Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2021
- Goals
- Record
- dadcfbaa-928f-461d-a801-ab418dca3a96
- Published
- 2026-10-30
Universitas Dian Nuswantoro conducted a study on osteosarcoma classification using a convolutional neural network (CNN). The activity involved Asmaria, Talitha; Mayasari, Dita Ayu; Heryanto, M.A.; Kurniatie, Menik; Wati, Rahma; and Aurellia, Salsabila as participants. The study addressed the diagnostic context in which biopsy is the final procedure for distinguishing primary and secondary osteosarcomas. It also considered the difficulty caused by morphological diversity among specimens of neoplasms, which may lead to misdiagnosis.
The study aimed to classify cell viability in an osteosarcoma dataset using hematoxylin and eosin-stained samples. Matlab was used to construct the CNN to increase efficiency and precision. The CNN architecture consisted of six convolution layers, max-pooling layers, and fully connected layers for feature extraction. Data augmentation was applied to boost performance. The classification process included the non-tumor, non-viable, and viable classes.
The CNN classification process produced an average accuracy of 95.37 percent. The reported accuracy for the non-tumor class was 92.8 percent, for the non-viable class was 98 percent, and for the viable class was 95.239 percent. The study concluded that the CNN technique classified osteosarcoma with high accuracy. It also stated that the technique obtained better results compared with the manual process and could effectively help pathologists work. In relation to the stated goal, the activity constitutes health research focused on osteosarcoma cell-viability classification and a diagnostic process. The evidence records UDINUS's research activity, its computational method, and its reported classification findings.