EVALUASI KOMPREHENSIF TEKNIK PREPROCESSING UNTUK KLASIFIKASI MRI TUMOR OTAK DENGAN MOBILENETV3-LARGE

Sayyid Miftah Farid; Ade Irma Purnamasari; Denni Pratama; Mulyawan .

Authors

Sayyid Miftah Farid STMIK IKMI Cirebon, Indonesia
Ade Irma Purnamasari STMIK IKMI Cirebon, Indonesia
Denni Pratama STMIK IKMI Cirebon, Indonesia
Mulyawan . STMIK IKMI Cirebon, Indonesia

Keywords:

Preprocessing, MRI, Tumor Otak

Abstract

Klasifikasi tumor otak dari citra MRI menggunakan Convolutional Neural Network (CNN) memerlukan tahap preprocessing yang optimal untuk memaksimalkan performa model. Penelitian ini mengevaluasi lima teknik preprocessing berbeda (standard resize, Aspect Ratio Preservation dengan Padding, Center Crop, Adaptive Resize, dan Multi-Scale Input) pada arsitektur MobileNetV3-Large untuk klasifikasi biner tumor otak. Dataset terdiri dari 3.000 citra MRI dengan distribusi seimbang yang menunjukkan heterogenitas resolusi tinggi (291 kombinasi unik) dan variasi aspek rasio signifikan antar kelas (tumor 0,872 vs non-tumor 0,988). Lima eksperimen independen dilakukan dengan konfigurasi model identik menggunakan transfer learning dan fine-tuning 20 layer teratas. Hasil menunjukkan Multi-Scale Input mencapai akurasi tertinggi 99,56% namun dengan overhead komputasi severe (waktu inferensi 43,62 ms, model size 23,34 MB). Aspect Ratio Preservation + Padding mencapai akurasi 98,89% dengan efisiensi superior (waktu inferensi 16,60 ms, overhead +6,1%, model size 20,52 MB), peningkatan signifikan 4,89 poin dari Baseline Standard Resize (94,00%). Analisis confusion matrix menunjukkan Aspect Ratio + Padding mencapai keseimbangan optimal antara sensitivitas (99,11%) dan spesifisitas dengan Nilai false positive 1,33% dan Nilai false negative 0,89%. Penelitian ini mengonfirmasi bahwa preservasi aspek rasio merupakan strategi preprocessing optimal untuk lightweight CNN pada klasifikasi MRI tumor otak, memberikan keseimbangan terbaik antara akurasi diagnostik tinggi dan efisiensi komputasi untuk deployment pada perangkat dengan keterbatasan sumber daya.

References

N. Ashar, G. Raut, V. Trivedi, S. Vishvakarma, and A. Kumar, “Quantmac: enhancing hardware performance in dnns with quantize enabled multiply-accumulate unit,” IEEE Access, vol. 12, pp. 43600–43614, 2024, doi: 10.1109/access.2024.3379906.

P. Yan and G. Lu, “Design of an intelligent educational evaluation system using deep learning,” IEEE Access, vol. 11, pp. 29790–29799, 2023, doi: 10.1109/access.2023.3260979.

M. Al-Andoli, S. Tan, K. Sim, M. Seera, and C. Lim, “A parallel ensemble learning model for fault detection and diagnosis of industrial machinery,” IEEE Access, vol. 11, pp. 39866–39878, 2023, doi: 10.1109/access.2023.3267089.

E. Manor and S. Greenberg, “Custom hardware inference accelerator for tensorflow lite for microcontrollers,” IEEE Access, vol. 10, pp. 73484–73493, 2022, doi: 10.1109/access.2022.3189776.

M. Musthafa, I. Manimozhi, T. Mahesh, and S. Guluwadi, “Optimizing double-layered convolutional neural networks for efficient lung cancer classification through hyperparameter optimization and advanced image pre-processing techniques,” BMC Med. Inform. Decis. Mak., vol. 24, no. 1, 2024, doi: 10.1186/s12911-024-02553-9.

H. Lu, X. Feng, and J. Zhang, “Early detection of cardiorespiratory complications and training monitoring using wearable ecg sensors and cnn,” BMC Med. Inform. Decis. Mak., vol. 24, no. 1, 2024, doi: 10.1186/s12911-024-02599-9.

I. Orășan, A. Bublea, and C. Caleanu, “Deep learning-based eye gaze estimation for automotive applications using knowledge distillation,” IEEE Access, vol. 11, pp. 120741–120753, 2023, doi: 10.1109/access.2023.3325134.

H. Uyanik et al., “A multi-input convolutional neural networks model for earthquake precursor detection based on ionospheric total electron content,” Remote Sens. (Basel)., vol. 15, no. 24, p. 5690, 2023, doi: 10.3390/rs15245690.

