本研究改進了2019 年Campanella. 等人發表於Nature Medicine 的MIL_RNN
[1] 方法，並提出了兩種預測子宮內膜染色全玻片影像微衛星狀態(Microsatellite
status) 的模型：Proposed method 1 與Proposed method 2。為確保提出模型的適應性
與可訓練性，我們使用了三個不同的數據集進行訓練與驗證。數據集一包含H&E
染色的子宮內膜微陣列（Tissue microarray, TMA）全玻片影像（WSIs），數據集二
包含H&E 染色的子宮內膜癌G1G2 級別的WSIs，數據集三則包含H&E 染色的子
宮內膜癌G3 級別的WSIs。
• 在數據集一中，Proposed method 1 與Proposed method 2 在準確率(Accuracy)、
精確率(Precision)、召回率(Recall) 和F1-分數(F1-score) 等指標上均優於原
始的MIL_RNN 方法。也將數據集一分別測試在七種不同模型，結果顯示
Proposed method 1 在準確率、召回率和F1-分數上表現最好。。
• 在數據集二和數據集三中，我們將在數據集一中表現前兩名的弱監督學
習方法Proposed method 1 和CLAM [2] 與原始的MIL_RNN 模型進行比較。
在數據集二中Proposed method 1 與CLAM 的F1 分數最高，在數據集三中
Proposed method 1 的F1-分數為最高。
• 最後，根據這三組數據集的結果，我們推論多實例學習模型的F1-分數與訓
練集數據量成正比，而Proposed method 1 的特異度(Specificity) 與訓練集中
的類別數量差異成反比。
綜上所述，本論文提出了兩種預測子宮內膜染色全玻片影像微衛星狀態的模
型，這些模型在性能上優於原始的MIL_RNN 方法。它們在不同的數據集上表現
出色，並為模型性能與訓練集特徵之間的關係提供了洞見。

This study improves upon the MIL_RNN method proposed by Campanella et al.
in Nature Medicine in 2019 [1] and proposes two models, Proposed method 1 and Proposed
method 2, for predicting the microsatellite status of endometrial cancer from wholeslide
images stained with hematoxylin and eosin (H&E). To ensure the adaptability and
trainability of the models, three different datasets were used for training and validation.
Dataset 1 consisted of H&E-stained tissue microarray (TMA) whole-slide images (WSIs)
of endometrial samples, Dataset 2 included H&E-stained WSIs of endometrial carcinoma
G1G2, and Dataset 3 comprised H&E-stained WSIs of endometrial carcinoma G3.
• In Dataset 1, both Proposed method 1 and Proposed method 2 exhibited superior
performance in terms of accuracy, precision, recall, and F1-score compared to the
original MIL_RNN method. Also, Dataset 1 was tested on seven different models
individually, and the results showed that Proposed Method 1 performed the best in
terms of accuracy, recall, and F1 score.
• In Dataset 2 and Dataset 3, Proposed method 1 and the weakly supervised learning
method CLAM [2], both ranking in the top two in Dataset 1, were compared with
the original MIL_RNN model. Proposed method 1 and CLAM achieved the highest
F1-scores in Dataset 2, while Proposed method 1 attained the highest F1-score in
Dataset 3.
• Lastly, based on the results from these three datasets, we inferred that the F1-score
of the multiple instance learning model is directly proportional to the amount of
training data, and the specificity of Proposed method 1 is inversely proportional to
the difference in class quantities in the training set.
In conclusion, this study proposes two models, Proposed method 1 and Proposed method
2, for predicting the microsatellite status of endometrial cancer from H&E-stained wholeslide
images, which outperform the original MIL_RNN method in terms of performance.
These models demonstrate excellent performance across different datasets and provide insights
into the relationship between model performance and training set characteristics.

[1] G. Campanella, M. G. Hanna, L. Geneslaw, A. Miraflor, V. Werneck Krauss Silva,
K. J. Busam, E. Brogi, V. E. Reuter, D. S. Klimstra, and T. J. Fuchs, “Clinical-grade
computational pathology using weakly supervised deep learning on whole slide images,”
Nature medicine, vol. 25, no. 8, pp. 1301–1309, 2019.
[2] M. Y. Lu, D. F. Williamson, T. Y. Chen, R. J. Chen, M. Barbieri, and F. Mahmood,
“Data-efficient and weakly supervised computational pathology on whole-slide images,”
Nature biomedical engineering, vol. 5, no. 6, pp. 555–570, 2021.
