在本研究中我們針對眼底螢光照影與 NOTCH3 基因異常進行研究。眼底螢光照影是將螢光劑注入血管後在藉由攝影設備所拍攝出來的影像。 NOTCH3 基因異常的病人通常會比正常的人得到腦中風的機會還大，並且會影響到視網膜血管。我們的研究主要是透過從眼底影像中提取的特徵，並利用決策樹模型來判斷受測者是否存在 NOTCH3 基因異常。在研究的過程中我們先對眼底影像先進行文字清除、影像亮度的歸一化跟影像拼接等處理後，再從其影像中提取特徵，接著訓練出決策樹，並預測每張眼底影像的類別。最後我們得到一個精準的決策樹來判斷受試者是否有 NOTCH3 基因異常，其精準度為0.9286。

In this research, we investigated the fluorescein angiography (FA) of fundus images in relation to the NOTCH3 gene anomaly. FA images are obtained by an imaging technique that involves injecting a fluorescent dye into the blood vessels and capturing images with specialized cameras. The risk of stroke with the NOTCH3 gene anomaly of patients is higher than normal individuals, and the NOTCH3 gene anomaly affects the retinal blood vessels in patients. The main goal is to determine the NOTCH3 gene anomaly of patients by a decision tree that uses features extracted from fundus images. In the process, we first removed any text from images, performed brightness normalization, image stitching for fundus images, and extracted features from fundus images. Subsequently, a decision tree was trained using these features to predict the class of each fundus image. Finally, we obtained the accurate decision tree, and the accuracy is 0.9286.

[1] H. Bay, A. Ess, T. Tuytelaars, and L. Van Gool, Speeded-up robust features (surf), Computer Vision and Image Understanding, 110 (2008), pp. 346–359. Similarity Matching in Computer Vision and Multimedia.
[2] L. Breiman, J. H. Friedman, R. A. Olshen, and C. J. Stone, Classification and regression trees, 1984.
[3] R. Cumurciuc, P. Massin, M. Pâques, V. Krisovic, A. Gaudric, M. G. Bousser, and H. Chabriat, Retinal abnormalities in cadasil: a retrospective study of 18 patients, Journal of Neurology, Neurosurgery & Psychiatry, 75 (2004), pp. 1058 – 1060.
[4] M. Foracchia, E. Grisan, and A. Ruggeri, Luminosity and contrast normalization in retinal images, Medical Image Analysis, 9 (2005), pp. 179–190.
[5] E. D. Goldstein, A. Shakoor, and J. J. Majersik, Branch retinal vein occlusion and venous abnormalities in cadasil, The Neurologist, 25 (2020), pp. 178 – 179.
[6] G. D. Joshi and J. Sivaswamy, Colour retinal image enhancement based on domain knowledge, in 2008 Sixth Indian Conference on Computer Vision, Graphics and Image Processing, 2008, pp. 591–598.
[7] C. K. Lam, D. Yi, M. Guo, and T. Lindsey, Automated detection of diabetic retinopathy using deep learning, AMIA Summits on Translational Science Proceedings, 2018 (2018), pp. 147 – 155.
[8] T. Lindeberg, Scale invariant feature transform, Scholarpedia, 7 (2012), p. 10491.
[9] D. Lowe, Object recognition from local scale-invariant features, in Proceedings of the Seventh IEEE International Conference on Computer Vision, vol. 2, 1999, pp. 1150– 1157 vol.2.
[10] D. G. Lowe, Distinctive image features from scale-invariant keypoints, International Journal of Computer Vision, 60 (2004), pp. 91–110.
[11] P. Nelis, I. Kleffner, M. C. Burg, C. R. Clemens, M. Alnawaiseh, J. Motte, M. Marziniak, N. Eter, and F. Alten, Oct-angiography reveals reduced vessel density in the deep retinal plexus of cadasil patients, Scientific Reports, 8 (2018).
[12] N. Otsu, A threshold selection method from gray-level histograms, IEEE Transactions on Systems, Man, and Cybernetics, 9 (1979), pp. 62–66.
[13] W. F. Robinson, S. L. Galetta, L. F. McCluskey, M. S. Forman, and L. J. Balcer, Retinal findings in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (cadasil)., Survey of ophthalmology, 455 (2001), pp. 445–8.
[14] D. Schoemaker and J. F. Arboleda-Velasquez, Notch3 signaling and aggregation as targets for the treatment of cadasil and other notch3-associated small-vessel diseases, The American Journal of Pathology, 191 (2021), pp. 1856–1870.
[15] J. Zaragoza, T.-J. Chin, M. S. Brown, and D. Suter, As-projective-as-possible image stitching with moving dlt, in 2013 IEEE Conference on Computer Vision and Pattern Recognition, 2013, pp. 2339–2346.
[16] K. Zuiderveld, Contrast Limited Adaptive Histogram Equalization, Academic Press Professional, Inc., USA, 1994, p. 474–485.