Skin Lesion Classification Using CNNs With Patch-Based Attention and Diagnosis-Guided Loss Weighting (2019-05-07T00:00:00.000000Z)

TL;DR

The new diagnosis-guided loss weighting method outperforms other methods and allows for effective training when facing class imbalance, and pretrained architectures can be readily used with high-resolution images without downsampling.

Abstract

<italic>Objective:</italic> 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. <italic>Methods:</italic> 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. <italic>Results:</italic> Our patch-based attention mechanism outperforms previous methods and improves the mean sensitivity by <inline-formula><tex-math notation="LaTeX">$\text{7}\%$</tex-math></inline-formula>. Class balancing significantly improves the mean sensitivity and we show that our diagnosis-guided loss weighting method improves the mean sensitivity by <inline-formula><tex-math notation="LaTeX">$\text{3}\%$</tex-math></inline-formula> over normal loss balancing. <italic>Conclusion:</italic> 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. <italic>Significance:</italic> 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.

References39 items

DermaKNet: Incorporating the Knowledge of Dermatologists to Convolutional Neural Networks for Skin Lesion Diagnosis

Seven-Point Checklist and Skin Lesion Classification Using Multitask Multimodal Neural Nets

(Preprint)

Skin Lesion Diagnosis using Ensembles, Unscaled Multi-Crop Evaluation and Loss Weighting

Clinical Skin Lesion Diagnosis Using Representations Inspired by Dermatologist Criteria

The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions

Automatic differentiation in PyTorch

Squeeze-and-Excitation Networks

A survey on deep learning in medical image analysis

Dermatologist-level classification of skin cancer with deep neural networks

Skin Cancer: Epidemiology, Disease Burden, Pathophysiology, Diagnosis, and Therapeutic Approaches

Aggregated Residual Transformations for Deep Neural Networks

Multi-resolution-Tract CNN with Hybrid Pretrained and Skin-Lesion Trained Layers

Densely Connected Convolutional Networks

Skin lesion analysis toward melanoma detection: A challenge at the 2017 International symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC)

LSTM-CF: Unifying Context Modeling and Fusion with LSTMs for RGB-D Scene Labeling

Deep features to classify skin lesions

Efficient multi‐scale 3D CNN with fully connected CRF for accurate brain lesion segmentation

Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning?

Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning

Rethinking the Inception Architecture for Computer Vision

Deep Learning, Sparse Coding, and SVM for Melanoma Recognition in Dermoscopy Images

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Adam: A Method for Stochastic Optimization

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

Intrinsic Melanin and Hemoglobin Colour Components for Skin Lesion Malignancy Detection

Skin lesion diagnosis from images using novel ensemble classification techniques

The US dermatology workforce: a specialty remains in shortage.

Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: a meta‐analysis of studies performed in a clinical setting

Diagnostic accuracy of dermoscopy/dermatoscopy

Interactive Atlas of Dermoscopy

Epiluminescence microscopy for the diagnosis of doubtful melanocytic skin lesions. Comparison of the ABCD rule of dermatoscopy and a new 7-point checklist based on pattern analysis.

The ABCD rule of dermatoscopy. High prospective value in the diagnosis of doubtful melanocytic skin lesions.

Skin lesion classification from dermoscopic images using deep learning techniques

Dependency Exploitation: A Unified CNN-RNN Approach for Visual Emotion Recognition

Cancer statistics, 2017

Interactive atlas of dermoscopy

Dermoscopy: a tutorial (Book) and CD-ROM

“The International Skin Imaging Collaboration: Melanoma Project,”