[3] O. Raglan, I. Kalliala, G. Markozannes, S. Cividini, M. J. Gunter, J. Nautiyal,
H. Gabra, E. Paraskevaidis, P. Martin-Hirsch, K. K. Tsilidis, et al., “Risk factors
for endometrial cancer: An umbrella review of the literature,” International journal
of cancer, vol. 145, no. 7, pp. 1719–1730, 2019.
[4] S. F. Lax, E. S. Pizer, B. M. Ronnett, and R. J. Kurman, “Comparison of estrogen
and progesterone receptor, ki-67, and p53 immunoreactivity in uterine endometrioid
carcinoma and endometrioid carcinoma with squamous, mucinous, secretory, and
ciliated cell differentiation,” Human pathology, vol. 29, no. 9, pp. 924–931, 1998.
[5] J. V. Bokhman, “Two pathogenetic types of endometrial carcinoma,” Gynecologic
oncology, vol. 15, no. 1, pp. 10–17, 1983.
[6] A. K. Green, J. Feinberg, and V. Makker, “A review of immune checkpoint blockade
therapy in endometrial cancer,” American Society of Clinical Oncology Educational
Book, vol. 40, pp. 238–244, 2020.
[7] G. F. Fleming, “Second-line therapy for endometrial cancer: the need for better options,”
Obstetrical & Gynecological Survey, vol. 71, no. 7, pp. 406–408, 2016.
[8] M. Vicky, H. MacKay, I. Ray-Coquard, D. A. Levine, S. N. Westin, A. Daisuke, and
O. Ana, “Endometrial cancer (primer),” Nature Reviews: Disease Primers, vol. 7,
no. 1, 2021.
[9] A. de la Chapelle and H. Hampel, “Clinical relevance of microsatellite instability in
colorectal cancer,” Journal of Clinical Oncology, vol. 28, no. 20, p. 3380, 2010.
[10] S. J. Salipante, S. M. Scroggins, H. L. Hampel, E. H. Turner, and C. C. Pritchard,
“Microsatellite instability detection by next generation sequencing,” Clinical chemistry,
vol. 60, no. 9, pp. 1192–1199, 2014.
[11] G. Shamai, A. Livne, A. Polónia, E. Sabo, A. Cretu, G. Bar-Sela, and R. Kimmel,
“Deep learning-based image analysis predicts pd-l1 status from h&e-stained
histopathology images in breast cancer,” Nature Communications, vol. 13, no. 1,
p. 6753, 2022.
[12] N. Ma, X. Zhang, H.-T. Zheng, and J. Sun, “Shufflenet v2: Practical guidelines for
efficient cnn architecture design,” in Proceedings of the European conference on
computer vision (ECCV), pp. 116–131, 2018.
[13] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,”
in Proceedings of the IEEE conference on computer vision and pattern recognition,
pp. 770–778, 2016.
[14] N. Coudray, P. S. Ocampo, T. Sakellaropoulos, N. Narula, M. Snuderl, D. Fenyö,
A. L. Moreira, N. Razavian, and A. Tsirigos, “Classification and mutation prediction
from non–small cell lung cancer histopathology images using deep learning,” Nature
Medicine, vol. 24, pp. 1559–1567, Sept. 2018.
[15] Y. Tolkach, T. Dohmgörgen, M. Toma, and G. Kristiansen, “High-accuracy prostate
cancer pathology using deep learning,” Nature Machine Intelligence, vol. 2, no. 7,
pp. 411–418, 2020.
[16] C.-W. Wang, Y.-C. Lee, C.-C. Chang, Y.-J. Lin, Y.-A. Liou, P.-C. Hsu, C.-C. Chang,
A.-K.-O. Sai, C.-H. Wang, and T.-K. Chao, “A weakly supervised deep learning
method for guiding ovarian cancer treatment and identifying an effective biomarker,”
Cancers, vol. 14, no. 7, p. 1651, 2022.
[17] Z.-H. Zhou, “A brief introduction to weakly supervised learning,” National science
review, vol. 5, no. 1, pp. 44–53, 2018.
[18] B. Settles, “Active learning literature survey,” 2009.
[19] O. Chapelle, B. Scholkopf, and A. Zien, “Semi-supervised learning (chapelle, o. et
al., eds.; 2006)[book reviews],” IEEE Transactions on Neural Networks, vol. 20,
no. 3, pp. 542–542, 2009.
[20] Z.-H. Zhou and M. Li, “Semi-supervised learning by disagreement,” Knowledge and
Information Systems, vol. 24, pp. 415–439, 2010.
[21] D. D. Lewis, “A sequential algorithm for training text classifiers: Corrigendum and
additional data,” in Acm Sigir Forum, vol. 29, pp. 13–19, ACM New York, NY, USA,
1995.
