Pengembangan Deep Learning untuk Sistem Deteksi Dini Komplikasi Kaki Diabetik Menggunakan Citra Termogram
DOI:
https://doi.org/10.25126/jtiik.1067382Abstrak
Prevalensi komplikasi kaki diabetik secara global mencapai 66% dengan resiko amputasi 20 kali lebih tinggi pada pasien diabetes mellitus. Tindakan pencegahan melalui deteksi dini komplikasi kaki diabetik mutlak dilakukan untuk meminimalisasi resiko amputasi. Penelitian sebelumnya menunjukkan validitas dan akurasi yang tinggi (mencapai 100%) dari sistem deteksi dini komplikasi kaki diabetik menggunakan termografi berbasis kecerdasan buatan. Namun sebagian besar penelitian tersebut terlalu berfokus pada peningkatan performa dan tidak memperhatikan aspek biaya komputasi yang berperan penting pada proses deployment model. Pada penelitian ini dirancang empat model deep convolutional neural network dengan prinsip Occam’s razor melalui pengaturan hyperparameter pada aspek struktur algoritma berupa jumah layer dan aspek optimasi berupa tipe optimizer. Penelitian bertujuan mengembangkan algoritma deep convolutional neural network untuk menghasilkan sistem deteksi dini komplikasi kaki diabetik dengan biaya komputasi terendah (jumlah parameter paling sedikit) dan mempertahankan kemampuan deteksi tetap tinggi (nilai rata-rata parameter evaluasi tertinggi). Data yang digunakan merupakan data primer berupa citra termogram telapak kaki dari RSUP. Dr. Sardjito Yogyakarta yang terdiri dari 20 subjek diabetes mellitus dan 20 subjek kontrol (sehat). Pengambilan data primer dilakukan menggunakan kamera thermal merek HIKMICRO B20 dengan resolusi inframerah 256x192 yang telah memenuhi standar internasional (IACT) untuk menghasilkan citra termogram dua dimensi. Hasil penelitian menunjukkan model 4 dengan Adam optimizer dan pengaturan hyperparameter tertentu merupakan model terbaik dengan jumlah parameter model paling sedikit yaitu 1.570.594 juta dan nilai rata-rata parameter evaluasi tetap tinggi sebesar 96%. Selain arsitektur deep convolutional neural network model 4, kontribusi penelitian yang didapatkan dari penelitian ini adalah penggunaan variasi ukuran filter 3x3, 2x2, dan 1x1 dengan jumlah convolutional layer yang tetap dan pengurangan jumlah hidden layer pada struktur algoritma mampu menurunkan jumlah parameter model dengan tetap mempertahankan kemampuan deteksi yang tinggi. Selain itu penelitian yang dilakukan merupakan penelitian pembuka atau pendahuluan mengenai perancangan sistem deteksi dini komplikasi kaki diabetik menggunakan termografi berbasis kecerdasan buatan deep learning di Indonesia.
Abstract
The prevalence of diabetic foot complications globally reaches 66% with a 20 times higher risk of amputation in patients with diabetes mellitus. Preventive measures through early detection of diabetic foot complications are necessary to minimize the risk of amputation. Previous studies have shown high validity and accuracy (up to 100%) of the early detection system of diabetic foot complications using artificial intelligence-based thermography. However, most of these studies focused too much on improving performance and did not pay attention to the computational cost aspect. In this study, four deep convolutional neural network models were designed with Occam's razor principle through hyperparameter settings on the algorithm structure aspect in the form of number of layers and optimization aspect in the form of optimizer type. The research aims to develop a deep convolutional neural network algorithm to produce an early detection system for diabetic foot complications with the lowest computational cost (least number of parameters) and maintain high detection capability (highest average value of evaluation parameters). The data used is primary data in the form of foot thermogram images from the General Hospital. Dr. Sardjito Yogyakarta consisting of 20 diabetes mellitus subjects and 20 control (healthy) subjects. Primary data collection was carried out using a thermal camera brand HIKMICRO B20 with 256x192 infrared resolution that has met international standards (IACT) to produce a two-dimensional color thermogram image. The results show that model 4 with Adam optimizer and certain hyperparameter settings is the best model with the least number of model parameters, namely 1,570,594 million and the average value of evaluation parameters remains high at 96%. In addition to the deep convolutional neural network architecture model 4, the research contribution obtained from this research is the use of filter size variations of 3x3, 2x2, and 1x1 with a fixed number of convolutional layers and a reduction in the number of hidden layers in the algorithm structure can reduce the number of model parameters while maintaining high detection capability. In addition, the research conducted can be an opening or preliminary research on the design of an early detection system for diabetic foot complications using deep learning artificial intelligence-based thermography in Indonesia.
