Harits Abdurrohman, Robih Dini, Arief Purnama Muharram
Sari
The HAM10000 Dataset merupakan koleksi besar citra dermatoskopi untuk lesi kulit berpigmen yang umum. The HAM10000 Dataset terdiri atas 10.015 data citra lesi kulit berpigmen yang terbagi untuk penyakit Bowen, karsinoma sel basal, benign keratosis-like lesion, dermatofibroma, melanoma, melanocytic nevi, dan lesi vaskular. Data citra yang terdapat dalam dataset telah terkonfirmasi baik melalui histopatologi, pemeriksaan follow-up, konsensus pakar, maupun konfirmasi melalui in-vivo confocal microscopy. Pada penelitian ini kami melakukan pengujian performa terhadap model deep learning dan melakukan evaluasi. Tahap pre-processing citra meliputi analisis distribusi citra pada setiap kelas lesi, pengelompokan ulang kelas lesi berdasarkan letak pada bagian tubuh, dan augmentasi citra. Oleh karena keterbatasan data citra setelah dilakukan analisis distribusi maka model yang dibangun pada penelitian ini hanya berfokus pada kelas lesi untuk abdomen, punggung, ekstremitas atas dan bawah. Evaluasi ini dilakukan terhadap beberapa metode yang terkenal InceptionV3, MobileNet dan MobileNetV2. Ukuran performa yang dilakukan meliputi analisis confusion matrix yakni dengan mengambil nilai precision dan recall, dan f1-score.
Kata Kunci
Deep Learning, Performance, The HAM10000 Dataset, Skin Lesion.
Teks Lengkap
Referensi
Binder, M. et al., 1994. Application of an artificial neural network in epiluminescence microscopy pattern analysis of pigmented skin lesions: a pilot study. British Journal of Dermatology, 4(130), pp. 460-465.
Dechter, R., 1986. Learning While Searching in Constraint-Satisfaction-Problems. University of California, Computer Science Department, Cognitive Systems Laboratory.
He, K., Zhang, X., Ren, S. & Sun, J., 2016. Deep residual learning for image recognition. Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 770-778.
Howard, A. G. et al., 2017. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint .
Hussain, Z., Gimenez, F., Yi, D. & Rubin, D., 2017. Differential Data Augmentation Techniques for Medical Imaging Classification Tasks. AMIA Annual Symposium Proceedings, Volume 2017, p. 979.
Kotsiantis, S. B., Zaharakis, I. & Pintelas, P., 2007. Supervised machine learning: A review of classification techniques. Emerging artificial intelligence applications in computer engineering, Volume 160, pp. 3-24.
LeCun, Y., Haffner, P., Bottou, L. & Bengio, Y., 1999. Object recognition with gradient-based learning. Shape, contour and grouping in computer vision, pp. 319-345.
Rosendahl, C., Tschandl, P., Cameron, A. & Kittler, H., 2011. Diagnostic accuracy of dermatoscopy for melanocytic and nonmelanocytic pigmented lesions. J Am Acad Dermatol, Volume 64, p. 1068–1073.
Sandler, M. et al., 2018. MobileNetV2: Inverted Residuals and Linear Bottlenecks. IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4510-4520.
Sifre, L. & Mallat, S., 2014. Rigid-motion scattering for image classification. Doctoral dissertation, PhD thesis, Ph. D. thesis.
Stevenson, A. D., Mickan, S., Mallett, S. & Ayya, M., 2013. Systematic review of diagnostic accuracy of reflectance confocal microscopy. Dermatol Pract Concept, Volume 3, p. 19–27.
Szegedy, C. et al., 2016. Rethinking the inception architecture for computer vision. Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2818-2826.
Tschandl, P., Rosendahl, C. & Kittler, H., 2018. The HAM10000 Dataset: A Large Collection of Multi-Source Dermatoscopic Images of Common Pigmented Skin Lesions. arXiv preprint .
DOI