[22] H. S. Seung, M. Opper, and H. Sompolinsky, “Query by committee,” in Proceedings
of the fifth annual workshop on Computational learning theory, pp. 287–294, 1992.
[23] N. Abe, “Query learning strategies using boosting and bagging,” in International
Conference on Machine Learning, 1998, pp. 1–9, 1998.
[24] H. T. Nguyen and A. Smeulders, “Active learning using pre-clustering,” in Proceedings
of the twenty-first international conference on Machine learning, p. 79, 2004.
[25] S. Dasgupta and D. Hsu, “Hierarchical sampling for active learning,” in Proceedings
of the 25th international conference on Machine learning, pp. 208–215, 2008.
[26] D. J. Miller and H. Uyar, “A mixture of experts classifier with learning based on both
labelled and unlabelled data,” Advances in neural information processing systems,
vol. 9, 1996.
[27] A. Blum and S. Chawla, “Learning from labeled and unlabeled data using graph
mincuts,” 2001.
[28] O. Chapelle and A. Zien, “Semi-supervised classification by low density separation,”
in International workshop on artificial intelligence and statistics, pp. 57–64, PMLR,
2005.
[29] A. Blum and T. Mitchell, “Combining labeled and unlabeled data with co-training,”
in Proceedings of the eleventh annual conference on Computational learning theory,
pp. 92–100, 1998.
[30] T. G. Dietterich, R. H. Lathrop, and T. Lozano-Pérez, “Solving the multiple instance
problem with axis-parallel rectangles,” Artificial intelligence, vol. 89, no. 1-2,
pp. 31–71, 1997.
[31] J. Foulds and E. Frank, “A review of multi-instance learning assumptions,” The
knowledge engineering review, vol. 25, no. 1, pp. 1–25, 2010.
[32] Z.-H. Zhou, “Multi-instance learning from supervised view,” Journal of Computer
Science and Technology, vol. 21, no. 5, pp. 800–809, 2006.
[33] Z.-H. Zhou and M.-L. Zhang, “Solving multi-instance problems with classifier
ensemble based on constructive clustering,” Knowledge and information systems,
vol. 11, pp. 155–170, 2007.
[34] J. Amores, “Multiple instance classification: Review, taxonomy and comparative
study,” Artificial intelligence, vol. 201, pp. 81–105, 2013.
[35] B. Frénay and M. Verleysen, “Classification in the presence of label noise: a survey,”
IEEE transactions on neural networks and learning systems, vol. 25, no. 5, pp. 845–
869, 2013.
[36] F. Muhlenbach, S. Lallich, and D. A. Zighed, “Identifying and handling mislabelled
instances,” Journal of Intelligent Information Systems, vol. 22, no. 1, pp. 89–109,
2004.
[37] Y. Bengio, A. Courville, and P. Vincent, “Representation learning: A review and
new perspectives,” IEEE transactions on pattern analysis and machine intelligence,
vol. 35, no. 8, pp. 1798–1828, 2013.
[38] K. He, X. Zhang, S. Ren, and J. Sun, “Delving deep into rectifiers: Surpassing
human-level performance on imagenet classification,” in Proceedings of the IEEE
international conference on computer vision, pp. 1026–1034, 2015.
[39] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep
convolutional neural networks,” Communications of the ACM, vol. 60, no. 6, pp. 84–
90, 2017.
[40] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied
to document recognition,” Proceedings of the IEEE, vol. 86, no. 11, pp. 2278–2324,
1998.
[41] G. Litjens, T. Kooi, B. E. Bejnordi, A. A. A. Setio, F. Ciompi, M. Ghafoorian, J. A.
Van Der Laak, B. Van Ginneken, and C. I. Sánchez, “A survey on deep learning in
medical image analysis,” Medical image analysis, vol. 42, pp. 60–88, 2017.
[42] K. Fukushima, “Neocognitron: A self-organizing neural network model for a mechanism
of pattern recognition unaffected by shift in position,” Biological cybernetics,
vol. 36, no. 4, pp. 193–202, 1980.
[43] Y. LeCun, Y. Bengio, et al., “Convolutional networks for images, speech, and time
series,” The handbook of brain theory and neural networks, vol. 3361, no. 10,
p. 1995, 1995.
[44] G. E. Hinton, A. Krizhevsky, and I. Sutskever, “Imagenet classification with deep
convolutional neural networks,” Advances in neural information processing systems,
vol. 25, no. 1106-1114, p. 1, 2012.
[45] D. Shen, G. Wu, and H.-I. Suk, “Deep learning in medical image analysis,” Annual
review of biomedical engineering, vol. 19, pp. 221–248, 2017.
