Knowledge transfer for melanoma screening with deep learning

Menegola, Afonso; Fornaciali, Michel; Pires, Ramon; Bittencourt, Flávia Vasques; Avila, Sandra; Valle, Eduardo

doi:10.1109/isbi.2017.7950523

Cited by 164 publications

(114 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Nonetheless, for medical tasks, obtaining a large amount of data to train a CNN is quite challenging. To overcome this issue, all these works used transfer learning, a well-known technique where a model trained for a given source task is partially reused for a new target task [47]. Thereby, the models were initialized using the weights from the ImageNet dataset [49] and then fine-tuned using their own dataset.…”

Section: Convolutional Neural Networkmentioning

confidence: 99%

The impact of patient clinical information on automated skin cancer detection

Pacheco

Krohling

2020

Computers in Biology and Medicine

155

View full text Add to dashboard Cite

Skin cancer is one of the most common types of cancer around the world. For this reason, over the past years, different approaches have been proposed to assist detect it. Nonetheless, most of them are based only on dermoscopy images and do not take into account the patient clinical information. In this work, first, we present a new dataset that contains clinical images, acquired from smartphones, and patient clinical information of the skin lesions. Next, we introduce a straightforward approach to combine the clinical data and the images using different well-known deep learning models. These models are applied to the presented dataset using only the images and combining them with the patient clinical information. We present a comprehensive study to show the impact of the clinical data on the final predictions. The results obtained by combining both sets of information show a general improvement of around 7% in the balanced accuracy for all models. In addition, the statistical test indicates significant differences between the models with and without considering both data. The improvement achieved shows the potential of using patient clinical information in skin cancer detection and indicates that this piece of information is important to leverage skin cancer detection systems.

show abstract

Section: Convolutional Neural Networkmentioning

confidence: 99%

The impact of patient clinical information on automated skin cancer detection

Pacheco

Krohling

2020

Computers in Biology and Medicine

155

View full text Add to dashboard Cite

show abstract

“…Table 5 and Figure 7 show a comparison between the proposed method and other representative methods that used ISBI 2016 dataset in terms of the ACC, SEN, SPE, DIC, and JAC. [34] 91.10 95.70 94.9 87.7 82.9 Codella et al [35] 69.30 83.60 80.70 Menegola et al [25] 47.60 88.10 79.20 Vasconcelos et al [36] 74.60 84.50 82.50 Oliveira et al [27] 91.80 96.70 27.70 Figure 7 and Table 5 above show that the best-segmented result from ISBI 2016 images was that of the proposed method by better delineating the border region of the lesion and better performance parameters. Also, it can be observed that the closer best-segmented result was that of Oliveira et al [27] in terms of SEN and SPE.…”

Section: Resultsmentioning

confidence: 99%

“…In Reference [24], new boundary features are presented based on the color variation of the pigmented lesion images to influence higher performance in skin tumor diagnosis with the multilayer perceptron neural network. Menegola et al [25] used knowledge transfer learning for skin tumors lesion screening with deep learning to reach high confidence, whereas for difficult lesions performance is still little better than chance. Xie et al [26] developed a new algorithm with a self-generating neural network (SGNN) and features descriptive of the tumor color to get high accuracy with the classifier model for melanomas and benign case.…”

Section: Introductionmentioning

confidence: 99%

Skin Lesion Segmentation Method for Dermoscopy Images Using Artificial Bee Colony Algorithm