T. Devi and K. Saji, “A stacked custom convolution neural network for voxel-based human brain morphometry classification,” Sci. Rep., vol. 15, no. 1, 2025, doi: 10.1038/s41598-025-17331-4.

G. Datta et al., “A processing-in-pixel-in-memory paradigm for resource-constrained tinyml applications,” Sci. Rep., vol. 12, no. 1, 2022, doi: 10.1038/s41598-022-17934-1.

X. Tian, L. Shi, Y. Luo, and X. Zhang, “Garbage classification algorithm based on improved mobilenetv3,” IEEE Access, vol. 12, pp. 44799–44807, 2024, doi: 10.1109/access.2024.3381533.

B. Olimov, B. Subramanian, R. Ugli, J. Kim, and J. Kim, “Consecutive multiscale feature learning-based image classification model,” Sci. Rep., vol. 13, no. 1, 2023, doi: 10.1038/s41598-023-30480-8.

O. Oyelade and A. Ezugwu, “A novel wavelet decomposition and transformation convolutional neural network with data augmentation for breast cancer detection using digital mammogram,” Sci. Rep., vol. 12, no. 1, 2022, doi: 10.1038/s41598-022-09905-3.

J. Rahman, M. Hanif, O. Rehman, U. Haider, S. Qaisar, and P. Plawiak, “Stages prediction of alzheimer’s disease with shallow 2d and 3d cnns from intelligently selected neuroimaging data,” Sci. Rep., vol. 15, no. 1, 2025, doi: 10.1038/s41598-025-93560-x.

F. Zhang, F. Wang, W. Zhang, Q. Wang, Y. Liu, and Z. Jiang, “Rotu-net: an innovative u-net with local rotation for medical image segmentation,” IEEE Access, vol. 12, pp. 21114–21128, 2024, doi: 10.1109/access.2024.3363410.

U. Lilhore et al., “Prevalence and risk factors analysis of postpartum depression at early stage using hybrid deep learning model,” Sci. Rep., vol. 14, no. 1, 2024, doi: 10.1038/s41598-024-54927-8.

C. Tinauer et al., “Interpretable brain disease classification and relevance-guided deep learning,” Sci. Rep., vol. 12, no. 1, 2022, doi: 10.1038/s41598-022-24541-7.

J. Liu, T. Wang, A. Skidmore, Y. Sun, P. Jia, and K. Zhang, “Integrated 1d, 2d, and 3d cnns enable robust and efficient land cover classification from hyperspectral imagery,” Remote Sens. (Basel)., vol. 15, no. 19, p. 4797, 2023, doi: 10.3390/rs15194797.

J. Pandian, R. Thirunavukarasu, and E. Kotei, “A novel convolutional neural network model for automatic speaker identification from speech signals,” IEEE Access, vol. 12, pp. 51381–51394, 2024, doi: 10.1109/access.2024.3385858.

H. Guo and L. Ren, “A marine small-targets classification algorithm based on improved convolutional neural networks,” Remote Sens. (Basel)., vol. 15, no. 11, p. 2917, 2023, doi: 10.3390/rs15112917.

X. Xu, S. Huang, and H. Lai, “Lightweight semantic segmentation network leveraging class-aware contextual information,” IEEE Access, vol. 11, pp. 144722–144734, 2023, doi: 10.1109/access.2023.3345790.

F. Siraj, S. Ayon, M. Samad, J. Uddin, and K. Choi, “Few-shot lightweight squeezenet architecture for induction motor fault diagnosis using limited thermal image dataset,” IEEE Access, vol. 12, pp. 50986–50997, 2024, doi: 10.1109/access.2024.3385430.

F. Alantali, Y. Halawani, B. Mohammad, and M. Al-Qutayri, “Slid: exploiting spatial locality in input data as a computational reuse method for efficient cnn,” IEEE Access, vol. 9, pp. 57179–57187, 2021, doi: 10.1109/access.2021.3071409.

C. Lee, J. Kong, and A. Munir, “Arithmetic coding-based 5-bit weight encoding and hardware decoder for cnn inference in edge devices,” IEEE Access, vol. 9, pp. 166736–166749, 2021, doi: 10.1109/access.2021.3136888.

Y. Kok et al., “Classification of osteoarthritic and healthy cartilage using deep learning with raman spectra,” Sci. Rep., vol. 14, no. 1, 2024, doi: 10.1038/s41598-024-66857-6.

Z. A. Hasibuan, Metodologi penelitian pada bidang ilmu komputer dan teknologi informasi: Konsep, teknik, dan aplikasi. Fakultas Ilmu Komputer Universitas Indonesia, 2007.