SISC: End-to-end Interpretable Discovery Radiomics-Driven Lung Cancer\n Prediction via Stacked Interpretable Sequencing Cells

2019/01/14 by Vignesh Sankar, Devinder Kumar, Sankar, Vignesh +7
Computer Science · Medicine · #AI in cancer detection #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Lung Cancer Diagnosis and Treatment #Lung Cancer Treatments and Mutations #Radiomics and Machine Learning in Medical Imaging

paper · pdf · doi:10.48550/arxiv.1901.04641

openalex publication_date 2019/01/14 · openalex created_date 2022/07/30 · openalex updated_date 2026/07/28

Abstract

Objective: Lung cancer is the leading cause of cancer-related death\nworldwide. Computer-aided diagnosis (CAD) systems have shown significant\npromise in recent years for facilitating the effective detection and\nclassification of abnormal lung nodules in computed tomography (CT) scans.\nWhile hand-engineered radiomic features have been traditionally used for lung\ncancer prediction, there have been significant recent successes achieving\nstate-of-the-art results in the area of discovery radiomics. Here, radiomic\nsequencers comprising of highly discriminative radiomic features are discovered\ndirectly from archival medical data. However, the interpretation of predictions\nmade using such radiomic sequencers remains a challenge. Method: A novel\nend-to-end interpretable discovery radiomics-driven lung cancer prediction\npipeline has been designed, build, and tested. The radiomic sequencer being\ndiscovered possesses a deep architecture comprised of stacked interpretable\nsequencing cells (SISC). Results: The SISC architecture is shown to outperform\nprevious approaches while providing more insight in to its decision making\nprocess. Conclusion: The SISC radiomic sequencer is able to achieve\nstate-of-the-art results in lung cancer prediction, and also offers prediction\ninterpretability in the form of critical response maps. Significance: The\ncritical response maps are useful for not only validating the predictions of\nthe proposed SISC radiomic sequencer, but also provide improved\nradiologist-machine collaboration for effective diagnosis.\n

Citations

• Understanding Neural Networks Through Deep Visualization
• The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI): A Completed Reference Database of Lung Nodules on CT Scans
• Striving for Simplicity: The All Convolutional Net
• How to Explain Individual Classification Decisions
• Multiresolution gray-scale and rotation invariant texture classification with local binary patterns
• ImageNet classification with deep convolutional neural networks
• Discovery Radiomics via StochasticNet Sequencers for Cancer Detection
• Multi-crop Convolutional Neural Networks for lung nodule malignancy suspiciousness classification
• Visualizing Deep Neural Network Decisions: Prediction Difference Analysis
• DeepLung: Deep 3D Dual Path Nets for Automated Pulmonary Nodule Detection and Classification
• Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps
• Learning Multiple Layers of Features from Tiny Images

Related

• Discovery Radiomics for Pathologically-Proven Computed Tomography Lung\n Cancer Prediction
• Discovery Radiomics via Evolutionary Deep Radiomic Sequencer Discovery for Pathologically-Proven Lung Cancer Detection
• Discovery Radiomics via StochasticNet Sequencers for Cancer Detection
• EMeRALDS: Electronic Medical Record Driven Automated Lung Nodule Detection and Classification in Thoracic CT Images
• Genotype-Guided Radiomics Signatures for Recurrence Prediction of Non-Small-Cell Lung Cancer