Epidermal thickness measurement on skin OCT using time-efficient deep learning with graph search

Lin, Chieh-Hsi; Rajabi‐Estarabadi, Ali; May, Julia; Pang, Yanzhen; Tsoukas, Maria M.; Dai, Yang; Avanaki, Kamran

doi:10.1117/12.2613041

Cited by 5 publications

(3 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[18] show classification of skin cancer using a single CNN model, trained end-to-end from real skin lesion images directly. [19] propose a time-efficient layer segmentation method developed on central unit processors (CPUs). CNN-GS aims to automatically segment two defined boundaries to calculate the epidermal thickness.…”

Section: Introduction and Releated Workmentioning

confidence: 99%

Advancing Skin Cancer Diagnosis through the Comparison of SHAP and Layer-wise Relevance Propagation (LRP)

ÖZKURT

2024

Preprint

View full text Add to dashboard Cite

Skin cancer detection stands at the intersection of cutting-edge technology and critical healthcare needs. In this study, we employ a Convolutional Neural Network (CNN) on the HAM10000 dataset, showcasing commendable accuracy. However, the inherent opacity of deep learning models sparks a quest for interpretability. We introduce SHAP (SHapley Additive exPlanations) and Layerwise analysis to unravel the decision-making intricacies. SHAP provides a global spotlight on feature importance, while Layerwise analysis offers a step-by-step exploration of the model's learning process. Beyond model performance, our goal is to inject transparency, trust, and innovation into skin cancer diagnosis, paving the way for a novel approach to humanizing artificial intelligence decisions in healthcare. Future work includes testing interpretation methods on diverse datasets and applying them to other healthcare domains for a transformative impact. This study not only contributes to understanding AI decisions in skin cancer diagnosis but also introduces methodologies that may revolutionize interpretability in healthcare AI applications.

show abstract

Section: Introduction and Releated Workmentioning

confidence: 99%

Advancing Skin Cancer Diagnosis through the Comparison of SHAP and Layer-wise Relevance Propagation (LRP)

ÖZKURT

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…These denoising techniques include signal processing tools, machine and deep learning tools and some other innovative algorithms. Deep learning has the potential to learn complex relationships and excel in extracting meaning information [1][2][3][4][5][6][7][8][9][10]. In paper [11], the author proposes the application of Convolutional Neural Network Denoising Auto-Encoders to intelligently filter noise in aircraft engine gas path health signals, enhancing signal quality for improved diagnostic accuracy.…”

Section: Introductionmentioning

confidence: 99%

Signal denoising for thermal acoustic imaging

Aggrawal,

Islam,

Zafar

et al. 2024

Photons Plus Ultrasound: Imaging and Sensing 2024

Self Cite

View full text Add to dashboard Cite

Thermoacoustic imaging (TAI) combines microwave energy's penetration depth with ultrasound's spatial resolution for medical imaging. Denoising is crucial in TAI to refine low energy thermoacoustic signals, overcoming depth limitations and improving imaging precision. We utilized the MI-TAT system to capture signals from different phantoms and gather data for training and validation. Our architectural approach harnesses both time and spatial signal features, enabling the design of an advanced deep-learning model.

show abstract

“…Specifically, retinal optical coherence tomography images have been tested extensively via deep learning to identify diagnoses, clinical features, and anatomical retinal layer segmentation [7,8]. Similar feats have been accomplished by dermatologists, using OCT images for detection of skin pathologies and layer segmentation [9,10]. While the potential success of unsupervised learning and deep learning is promising, they are not without shortcomings.…”

Section: Introductionmentioning

confidence: 99%

Facilitating deep learning through preprocessing of optical coherence tomography images

Winebrake

Kovacs

2023

BMC Ophthalmol

View full text Add to dashboard Cite

Background While deep learning has delivered promising results in the field of ophthalmology, the hurdle to complete a deep learning study is high. In this study, we aim to facilitate small scale model trainings by exploring the role of preprocessing to reduce computational burden and accelerate learning. Methods A small subset of a previously published dataset containing optical coherence tomography images of choroidal neovascularization, drusen, diabetic macula edema, and normal macula was modified using Fourier transformation and bandpass filter, producing high frequency images, original images, and low frequency images. Each set of images was trained with the same model, and their performances were compared. Results Compared to that with the original image dataset, the model trained with the high frequency image dataset achieved an improved final performance and reached maximum performance much earlier (in fewer epochs). The model trained with low frequency images did not achieve a meaningful performance. Conclusion Appropriate preprocessing of training images can accelerate the training process and can potentially facilitate modeling using artificial intelligence when limited by sample size or computational power.

show abstract

Epidermal thickness measurement on skin OCT using time-efficient deep learning with graph search

Cited by 5 publications

References 26 publications

Advancing Skin Cancer Diagnosis through the Comparison of SHAP and Layer-wise Relevance Propagation (LRP)

Advancing Skin Cancer Diagnosis through the Comparison of SHAP and Layer-wise Relevance Propagation (LRP)

Signal denoising for thermal acoustic imaging

Facilitating deep learning through preprocessing of optical coherence tomography images

Contact Info

Product

Resources

About