True Structured Network Surgery of EfficientNet-B3 with Focal Loss for Highly Imbalanced Dermoscopic Image Analysis
-
Argha Orion Silitonga
*
Universitas Klabat, Minahasa Utara,
Indonesia
(*) Corresponding Author
Abstract
Skin tumor classification relies heavily on deep convolutional neural networks, which are computationally expensive for edge deployment. This study proposes a resource-efficient pipeline using true structured network surgery on an EfficientNet-B3 architecture. To address the extreme class imbalance in the ISIC dataset, we implemented minority class oversampling alongside a custom Binary Focal Loss function. Our experiments evaluated 0%, 20%, and 50% structural channel pruning utilizing L2-norm magnitude importance. The baseline model achieved an Area Under the Curve (AUC) of 0.9314 with 10.70 million parameters. At 50% pruning, the parameter footprint dropped by 74% to 2.79 million, and floating-point operations (FLOPs) decreased by 71% to 0.55 G. Despite these reductions, latency metrics indicated a hardware execution plateau on an NVIDIA RTX 3060 Laptop GPU, stabilizing at approximately 12.93 ms per image. By dynamically optimizing the decision threshold using Youden's J statistic, the 50% pruned model maintained a high malignant sensitivity (recall) of 81.59%, despite an overall accuracy reduction to 76.69%. These results demonstrate that combining physical tensor pruning with focal loss yields highly compressed models capable of robust malignant lesion recall. Furthermore, post-hoc visual interpretability via Grad-CAM verified that these compressed architectures retained their clinical diagnostic integrity by accurately localizing genuine pathological features. The primary contributions of this research include the demonstration of true structural channel pruning to reduce model parameters by 74% for edge deployment, the preservation of an 81.59% malignant sensitivity in a highly compressed network using Binary Focal Loss and dynamic thresholding via Youden's J statistic, and the verification of post-pruning diagnostic integrity using Grad-CAM. This pipeline presents a viable pathway for deploying lightweight, accurate dermatological diagnostic tools in resource-constrained clinical environments.
Downloads
References
“International Agency for Research on Cancer, Skin Cancer,” Skin Cancer. Accessed: Sep. 08, 2026. [Online]. Available: https://www.iarc.who.int/cancer-type/skin-cancer
“Skin Cancer Information,” The Skin Cancer Foundation. Accessed: Sep. 08, 2026. [Online]. Available: https://www.skincancer.org/skin-cancer-information/
J. Heaton, “Ian Goodfellow, Yoshua Bengio, and Aaron Courville: Deep learning,” Genet Program Evolvable Mach, vol. 19, no. 1–2, pp. 305–307, Jun. 2018, doi: 10.1007/s10710-017-9314-z.
B. Shetty, R. Fernandes, A. P. Rodrigues, R. Chengoden, S. Bhattacharya, and K. Lakshmanna, “Skin Lesion Classification of Dermoscopic Images Using Machine Learning and Convolutional Neural Network,” Sci Rep, vol. 12, no. 1, p. 18134, Oct. 2022, doi: 10.1038/s41598-022-22644-9.
A. Esteva et al., “Deep Learning-enabled Medical Computer Vision,” npj Digit. Med., vol. 4, no. 1, p. 5, Jan. 2021, doi: 10.1038/s41746-020-00376-2.
N. Codella et al., “Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC),” 2019, arXiv. doi: 10.48550/ARXIV.1902.03368.
Md. M. Hossain, Md. M. Hossain, Most. B. Arefin, F. Akhtar, and J. Blake, “Combining State-of-the-Art Pre-Trained Deep Learning Models: A Noble Approach for Skin Cancer Detection Using Max Voting Ensemble,” Diagnostics, vol. 14, no. 1, p. 89, Dec. 2023, doi: 10.3390/diagnostics14010089.
K. He et al., “Transformers in Medical Image Analysis,” Intelligent Medicine, vol. 3, no. 1, pp. 59–78, Feb. 2023, doi: 10.1016/j.imed.2022.07.002.
A. F. Sönmez, S. Çakar, F. Cerezci̇, M. Kotan, İ. Deli̇Başoğlu, and G. Çi̇T, “Deep Learning-Based Classification of Dermoscopic Images for Skin Lesions,” Sakarya University Journal of Computer and Information Sciences, vol. 6, no. 2, pp. 114–122, Aug. 2023, doi: 10.35377/saucis...1314638.
