Comparison of Deep Learning Methods for Osteosarcoma Classification Kemik T m r Siniflandirmasi i in Derin grenme Y ntemlerinin Kar sila stirilmasi

Emiroglu M. B., BİLGİN G.

2025 Innovations in Intelligent Systems and Applications Conference, ASYU 2025, Bursa, Türkiye, 10 - 12 Eylül 2025, (Tam Metin Bildiri) identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/asyu67174.2025.11208414
  • Basıldığı Şehir: Bursa
  • Basıldığı Ülke: Türkiye
  • Anahtar Kelimeler: Deep learning, Discrete cosine transform, Fast Fourier transform, HistoTransformer, Osteosarcoma classification, Signal processing, Wavelet transform
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

Osteosarcoma tumors present significant diagnostic challenges due to their complex and heterogeneous structures. This study investigates the integration of advanced signal processing techniques - Discrete Cosine Transform (DCT), Fast Fourier Transform (FFT), and Wavelet Transform (WT) - with state-ofthe- art deep learning models to improve osteosarcoma tumor classification accuracy. A comprehensive workflow leveraging feature extraction through three different transformation techniques is proposed, and the effectiveness of these techniques is compared. The performance of various deep learning models, including Convolutional Neural Networks (CNNs) and specialized hybrid architectures like DCT-HistoTransformer, is evaluated and benchmarked. Experimental results demonstrate the effectiveness of the proposed approach, highlighting the potential of DCT, FFT, and WT-based methods to achieve high diagnostic accuracy and support medical professionals in the early diagnosis and treatment planning of osteosarcoma.