Deep Learning for Lung Cancer Detection: Tackling the Kaggle Data Science Bowl 2017 Challenge

Kuan, Kingsley; Ravaut, Mathieu; Manek, Gaurav; Chen, Huiling; Lin, Jie; Nazir, Babar; Chen, Cen; Howe, Tse Chiang; Zeng, Zeng; Chandrasekhar, Vijay

doi:10.48550/arxiv.1705.09435

Cited by 16 publications

(16 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(2) Semi-supervised GANs for image-resolution Training datasets: Market-1501 [155], Tibia Dataset [164], Abdominal Dataset [165], CUHK03 [166], MSMT17 [167], LUNA [168], Data Science Bowl 2017 (DSB) [169], UKDHP [170], SG [170] and UKBB [171].…”

Section: Modelsmentioning

confidence: 99%

Generative Adversarial Networks for Image Super-Resolution: A Survey

Tian¹,

Zhang²,

Lin³

et al. 2022

Preprint

View full text Add to dashboard Cite

Single image super-resolution (SISR) has played an important role in the field of image processing. Recent generative adversarial networks (GANs) can achieve excellent results on low-resolution images with small samples. However, there are little literatures summarizing different GANs in SISR. In this paper, we conduct a comparative study of GANs from different perspectives. We first take a look at developments of GANs. Second, we present popular architectures for GANs in big and small samples for image applications. Then, we analyze motivations, implementations and differences of GANs based optimization methods and discriminative learning for image super-resolution in terms of supervised, semi-supervised and unsupervised manners. Next, we compare performance of these popular GANs on public datasets via quantitative and qualitative analysis in SISR. Finally, we highlight challenges of GANs and potential research points for SISR.

show abstract

Section: Modelsmentioning

confidence: 99%

Generative Adversarial Networks for Image Super-Resolution: A Survey

Tian¹,

Zhang²,

Lin³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Classification problems refer to tasks aimed at identifying the category of a new observation with attributes only by using a model learned from a training data set with both attributes and known category for each instance (Alpaydin, 2020). For example, in the Kaggle Data Science Bowl 2017, CT scan images and expert-labeled lung cancer status were used to train a model for classifying new images into cancer versus no cancer (Kuan et al, 2017). To solicit solutions for classification problems, open contests are often adopted to source algorithms from a vast group of people (Hall, 2016;Levin, 2000).…”

Section: Literature Reviewmentioning

confidence: 99%

Three cobblers worth the mastermind? The potential of ensemble in crowdsourced classification problems

2021

View full text Add to dashboard Cite

Classification problems, where the objective is to identify the class labels of given data points, are most often the subject of open contests, in which solvers compete for awards offered by solution seekers. Extant literature in open contests has studied both the winner‐takes‐all and top‐K award schemes, in which the award is granted to the best one and the best K solutions, respectively. However, in comparing these two schemes, researchers have never considered that under a top‐K award scheme, seekers may often benefit from aggregating the solutions as an ensemble, which could achieve a performance that is superior to any individual solution. The practice of ensemble is very common especially in classification and other data science problems, and in this work, we are the first to model it with the scope of deriving the optimal award scheme in open contests. Our results formally show that, given the option of ensemble, a top‐K award scheme may have the advantage to grant a higher profit to the seeker than the winner‐takes‐all award scheme. Further, the optimal number of awards is positively affected by contest parameters including the solvers' expertise, the number of solvers, and the return on effort, whereas it is negatively affected by the technical uncertainty when such uncertainty is sufficiently large. These findings are consistent with the increasing adoption of the top‐K award scheme in contests held on Kaggle and other similar platforms.

show abstract

“…In the experiments, we use the LIDC-IDRI dataset [33,34] with the LUNA16's settings [35]. Specially, the CTs with slice thickness greater than 3mm, slice spacing inconsistent or missing slices, are removed from the LIDC-IDRI dataset.…”

Section: Datasetmentioning

confidence: 99%

Learning Efficient, Explainable and Discriminative Representations for Pulmonary Nodules Classification

Jiang

Shen

Gao

et al. 2021

Preprint

View full text Add to dashboard Cite

Automatic pulmonary nodules classification is significant for early diagnosis of lung cancers. Recently, deep learning techniques have enabled remarkable progress in this field. However, these deep models are typically of high computational complexity and work in a black-box manner. To combat these challenges, in this work, we aim to build an efficient and (partially) explainable classification model. Specially, we use neural architecture search (NAS) to automatically search 3D network architectures with excellent accuracy/speed trade-off. Besides, we use the convolutional block attention module (CBAM) in the networks, which helps us understand the reasoning process. During training, we use A-Softmax loss to learn angularly discriminative representations. In the inference stage, we employ an ensemble of diverse neural networks to improve the prediction accuracy and robustness. We conduct extensive experiments on the LIDC-IDRI database. Compared with previous state-of-the-art, our model shows highly comparable performance by using less than 1/40 parameters. Besides, empirical study shows that the reasoning process of learned networks is in conformity with physicians' diagnosis. Related code and results have been released at: https://github.com/ fei-hdu/NAS-Lung.

show abstract

Deep Learning for Lung Cancer Detection: Tackling the Kaggle Data Science Bowl 2017 Challenge

Cited by 16 publications

References 12 publications

Generative Adversarial Networks for Image Super-Resolution: A Survey

Generative Adversarial Networks for Image Super-Resolution: A Survey

Three cobblers worth the mastermind? The potential of ensemble in crowdsourced classification problems

Learning Efficient, Explainable and Discriminative Representations for Pulmonary Nodules Classification

Contact Info

Product

Resources

About