M. J. Beirami, F. Gruzmark, R. Manwar, M. Tsoukas, and K. Avanaki, “Lightweight Attention-Based Architecture for Accurate Melanoma Recognition,” Electronics, vol. 14, no. 21, p. 4281, Oct. 2025, doi: 10.3390/electronics14214281.
P. Tschandl, C. Rosendahl, and H. Kittler, “The HAM10000 Dataset a Large Collection of Multi-source Dermatoscopic Images of Common Pigmented Skin Lesions,” Sci Data, vol. 5, no. 1, p. 180161, Aug. 2018, doi: 10.1038/sdata.2018.161.
M. M’hamedi, M. Merzoug, M. Hadjila, and A. Bekkouche, “Enhancing Melanoma Skin Cancer Classification through Data Augmentation,” TELKOMNIKA, vol. 22, no. 5, p. 1209, Jul. 2024, doi: 10.12928/telkomnika.v22i5.26106.
A. Akram, J. Rashid, M. A. Jaffar, M. Faheem, and R. U. Amin, “Segmentation and Classification of Skin Lesions Using Hybrid Deep Learning Method in The Internet of Medical Things,” Skin Res Technol, vol. 29, no. 11, p. e13524, Nov. 2023, doi: 10.1111/srt.13524.
E. S. Nugroho, I. Ardiyanto, and H. A. Nugroho, “Enhancing Dermoscopic Pigmented Skin Lesion Classification: A Refined Approach Using The Pre-trained Inception-V3 Architecture,” Narra J, vol. 5, no. 2, p. e1852, Apr. 2025, doi: 10.52225/narra.v5i2.1852.
C. M. Nguyen and T.-S. Hy, “Efficient Deep Learning for Medical Imaging: Bridging the Gap Between High-Performance AI and Clinical Deployment,” Jan. 31, 2026, arXiv: arXiv:2602.00910. doi: 10.48550/arXiv.2602.00910.
Y. A. Amer, H. I. Saleh, and O. A. Nasr, “Optimized AlexNet Pruning for Edge-Based Medical Diagnostics,” IEEE Access, vol. 13, pp. 139129–139140, 2025, doi: 10.1109/ACCESS.2025.3593453.
T. Elwood, “Channel Pruning for Lightweight UNet in Medical Image Segmentation,” vol. 5, no. 5, May 2025, doi: 10.5281/ZENODO.15455265.
A. G. C. Pacheco and R. A. Krohling, “The Impact of Patient Clinical Information on Automated Skin Cancer Detection,” Computers in Biology and Medicine, vol. 116, p. 103545, Jan. 2020, doi: 10.1016/j.compbiomed.2019.103545.
A. K. Adepu, S. Sahayam, U. Jayaraman, and R. Arramraju, “Melanoma Classification from Dermatoscopy Images Using Knowledge Distillation for Highly Imbalanced Data,” Computers in Biology and Medicine, vol. 154, p. 106571, Mar. 2023, doi: 10.1016/j.compbiomed.2023.106571.
M. N. Mohammed-Ramzi and A. M. Aladdin, “Skin Lesion Classification Using CNN Model and Augmented Dataset,” Journal of Soft Computing and Computer Applications, vol. 3, no. 1, Jun. 2026, doi: 10.70403/3008-1084.1026.
A. Mahbod, G. Schaefer, C. Wang, G. Dorffner, R. Ecker, and I. Ellinger, “Transfer Learning Using a Multi-scale and Multi-network Ensemble for Skin Lesion Classification,” Computer Methods and Programs in Biomedicine, vol. 193, p. 105475, Sep. 2020, doi: 10.1016/j.cmpb.2020.105475.
M. Hassanzad and K. Hajian-Tilaki, “Methods of determining optimal cut-point of diagnostic biomarkers with application of clinical data in ROC analysis: an update review,” BMC Med Res Methodol, vol. 24, no. 1, p. 84, Apr. 2024, doi: 10.1186/s12874-024-02198-2.
Bila bermanfaat silahkan share artikel ini
Berikan Komentar Anda terhadap artikel True Structured Network Surgery of EfficientNet-B3 with Focal Loss for Highly Imbalanced Dermoscopic Image Analysis
Pages: 147-156
Copyright (c) 2026 Argha Orion Silitonga
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under Creative Commons Attribution 4.0 International License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (Refer to The Effect of Open Access).
