum Machine Learning. Journal of Information Systems, Digitization and Business, 5(3), 01–07. https://doi.org/10.38142/jisdb.v5i3.2116
Rayed, M. E., Islam, S. M. S., Niha, S. I., Jim, J. R., Kabir, M. M., & Mridha, M. F. (2024). Deep learning for medical image segmentation: State-of-the-art advancements and challenges. Informatics in Medicine Unlocked, 47, 101504. https://doi.org/10.1016/J.IMU.2024.101504
Renaldo, N., Susanti, W., Yuliendi, R. R., Marlim, Y. N., Kurniawan, W. J., & Tendra, G. (2025). Innovations in Computer Vision for External Disease Identification. Journal of Information Systems, Digitization and Business, 4(1), 1–7.
Tarekegn, A. N., Giacobini, M., & Michalak, K. (2021). A review of methods for imbalanced multi-label classification. Pattern Recognition, 118, 107965. https://doi.org/10.1016/J.PATCOG.2021.107965
Yu, H. C., & Chen, Y. K. (2026). Improved FTIR-based classification for food authentication using a topological ensemble framework. Current Research in Food Science, 12, 101441. https://doi.org/10.1016/J.CRFS.2026.101441
Zhang, X., Wang, H., Zhou, H., Xiao, D., Xiong, H., Cheng, J., He, Y., Feng, Q., & Yang, J. (2026). An AI-powered diagnostic system for grading and invasion of non-muscle-invasive bladder cancer via TURBT specimens: A multicenter study. IScience, 29(7), 116651. https://doi.org/10.1016/J.ISCI.2026.116651