Downloads
Referensi
ALZUBAIDI, L., FADHEL, M. A., OLEIWI, S. R., AL-SHAMMA, O., & ZHANG, J., 2020. DFU_QUTNet: diabetic foot ulcer classification using novel deep convolutional neural network, Multimedia Tools and Applications, 79(21–22), 15655–15677.
CHAWLA, A., CHAWLA, R., & JAGGI, S., 2016. Microvasular and macrovascular complications in diabetes mellitus: Distinct or continuum?, In Indian Journal of Endocrinology and Metabolism, Vol. 20, Issue 4, pp. 546–553. Medknow Publications.
CHOLLET, F., 2016. Xception: Deep Learning with Depthwise Separable Convolutions. http://arxiv.org/abs/1610.02357
CRUZ-VEGA, I., HERNANDEZ-CONTRERAS, D., PEREGRINA-BARRETO, H., RANGEL-MAGDALENO, J. DE J., & RAMIREZ-CORTES, J. M., 2020. Deep learning classification for diabetic foot thermograms, Sensors (Switzerland), 20(6).
FRYKBERG, R. G., GORDON, I. L., REYZELMAN, A. M., CAZZELL, S. M., FITZGERALD, R. H., ROTHENBERG, G. M., BLOOM, J. D., PETERSEN, B. J., LINDERS, D. R., NOUVONG, A., & NAJAFI, B., 2017. Feasibility and efficacy of a smart mat technology to predict development of diabetic plantar ulcers, Diabetes Care, 40(7), 973–980.
FRYKBERG, R. G., ZGONIS, T., ARMSTRONG, D. G., DRIVER, V. R., GIURINI, J. M., KRAVITZ, S. R., LANDSMAN, A. S., LAVERY, L. A., MOORE, J. C., SCHUBERTH, J. M., WUKICH, D. K., ANDERSEN, C., & VANORE, J. V., 2006. DIABETIC FOOT DISORDERS: A CLINICAL PRACTICE GUIDELINE (2006 revision), The Journal of Foot & Ankle Surgery, 45(5).
GOODFELLOW, I., BENGIO, Y., & COURVILLE, A., 2016. DEEP LEARNING, MIT Publisher. www.deeplearningbook.org
HE, K., ZHANG, X., REN, S., & SUN, J., 2015. Deep Residual Learning for Image Recognition. http://arxiv.org/abs/1512.03385
HERNANDEZ-CONTRERAS, D. A., PEREGRINA-BARRETO, H., RANGEL-MAGDALENO, J. D. J., & RENERO-CARRILLO, F. J., 2019. Plantar Thermogram Database for the Study of Diabetic Foot Complications, IEEE Access, 7, 161296–161307.
HERNANDEZ-CONTRERAS, D., PEREGRINA-BARRETO, H., RANGEL-MAGDALENO, J., & GONZALEZ-BERNAL, J., 2016. Narrative review: Diabetic foot and infrared thermography, In Infrared Physics and Technology, Vol. 78, pp. 105–117. Elsevier B.V.
HERNANDEZ-CONTRERAS, D., PEREGRINA-BARRETO, H., RANGEL-MAGDALENO, J., RAMIREZ-CORTES, J., & RENERO-CARRILLO, F., 2015. Automatic classification of thermal patterns in diabetic foot based on morphological pattern spectrum, Infrared Physics and Technology, 73, 149–157.
HUANG, G., LIU, Z., VAN DER MAATEN, L., & WEINBERGER, K. Q., 2016. Densely Connected Convolutional Networks. http://arxiv.org/abs/1608.06993
KELLEHER, D. J., 2019. Deep Learning. The MIT Press.
KHANDAKAR, A., CHOWDHURY, M. E. H., IBNE REAZ, M. BIN, MD ALI, S. H., HASAN, M. A., KIRANYAZ, S., RAHMAN, T., ALFKEY, R., BAKAR, A. A. A., & MALIK, R. A., 2021. A machine learning model for early detection of diabetic foot using thermogram images, Computers in Biology and Medicine, 137.
KHANDAKAR, A., CHOWDHURY, M. E. H., REAZ, M. B. I., ALI, S. H. M., ABBAS, T. O., ALAM, T., AYARI, M. A., MAHBUB, Z. B., HABIB, R., RAHMAN, T., TAHIR, A. M., BAKAR, A. A. A., & MALIK, R. A., 2022. Thermal Change Index-Based Diabetic Foot Thermogram Image Classification Using Machine Learning Techniques, Sensors, 22(5).
KHANDAKAR, A., CHOWDHURY, M. E. H., REAZ, M. B. I., ALI, S. H. M., KIRANYAZ, S., RAHMAN, T., CHOWDHURY, M. H., AYARI, M. A., ALFKEY, R., BAKAR, A. A. A., MALIK, R. A., & HASAN, A., 2022. A Novel Machine Learning Approach for Severity Classification of Diabetic Foot Complications Using Thermogram Images, Sensors, 22(11).