[46] L. Hou, D. Samaras, T. M. Kurc, Y. Gao, J. E. Davis, and J. H. Saltz, “Patchbased
convolutional neural network for whole slide tissue image classification,” in
Proceedings of the IEEE conference on computer vision and pattern recognition,
pp. 2424–2433, 2016.
[47] N. Coudray, P. S. Ocampo, T. Sakellaropoulos, N. Narula, M. Snuderl, D. Fenyö,
A. L. Moreira, N. Razavian, and A. Tsirigos, “Classification and mutation prediction
from non–small cell lung cancer histopathology images using deep learning,” Nature
medicine, vol. 24, no. 10, pp. 1559–1567, 2018.
[48] O. Maron and T. Lozano-Pérez, “A framework for multiple-instance learning,” Advances
in neural information processing systems, vol. 10, 1997.
[49] S. Andrews, I. Tsochantaridis, and T. Hofmann, “Support vector machines for
multiple-instance learning,” Advances in neural information processing systems,
vol. 15, 2002.
[50] Y. Xu, T. Mo, Q. Feng, P. Zhong, M. Lai, I. Eric, and C. Chang, “Deep learning of
feature representation with multiple instance learning for medical image analysis,”
in 2014 IEEE international conference on acoustics, speech and signal processing
(ICASSP), pp. 1626–1630, IEEE, 2014.
[51] Y. Xu, J.-Y. Zhu, I. Eric, C. Chang, M. Lai, and Z. Tu, “Weakly supervised
histopathology cancer image segmentation and classification,” Medical image analysis,
vol. 18, no. 3, pp. 591–604, 2014.
[52] Y. Xu, J.-Y. Zhu, E. Chang, and Z. Tu, “Multiple clustered instance learning for
histopathology cancer image classification, segmentation and clustering,” in 2012
IEEE Conference on Computer Vision and Pattern Recognition, pp. 964–971, IEEE,
2012.
[53] E. Cosatto, P.-F. Laquerre, C. Malon, H.-P. Graf, A. Saito, T. Kiyuna, A. Marugame,
and K. Kamijo, “Automated gastric cancer diagnosis on h&e-stained sections; ltraining
a classifier on a large scale with multiple instance machine learning,” in Medical
Imaging 2013: Digital Pathology, vol. 8676, pp. 51–59, SPIE, 2013.
[54] M. M. Dundar, S. Badve, V. C. Raykar, R. K. Jain, O. Sertel, and M. N. Gurcan,
“A multiple instance learning approach toward optimal classification of pathology
slides,” in 2010 20th International Conference on Pattern Recognition, pp. 2732–
2735, IEEE, 2010.
[55] B. Li, Y. Li, and K. W. Eliceiri, “Dual-stream multiple instance learning network for
whole slide image classification with self-supervised contrastive learning,” in Proceedings
of the IEEE/CVF conference on computer vision and pattern recognition,
pp. 14318–14328, 2021.
[56] J. Feng and Z.-H. Zhou, “Deep miml network,” in Proceedings of the AAAI conference
on artificial intelligence, vol. 31, 2017.
[57] Z.-H. Zhou and M.-L. Zhang, “Neural networks for multi-instance learning,” in Proceedings
of the International Conference on Intelligent Information Technology, Beijing,
China, pp. 455–459, Citeseer, 2002.
[58] C. H. Li, I. Gondra, and L. Liu, “An efficient parallel neural network-based multiinstance
learning algorithm,” The Journal of Supercomputing, vol. 62, pp. 724–740,
2012.
[59] Z. Chen, Z. Chi, H. Fu, and D. Feng, “Multi-instance multi-label image classification:
A neural approach,” Neurocomputing, vol. 99, pp. 298–306, 2013.
[60] O. Z. Kraus, J. L. Ba, and B. J. Frey, “Classifying and segmenting microscopy images
with deep multiple instance learning,” Bioinformatics, vol. 32, no. 12, pp. i52–i59,
2016.
[61] C. Zhang, J. Platt, and P. Viola, “Multiple instance boosting for object detection,”
Advances in neural information processing systems, vol. 18, 2005.
[62] J. Ramon and L. De Raedt, “Multi instance neural networks,” in Proceedings of the
ICML-2000 workshop on attribute-value and relational learning, pp. 53–60, 2000.
[63] J. Keeler, D. Rumelhart, and W. Leow, “Integrated segmentation and recognition of
hand-printed numerals,” Advances in neural information processing systems, vol. 3,
1990.
[64] M. Ilse, J. Tomczak, and M. Welling, “Attention-based deep multiple instance learning,”
in International conference on machine learning, pp. 2127–2136, PMLR, 2018.
