Paper reviews melanoma detection in dermoscopy images using image processing and machine learning.
problem Early detection and differentiation of melanoma from benign lesions.
method Lesion segmentation, clinical feature segmentation, machine learning algorithms.
result Machine learning algorithms can predict melanoma with high accuracy.
Computer systems for melanoma detection ranked by sensitivity, not specificity.
problem Improving computer systems for melanoma detection in dermoscopic images.
method Analyzed an open challenge in dermoscopic image classification, comparing five measures of diagnostic accuracy.
result Choice of performance measure significantly impacts ranking of computer systems.
U-Net network trained for skin lesion segmentation in ISIC challenge.
problem Skin lesion segmentation in medical imaging.
method Trained a U-Net network with adjusted loss function and post-processing.
result Improved accuracy in segmenting skin lesions.
MedAusbild team won ISIC challenge by classifying seven skin diseases.
problem Skin disease classification using medical images.
method Employed deep learning, specifically convolutional neural networks (CNNs), for seven skin disease classification.
result MedAusbild team won ISIC challenge 2018.
Efficient skin lesion analysis combines deep CNN and handcrafted features.
problem Skin lesion analysis for melanoma detection.
method Combines deep learning and handcrafted features, introduces hybrid features.
result 0.841 score on validation dataset using SVM classifier.
Deep learning models outperform dermatologists in skin cancer classification.
problem Skin cancer classification using dermoscopy images.
method Different pre-trained deep learning architectures (DenseNet 201, ResNet 152, Inception v3, InceptionResNet v2) applied on dermoscopy images.
result DenseNet 201 achieved highest AUC values for melanoma and basal cell carcinoma.
Deep learning models can misinterpret skin images, highlighting interpretability challenges.
problem Difficulty understanding deep learning model predictions for skin cancer.
method Exploring interpretability methods on a skin cancer dataset.
result Deep learning models can assign importance to irrelevant features.
P-Net predicts patient outcomes using sample networks.
problem Predicting patient outcomes based on sample data.
method P-Net constructs a network of patients, ranking and classifying them based on phenotype.
result P-Net performs similarly to classical methods in predicting patient outcomes.
Estimates cost savings from early cancer diagnosis.
problem Improving early cancer diagnosis to reduce treatment costs.
method Combining published cancer treatment cost estimates by stage with incidence rates by stage at diagnosis, and extrapolating to other cancer sites.
result Estimates U.S. national annual treatment cost-savings from early cancer diagnosis in the trillions.
VAE detects skin disease anomalies with high accuracy.
problem Anomaly detection in skin disease images.
method Variational Autoencoder (VAE) for deep learning.
result 0.779 AUCROC overall, 0.864 for melanoma, 0.872 for actinic keratosis.
BF-VI improves posterior approximation in complex models.
problem Inefficient posterior approximations in complex models.
method Combines normalizing flows and Bernstein polynomial transformations.
result BF-VI outperforms other VI methods in approximating complex multivariate posteriors.
Bayesian model improves cure fraction estimation in survival analysis.
problem Estimating cure fraction in survival analysis.
method Hierarchical Bayesian framework for semiparametric mixture cure model.
result Improved accuracy in capturing unobserved heterogeneity.
The paper tackles controlling gene regulatory networks with noisy measurements and uncertain inputs.
problem Controlling gene regulatory networks with indirect measurements and uncertain inputs.
method Modeling GRNs with POBDS, transforming to a Markov Decision Process, using Gaussian processes for cost function, and applying reinforcement learning and sparsification.
result Near-optimal control strategy for infinite-horizon control of GRNs is found.
Ensemble methods get better with more models if loss function is convex.
problem Whether ensemble methods improve with more models.
method Analyzing different ensemble methods and loss functions.
result Ensembles get better with more models only if the loss function is convex.
BAR reprograms black-box ML models for transfer learning with scarce data.
problem Transfer learning with limited data and resources.
method Zeroth-order optimization and multi-label mapping techniques to reprogram black-box models.
result BAR outperforms state-of-the-art methods and baseline transfer learning approaches.
Develops a method to estimate average hazard under non-proportional hazards without relying on proportional hazards assumption.
problem Estimation of treatment effects when hazards are non-proportional, leading to unstable hazard ratios.
method Semiparametric, doubly robust framework for covariate-adjusted average hazard estimation.
result Valid sqrt{n} inference with small bias and near-nominal confidence-interval coverage across proportional and non-proportional hazards settings.
EB-VAE combines tumor growth and dropout data for personalized treatment response modeling.
problem Challenges in integrating longitudinal tumor measurements, dropout information, and genetic covariates.
method Extended EB-VAE framework to jointly model longitudinal and time-to-event data, incorporating dropout hazard and genetic covariates.
result Hybrid decoder formulation yields consistent treatment-effect parameters and prior predictive performance comparable to neural decoder.
Deep learning skin lesion classifier explained using CAVs.
problem Limited acceptance of deep learning CAD systems due to opaque decision-making.
method Mapped human understandable concepts to RECOD model using CAVs.
result Classifier learns and encodes disease-related concepts in its latent representation.
ST-STORM separates semantic and appearance features for robust representation learning.
problem Traditional SSL methods fail to capture appearance cues in critical applications.
method Hybrid SSL framework with two latent streams, Content and Style, disentangled through gating mechanisms.
result The Style branch effectively isolates complex appearance phenomena without degrading semantic performance.
New method estimates causal effects of time-varying biomarkers on patient outcomes.
problem Estimating causal effects of time-dependent exposures in high-dimensional settings.
method Chronologically ordered PC-algorithm (COPC-algorithm) to extend IDA method.
result CPDAGs obtained using COPC-algorithm provide more accurate causal effect estimates and preserve chronological structure.
New DAM method improves AUC scores in medical image classification.
problem Maximizing AUC in large-scale medical image classification.
method Proposes AUC margin loss for robust optimization, conducts extensive empirical studies.
result Improves performance on four medical image classification tasks, achieving 1st place on Stanford CheXpert.