et al. 2018

View full text Add to dashboard Cite

Abstract:The occurrence rates of melanoma are rising rapidly, which are resulting in higher death rates. However, if the melanoma is diagnosed in Phase I, the survival rates increase. The segmentation of the melanoma is one of the largest tasks to undertake and achieve when considering both beneath and over the segmentation. In this work, a new approach based on the artificial bee colony (ABC) algorithm is proposed for the detection of melanoma from digital images. This method is simple, fast, flexible, and requires fewer parameters compared with other algorithms. The proposed approach is applied on the PH2, ISBI 2016 challenge, the ISBI 2017 challenge, and Dermis datasets. These bases contained images are affected by different abnormalities. The formation of the databases consists of images collected from different sources; they are bases with different types of resolution, lighting, etc., so in the first step, the noise was removed from the images by using morphological filtering. In the next step, the ABC algorithm is used to find the optimum threshold value for the melanoma detection. The proposed approach achieved good results in the conditions of high specificity. The experimental results suggest that the proposed method accomplished higher performance compared to the ground truth images supported by a Dermatologist. For the melanoma detection, the method achieved an average accuracy and Jaccard's coefficient in the range of 95. 24-97.61%, and 83.56-85.25% in these four databases. To show the robustness of this work, the results were compared to existing methods in the literature for melanoma detection. High values for estimation performance confirmed that the proposed melanoma detection is better than other algorithms, which demonstrates the highly differential power of the newly introduced features.

show abstract

“…While the latter one utilizes a pre-trained source DNN to initialize some of the weights of the target DNN, and training it as usual. 6 The problem of how to use the "off-the-shelf" features more effectively for specific target task induces many feature selection techniques 7 (filter, wrapper and embeddedbased methods, etc.). For the fine-tuning tactics, because of the dataset bias between source and target task, adapting a state-of-the-art deep model on a new domain still requires a significant amount of data, which is simply not available for many biomedical image analysis tasks.…”

Section: Introductionmentioning

confidence: 99%

“…12 Recently years, multisource transfer learning approach has been widely adopted for many medical imaging analysis tasks. 6,11,13,14 Its main advantage lies in the fact that the knowledge transferred from multiple databases (especially texture databases) should be more generalized than simply from natural images to medical imaging domain. And transferring from multisource datasets with different modalities can be viewed as regularization for the target task model to reduce the over-fitting phenomenon, which is suitable for small-sized medical imaging datasets.…”

Section: Introductionmentioning

confidence: 99%

Lymph node detection method based on multisource transfer learning and convolutional neural network

Peng

2019

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Recently years, convolutional neural networks (CNNs) have proven to be powerful tools for a broad range of computer vision tasks. However, training a CNN from scratch is difficult because it requires a large amount of labeled training data, which remains a challenge in medical imaging domain. To this end, deep transfer learning (TL) technique is widely used for many medical image tasks. In this paper, we propose a novel multisource transfer learning CNN model for lymph node detection.The mechanism behind it is straightforward. Point-wise (1 × 1) convolution is used to fuse multisource transfer learning knowledge. Concretely, we view the transferred features as priori domain knowledge and 1 × 1 convolutional operation is implemented after pre-trained convolution layers to adaptively combine the transfer information for target task. In order to learn non-linear transferred features and prevent over-fitting, we present an encode process for the pre-trained convolution kernels. At last, based on convolutional factorization technique, we train the proposed CNN model and the encoder process jointly, which improves the feasibility of our approach. The effectiveness of the proposed method is verified on lymph node (LN) dataset: 388 mediastinal LNs labeled by radiologists in 90 patient CT scans, and 595 abdominal LNs in 86 patient CT scans for LN detection. Our method demonstrates sensitivities of about 85%/71% at 3 FP/vol. and 92%/85% at 6 FP/vol. for mediastinum and abdomen respectively, which compares favorably to previous methods. K E Y W O R D S convolutional neural network, lymph node detection, multisource transfer learning, point-wise convolutional operation

show abstract

Knowledge transfer for melanoma screening with deep learning

Cited by 164 publications

References 16 publications

The impact of patient clinical information on automated skin cancer detection

The impact of patient clinical information on automated skin cancer detection

Skin Lesion Segmentation Method for Dermoscopy Images Using Artificial Bee Colony Algorithm

Lymph node detection method based on multisource transfer learning and convolutional neural network

Contact Info

Product

Resources

About