[65] Y. Schirris, E. Gavves, I. Nederlof, H. M. Horlings, and J. Teuwen, “Deepsmile:
Contrastive self-supervised pre-training benefits msi and hrd classification directly
from h&e whole-slide images in colorectal and breast cancer,” Medical Image Analysis,
vol. 79, p. 102464, 2022.
[66] S. Xie, R. Girshick, P. Dollár, Z. Tu, and K. He, “Aggregated residual transformations
for deep neural networks,” in Proceedings of the IEEE conference on computer vision
and pattern recognition, pp. 1492–1500, 2017.
[67] X. Zhang, X. Zhou, M. Lin, and J. Sun, “Shufflenet: An extremely efficient convolutional
neural network for mobile devices,” in Proceedings of the IEEE conference
on computer vision and pattern recognition, pp. 6848–6856, 2018.
[68] H. Salehinejad, S. Sankar, J. Barfett, E. Colak, and S. Valaee, “Recent advances in
recurrent neural networks,” arXiv preprint arXiv:1801.01078, 2017.
[69] Y. Bengio, P. Simard, and P. Frasconi, “Learning long-term dependencies with gradient
descent is difficult,” IEEE transactions on neural networks, vol. 5, no. 2, pp. 157–
166, 1994.
[70] I. Sutskever, J. Martens, and G. E. Hinton, “Generating text with recurrent neural
networks,” in Proceedings of the 28th international conference on machine learning
(ICML-11), pp. 1017–1024, 2011.
[71] T. Mikolov, A. Joulin, S. Chopra, M. Mathieu, and M. Ranzato, “Learning longer
memory in recurrent neural networks,” arXiv preprint arXiv:1412.7753, 2014.
[72] Z. Cai and N. Vasconcelos, “Cascade r-cnn: high quality object detection and instance
segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence,
2019.
[73] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,”
in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),
pp. 770–778, June 2016.
[74] N. Otsu, “A threshold selection method from gray-level histograms,” IEEE transactions
on systems, man, and cybernetics, vol. 9, no. 1, pp. 62–66, 1979.
[75] N. Qian, “On the momentum term in gradient descent learning algorithms,” Neural
networks, vol. 12, no. 1, pp. 145–151, 1999.
[76] R. Murali, R. N. Grisham, and R. A. Soslow, “The roles of pathology in targeted
therapy of women with gynecologic cancers,” Gynecologic oncology, vol. 148, no. 1,
pp. 213–221, 2018.
[77] K. S. Tewari, M. W. Sill, H. J. Long III, R. T. Penson, H. Huang, L. M. Ramondetta,
L. M. Landrum, A. Oaknin, T. J. Reid, M. M. Leitao, et al., “Improved survival
with bevacizumab in advanced cervical cancer,” New England Journal of Medicine,
vol. 370, no. 8, pp. 734–743, 2014.
[78] D. T. Le, J. N. Uram, H. Wang, B. R. Bartlett, H. Kemberling, A. D. Eyring, A. D.
Skora, B. S. Luber, N. S. Azad, D. Laheru, et al., “Pd-1 blockade in tumors with
mismatch-repair deficiency,” New England Journal of Medicine, vol. 372, no. 26,
pp. 2509–2520, 2015.
[79] E. V. Connor and P. G. Rose, “Management strategies for recurrent endometrial cancer,”
Expert Review of Anticancer Therapy, vol. 18, no. 9, pp. 873–885, 2018.
[80] W. Cao, X. Ma, J. V. Fischer, C. Sun, B. Kong, and Q. Zhang, “Immunotherapy
in endometrial cancer: rationale, practice and perspectives,” Biomarker Research,
vol. 9, no. 1, pp. 1–30, 2021.
[81] R. H. Andtbacka, H. L. Kaufman, F. Collichio, T. Amatruda, N. Senzer, J. Chesney,
K. A. Delman, L. E. Spitler, I. Puzanov, S. S. Agarwala, et al., “Talimogene laherparepvec
improves durable response rate in patients with advanced melanoma,”
Journal of clinical oncology, vol. 33, no. 25, pp. 2780–2788, 2015.
[82] R. K. Gill, S. Saksena, S. Tyagi, W. A. Alrefai, J. Malakooti, Z. Sarwar, J. R. Turner,
K. Ramaswamy, and P. K. Dudeja, “Serotonin inhibits na+/h+ exchange activity via
5-ht4 receptors and activation of pkcα in human intestinal epithelial cells,” Gastroenterology,
vol. 128, no. 4, pp. 962–974, 2005.