Research ExplorerPublicationsSkin Lesion Classification Using CNNs with Patch-Based Atten...

Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting

Standard

Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting. / Gessert, Nils; Sentker, Thilo; Madesta, Frederic; Schmitz, Rudiger; Kniep, Helge; Baltruschat, Ivo; Werner, Rene; Schlaefer, Alexander.

In: IEEE T BIO-MED ENG, Vol. 67, No. 2, 02.2020, p. 495-503.

Research output: SCORING: Contribution to journal › SCORING: Journal article › Research › peer-review

Harvard

Gessert, N, Sentker, T, Madesta, F, Schmitz, R, Kniep, H, Baltruschat, I, Werner, R & Schlaefer, A 2020, 'Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting', IEEE T BIO-MED ENG, vol. 67, no. 2, pp. 495-503. https://doi.org/10.1109/TBME.2019.2915839

APA

Gessert, N., Sentker, T., Madesta, F., Schmitz, R., Kniep, H., Baltruschat, I., Werner, R., & Schlaefer, A. (2020). Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting. IEEE T BIO-MED ENG, 67(2), 495-503. https://doi.org/10.1109/TBME.2019.2915839

Vancouver

Gessert N, Sentker T, Madesta F, Schmitz R, Kniep H, Baltruschat I et al. Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting. IEEE T BIO-MED ENG. 2020 Feb;67(2):495-503. https://doi.org/10.1109/TBME.2019.2915839

Bibtex

@article{27469a795fa444c18dfe108df03afcd8,
title = "Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting",
abstract = "OBJECTIVE: This paper addresses two key problems of skin lesion classification. The first problem is the effective use of high-resolution images with pretrained standard architectures for image classification. The second problem is the high-class imbalance encountered in real-world multi-class datasets.METHODS: To use high-resolution images, we propose a novel patch-based attention architecture that provides global context between small, high-resolution patches. We modify three pretrained architectures and study the performance of patch-based attention. To counter class imbalance problems, we compare oversampling, balanced batch sampling, and class-specific loss weighting. Additionally, we propose a novel diagnosis-guided loss weighting method that takes the method used for ground-truth annotation into account.RESULTS: Our patch-based attention mechanism outperforms previous methods and improves the mean sensitivity by [Formula: see text]. Class balancing significantly improves the mean sensitivity and we show that our diagnosis-guided loss weighting method improves the mean sensitivity by [Formula: see text] over normal loss balancing.CONCLUSION: The novel patch-based attention mechanism can be integrated into pretrained architectures and provides global context between local patches while outperforming other patch-based methods. Hence, pretrained architectures can be readily used with high-resolution images without downsampling. The new diagnosis-guided loss weighting method outperforms other methods and allows for effective training when facing class imbalance.SIGNIFICANCE: The proposed methods improve automatic skin lesion classification. They can be extended to other clinical applications where high-resolution image data and class imbalance are relevant.",
keywords = "Journal Article",
author = "Nils Gessert and Thilo Sentker and Frederic Madesta and Rudiger Schmitz and Helge Kniep and Ivo Baltruschat and Rene Werner and Alexander Schlaefer",
year = "2020",
month = feb,
doi = "10.1109/TBME.2019.2915839",
language = "English",
volume = "67",
pages = "495--503",
journal = "IEEE T BIO-MED ENG",
issn = "0018-9294",
publisher = "IEEE Computer Society",
number = "2",

}

RIS

TY - JOUR

T1 - Skin Lesion Classification Using CNNs with Patch-Based Attention and Diagnosis-Guided Loss Weighting

AU - Gessert, Nils

AU - Sentker, Thilo

AU - Madesta, Frederic

AU - Schmitz, Rudiger

AU - Kniep, Helge

AU - Baltruschat, Ivo

AU - Werner, Rene

AU - Schlaefer, Alexander

PY - 2020/2

Y1 - 2020/2

N2 - OBJECTIVE: This paper addresses two key problems of skin lesion classification. The first problem is the effective use of high-resolution images with pretrained standard architectures for image classification. The second problem is the high-class imbalance encountered in real-world multi-class datasets.METHODS: To use high-resolution images, we propose a novel patch-based attention architecture that provides global context between small, high-resolution patches. We modify three pretrained architectures and study the performance of patch-based attention. To counter class imbalance problems, we compare oversampling, balanced batch sampling, and class-specific loss weighting. Additionally, we propose a novel diagnosis-guided loss weighting method that takes the method used for ground-truth annotation into account.RESULTS: Our patch-based attention mechanism outperforms previous methods and improves the mean sensitivity by [Formula: see text]. Class balancing significantly improves the mean sensitivity and we show that our diagnosis-guided loss weighting method improves the mean sensitivity by [Formula: see text] over normal loss balancing.CONCLUSION: The novel patch-based attention mechanism can be integrated into pretrained architectures and provides global context between local patches while outperforming other patch-based methods. Hence, pretrained architectures can be readily used with high-resolution images without downsampling. The new diagnosis-guided loss weighting method outperforms other methods and allows for effective training when facing class imbalance.SIGNIFICANCE: The proposed methods improve automatic skin lesion classification. They can be extended to other clinical applications where high-resolution image data and class imbalance are relevant.

AB - OBJECTIVE: This paper addresses two key problems of skin lesion classification. The first problem is the effective use of high-resolution images with pretrained standard architectures for image classification. The second problem is the high-class imbalance encountered in real-world multi-class datasets.METHODS: To use high-resolution images, we propose a novel patch-based attention architecture that provides global context between small, high-resolution patches. We modify three pretrained architectures and study the performance of patch-based attention. To counter class imbalance problems, we compare oversampling, balanced batch sampling, and class-specific loss weighting. Additionally, we propose a novel diagnosis-guided loss weighting method that takes the method used for ground-truth annotation into account.RESULTS: Our patch-based attention mechanism outperforms previous methods and improves the mean sensitivity by [Formula: see text]. Class balancing significantly improves the mean sensitivity and we show that our diagnosis-guided loss weighting method improves the mean sensitivity by [Formula: see text] over normal loss balancing.CONCLUSION: The novel patch-based attention mechanism can be integrated into pretrained architectures and provides global context between local patches while outperforming other patch-based methods. Hence, pretrained architectures can be readily used with high-resolution images without downsampling. The new diagnosis-guided loss weighting method outperforms other methods and allows for effective training when facing class imbalance.SIGNIFICANCE: The proposed methods improve automatic skin lesion classification. They can be extended to other clinical applications where high-resolution image data and class imbalance are relevant.

KW - Journal Article

U2 - 10.1109/TBME.2019.2915839

DO - 10.1109/TBME.2019.2915839

M3 - SCORING: Journal article

C2 - 31071016

VL - 67

SP - 495

EP - 503

JO - IEEE T BIO-MED ENG

JF - IEEE T BIO-MED ENG

SN - 0018-9294

IS - 2

ER -