KHANDPUR, R. S., 2003. Handbook of Second Edition Biomedical Instrumentation (Second Edition). Tata McGraw-Hill Publishing Company Limited.
KINGMA, D. P., & BA, J., 2014. Adam: A Method for Stochastic Optimization. http://arxiv.org/abs/1412.6980
KROHN, JON., 2020. Deep Learning Illustrated (J. Lander, Ed.). Pearson Education, Inc.
MORI, T., NAGASE, T., TAKEHARA, K., OE, M., OHASHI, Y., AMEMIYA, A., NOGUCHI, H., UEKI, K., KADOWAKI, T., & SANADA, H., 2013. Morphological pattern classification system for plantar thermography of patients with diabetes, Journal of Diabetes Science and Technology, 7(5), 1102–1112.
MUNADI, K., SADDAMI, K., OKTIANA, M., ROSLIDAR, R., MUCHTAR, K., MELINDA, M., MUHARAR, R., SYUKRI, M., ABIDIN, T. F., & ARNIA, F., 2022. A Deep Learning Method for Early Detection of Diabetic Foot Using Decision Fusion and Thermal Images, Applied Sciences (Switzerland), 12(15).
MURALIDHARA, S., LUCIERI, A., DENGEL, A., & AHMED, S., 2022. Holistic multi-class classification & grading of diabetic foot ulcerations from plantar thermal images using deep learning, Health Information Science and Systems, 10(1).
NAGASE, T., SANADA, H., TAKEHARA, K., OE, M., IIZAKA, S., OHASHI, Y., OBA, M., KADOWAKI, T., & NAKAGAMI, G., 2011. Variations of plantar thermographic patterns in normal controls and non-ulcer diabetic patients: Novel classification using angiosome concept, Journal of Plastic, Reconstructive and Aesthetic Surgery, 64(7), 860–866.
RING, F., 2010. Thermal Imaging Today and Its Relevance to Diabetes, In J Diabetes Sci Technol, Vol. 4, Issue 4.
RUDER, S., 2016. An overview of gradient descent optimization algorithms. http://arxiv.org/abs/1609.04747
SANDLER, M., HOWARD, A., ZHU, M., ZHMOGINOV, A., & CHEN, L.-C., 2018. MobileNetV2: Inverted Residuals and Linear Bottlenecks. http://arxiv.org/abs/1801.04381
SHAH, B., & BHAVSAR, H., 2022. Time Complexity in Deep Learning Models, Procedia Computer Science, 215, 202–210.
SIMONYAN, K., & ZISSERMAN, A., 2014. Very Deep Convolutional Networks for Large-Scale Image Recognition. http://arxiv.org/abs/1409.1556
SUYANTO., 2018. Machine Learning. Informatika Bandung.
SZEGEDY, C., LIU, W., JIA, Y., SERMANET, P., REED, S., ANGUELOV, D., ERHAN, D., VANHOUCKE, V., & RABINOVICH, A., 2014. Going Deeper with Convolutions. http://arxiv.org/abs/1409.4842
SZEGEDY, C., VANHOUCKE, V., IOFFE, S., & SHLENS, J., 2015. Rethinking the Inception Architecture for Computer Vision.
VERMA, A., MEENPAL, T., & ACHARYA, B., 2022. Computational Cost Reduction of Convolution Neural Networks by Insignificant Filter Removal, In ROMANIAN JOURNAL OF INFORMATION SCIENCE AND TECHNOLOGY, Vol. 25, Issue 2.
Unduhan
Diterbitkan
Terbitan
Bagian
Lisensi
Hak Cipta (c) 2023 Jurnal Teknologi Informasi dan Ilmu Komputer
Artikel ini berlisensiCreative Commons Attribution-ShareAlike 4.0 International License.
Artikel ini berlisensi Creative Common Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
Penulis yang menerbitkan di jurnal ini menyetujui ketentuan berikut:
- Penulis menyimpan hak cipta dan memberikan jurnal hak penerbitan pertama naskah secara simultan dengan lisensi di bawah Creative Common Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) yang mengizinkan orang lain untuk berbagi pekerjaan dengan sebuah pernyataan kepenulisan pekerjaan dan penerbitan awal di jurnal ini.
- Penulis bisa memasukkan ke dalam penyusunan kontraktual tambahan terpisah untuk distribusi non ekslusif versi kaya terbitan jurnal (contoh: mempostingnya ke repositori institusional atau menerbitkannya dalam sebuah buku), dengan pengakuan penerbitan awalnya di jurnal ini.
- Penulis diizinkan dan didorong untuk mem-posting karya mereka online (contoh: di repositori institusional atau di website mereka) sebelum dan selama proses penyerahan, karena dapat mengarahkan ke pertukaran produktif, seperti halnya sitiran yang lebih awal dan lebih hebat dari karya yang diterbitkan. (Lihat Efek Akses Terbuka).
